Merge branch 'comfyanonymous:master' into master

.ci/update_windows/update.py

@@ -33,12 +33,12 @@ def pull(repo, remote_name='origin', branch='master'):
 
                 user = repo.default_signature
                 tree = repo.index.write_tree()
-                commit = repo.create_commit('HEAD',
-                                            user,
-                                            user,
-                                            'Merge!',
-                                            tree,
-                                            [repo.head.target, remote_master_id])
+                repo.create_commit('HEAD',
+                                    user,
+                                    user,
+                                    'Merge!',
+                                    tree,
+                                    [repo.head.target, remote_master_id])
                 # We need to do this or git CLI will think we are still merging.
                 repo.state_cleanup()
             else:

README.md

@@ -147,7 +147,7 @@ AMD users can install rocm and pytorch with pip if you don't have it already ins
 
 This is the command to install the nightly with ROCm 6.2 which might have some performance improvements:
 
-```pip install --pre torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/rocm6.2```
+```pip install --pre torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/rocm6.2.4```
 
 ### NVIDIA
 

api_server/routes/internal/internal_routes.py

@@ -40,7 +40,7 @@ class InternalRoutes:
             return web.json_response("".join([(l["t"] + " - " + l["m"]) for l in app.logger.get_logs()]))
 
         @self.routes.get('/logs/raw')
-        async def get_logs(request):
+        async def get_raw_logs(request):
             self.terminal_service.update_size()
             return web.json_response({
                 "entries": list(app.logger.get_logs()),

comfy/cldm/cldm.py

@@ -413,7 +413,6 @@ class ControlNet(nn.Module):
         out_output = []
         out_middle = []
 
-        hs = []
         if self.num_classes is not None:
             assert y.shape[0] == x.shape[0]
             emb = emb + self.label_emb(y)

comfy/controlnet.py

@@ -297,7 +297,6 @@ class ControlLoraOps:
     class Linear(torch.nn.Module, comfy.ops.CastWeightBiasOp):
         def __init__(self, in_features: int, out_features: int, bias: bool = True,
                     device=None, dtype=None) -> None:
-            factory_kwargs = {'device': device, 'dtype': dtype}
             super().__init__()
             self.in_features = in_features
             self.out_features = out_features
@@ -382,7 +381,6 @@ class ControlLora(ControlNet):
         self.control_model.to(comfy.model_management.get_torch_device())
         diffusion_model = model.diffusion_model
         sd = diffusion_model.state_dict()
-        cm = self.control_model.state_dict()
 
         for k in sd:
             weight = sd[k]
@@ -823,7 +821,7 @@ def load_t2i_adapter(t2i_data, model_options={}): #TODO: model_options
         for i in range(4):
             for j in range(2):
                 prefix_replace["adapter.body.{}.resnets.{}.".format(i, j)] = "body.{}.".format(i * 2 + j)
-            prefix_replace["adapter.body.{}.".format(i, j)] = "body.{}.".format(i * 2)
+            prefix_replace["adapter.body.{}.".format(i, )] = "body.{}.".format(i * 2)
         prefix_replace["adapter."] = ""
         t2i_data = comfy.utils.state_dict_prefix_replace(t2i_data, prefix_replace)
     keys = t2i_data.keys()

comfy/diffusers_convert.py

@@ -157,16 +157,23 @@ vae_conversion_map_attn = [
 ]
 
 
-def reshape_weight_for_sd(w):
+def reshape_weight_for_sd(w, conv3d=False):
     # convert HF linear weights to SD conv2d weights
-    return w.reshape(*w.shape, 1, 1)
+    if conv3d:
+        return w.reshape(*w.shape, 1, 1, 1)
+    else:
+        return w.reshape(*w.shape, 1, 1)
 
 
 def convert_vae_state_dict(vae_state_dict):
     mapping = {k: k for k in vae_state_dict.keys()}
+    conv3d = False
     for k, v in mapping.items():
         for sd_part, hf_part in vae_conversion_map:
             v = v.replace(hf_part, sd_part)
+        if v.endswith(".conv.weight"):
+            if not conv3d and vae_state_dict[k].ndim == 5:
+                conv3d = True
         mapping[k] = v
     for k, v in mapping.items():
         if "attentions" in k:
@@ -179,7 +186,7 @@ def convert_vae_state_dict(vae_state_dict):
         for weight_name in weights_to_convert:
             if f"mid.attn_1.{weight_name}.weight" in k:
                 logging.debug(f"Reshaping {k} for SD format")
-                new_state_dict[k] = reshape_weight_for_sd(v)
+                new_state_dict[k] = reshape_weight_for_sd(v, conv3d=conv3d)
     return new_state_dict
 
 

comfy/extra_samplers/uni_pc.py

@@ -703,7 +703,6 @@ class UniPC:
     ):
         # t_0 = 1. / self.noise_schedule.total_N if t_end is None else t_end
         # t_T = self.noise_schedule.T if t_start is None else t_start
-        device = x.device
         steps = len(timesteps) - 1
         if method == 'multistep':
             assert steps >= order

comfy/gligen.py

@@ -1,3 +1,4 @@
+import math
 import torch
 from torch import nn
 from .ldm.modules.attention import CrossAttention

comfy/hooks.py

@@ -130,7 +130,7 @@ class WeightHook(Hook):
                 weights = self.weights
             else:
                 weights = self.weights_clip
-        k = model.add_hook_patches(hook=self, patches=weights, strength_patch=strength)
+        model.add_hook_patches(hook=self, patches=weights, strength_patch=strength)
         registered.append(self)
         return True
         # TODO: add logs about any keys that were not applied

comfy/k_diffusion/deis.py

@@ -11,7 +11,6 @@ import numpy as np
 # Transfer from the input time (sigma) used in EDM to that (t) used in DEIS.
 
 def edm2t(edm_steps, epsilon_s=1e-3, sigma_min=0.002, sigma_max=80):
-    vp_sigma = lambda beta_d, beta_min: lambda t: (np.e ** (0.5 * beta_d * (t ** 2) + beta_min * t) - 1) ** 0.5
     vp_sigma_inv = lambda beta_d, beta_min: lambda sigma: ((beta_min ** 2 + 2 * beta_d * (sigma ** 2 + 1).log()).sqrt() - beta_min) / beta_d
     vp_beta_d = 2 * (np.log(torch.tensor(sigma_min).cpu() ** 2 + 1) / epsilon_s - np.log(torch.tensor(sigma_max).cpu() ** 2 + 1)) / (epsilon_s - 1)
     vp_beta_min = np.log(torch.tensor(sigma_max).cpu() ** 2 + 1) - 0.5 * vp_beta_d

comfy/latent_formats.py

@@ -352,3 +352,7 @@ class LTXV(LatentFormat):
         ]
 
         self.latent_rgb_factors_bias = [-0.0571, -0.1657, -0.2512]
+
+class HunyuanVideo(LatentFormat):
+    latent_channels = 16
+    scale_factor = 0.476986

comfy/ldm/audio/autoencoder.py

@@ -97,7 +97,7 @@ def get_activation(activation: Literal["elu", "snake", "none"], antialias=False,
         raise ValueError(f"Unknown activation {activation}")
 
     if antialias:
-        act = Activation1d(act)
+        act = Activation1d(act)  # noqa: F821 Activation1d is not defined
 
     return act
 

comfy/ldm/audio/dit.py

@@ -158,7 +158,6 @@ class RotaryEmbedding(nn.Module):
     def forward(self, t):
         # device = self.inv_freq.device
         device = t.device
-        dtype = t.dtype
 
         # t = t.to(torch.float32)
 
@@ -170,7 +169,7 @@ class RotaryEmbedding(nn.Module):
         if self.scale is None:
             return freqs, 1.
 
-        power = (torch.arange(seq_len, device = device) - (seq_len // 2)) / self.scale_base
+        power = (torch.arange(seq_len, device = device) - (seq_len // 2)) / self.scale_base  # noqa: F821 seq_len is not defined
         scale = comfy.ops.cast_to_input(self.scale, t) ** rearrange(power, 'n -> n 1')
         scale = torch.cat((scale, scale), dim = -1)
 
@@ -229,9 +228,9 @@ class FeedForward(nn.Module):
             linear_in = GLU(dim, inner_dim, activation, dtype=dtype, device=device, operations=operations)
         else:
             linear_in = nn.Sequential(
-                Rearrange('b n d -> b d n') if use_conv else nn.Identity(),
+                rearrange('b n d -> b d n') if use_conv else nn.Identity(),
                 operations.Linear(dim, inner_dim, bias = not no_bias, dtype=dtype, device=device) if not use_conv else operations.Conv1d(dim, inner_dim, conv_kernel_size, padding = (conv_kernel_size // 2), bias = not no_bias, dtype=dtype, device=device),
-                Rearrange('b n d -> b d n') if use_conv else nn.Identity(),
+                rearrange('b n d -> b d n') if use_conv else nn.Identity(),
                 activation
             )
 
@@ -246,9 +245,9 @@ class FeedForward(nn.Module):
 
         self.ff = nn.Sequential(
             linear_in,
-            Rearrange('b d n -> b n d') if use_conv else nn.Identity(),
+            rearrange('b d n -> b n d') if use_conv else nn.Identity(),
             linear_out,
-            Rearrange('b n d -> b d n') if use_conv else nn.Identity(),
+            rearrange('b n d -> b d n') if use_conv else nn.Identity(),
         )
 
     def forward(self, x):
@@ -346,18 +345,13 @@ class Attention(nn.Module):
 
         # determine masking
         masks = []
-        final_attn_mask = None # The mask that will be applied to the attention matrix, taking all masks into account
 
         if input_mask is not None:
             input_mask = rearrange(input_mask, 'b j -> b 1 1 j')
             masks.append(~input_mask)
 
         # Other masks will be added here later
-
-        if len(masks) > 0:
-            final_attn_mask = ~or_reduce(masks)
-
-        n, device = q.shape[-2], q.device
+        n = q.shape[-2]
 
         causal = self.causal if causal is None else causal
 

comfy/ldm/aura/mmdit.py

@@ -147,7 +147,6 @@ class DoubleAttention(nn.Module):
 
         bsz, seqlen1, _ = c.shape
         bsz, seqlen2, _ = x.shape
-        seqlen = seqlen1 + seqlen2
 
         cq, ck, cv = self.w1q(c), self.w1k(c), self.w1v(c)
         cq = cq.view(bsz, seqlen1, self.n_heads, self.head_dim)

comfy/ldm/common_dit.py

@@ -4,9 +4,12 @@ import comfy.ops
 def pad_to_patch_size(img, patch_size=(2, 2), padding_mode="circular"):
     if padding_mode == "circular" and (torch.jit.is_tracing() or torch.jit.is_scripting()):
         padding_mode = "reflect"
-    pad_h = (patch_size[0] - img.shape[-2] % patch_size[0]) % patch_size[0]
-    pad_w = (patch_size[1] - img.shape[-1] % patch_size[1]) % patch_size[1]
-    return torch.nn.functional.pad(img, (0, pad_w, 0, pad_h), mode=padding_mode)
+
+    pad = ()
+    for i in range(img.ndim - 2):
+        pad = (0, (patch_size[i] - img.shape[i + 2] % patch_size[i]) % patch_size[i]) + pad
+
+    return torch.nn.functional.pad(img, pad, mode=padding_mode)
 
 try:
     rms_norm_torch = torch.nn.functional.rms_norm

comfy/ldm/flux/layers.py

@@ -114,7 +114,7 @@ class Modulation(nn.Module):
 
 
 class DoubleStreamBlock(nn.Module):
-    def __init__(self, hidden_size: int, num_heads: int, mlp_ratio: float, qkv_bias: bool = False, dtype=None, device=None, operations=None):
+    def __init__(self, hidden_size: int, num_heads: int, mlp_ratio: float, qkv_bias: bool = False, flipped_img_txt=False, dtype=None, device=None, operations=None):
         super().__init__()
 
         mlp_hidden_dim = int(hidden_size * mlp_ratio)
@@ -141,8 +141,9 @@ class DoubleStreamBlock(nn.Module):
             nn.GELU(approximate="tanh"),
             operations.Linear(mlp_hidden_dim, hidden_size, bias=True, dtype=dtype, device=device),
         )
+        self.flipped_img_txt = flipped_img_txt
 
-    def forward(self, img: Tensor, txt: Tensor, vec: Tensor, pe: Tensor):
+    def forward(self, img: Tensor, txt: Tensor, vec: Tensor, pe: Tensor, attn_mask=None):
         img_mod1, img_mod2 = self.img_mod(vec)
         txt_mod1, txt_mod2 = self.txt_mod(vec)
 
@@ -160,12 +161,22 @@ class DoubleStreamBlock(nn.Module):
         txt_q, txt_k, txt_v = txt_qkv.view(txt_qkv.shape[0], txt_qkv.shape[1], 3, self.num_heads, -1).permute(2, 0, 3, 1, 4)
         txt_q, txt_k = self.txt_attn.norm(txt_q, txt_k, txt_v)
 
-        # run actual attention
-        attn = attention(torch.cat((txt_q, img_q), dim=2),
-                         torch.cat((txt_k, img_k), dim=2),
-                         torch.cat((txt_v, img_v), dim=2), pe=pe)
+        if self.flipped_img_txt:
+            # run actual attention
+            attn = attention(torch.cat((img_q, txt_q), dim=2),
+                             torch.cat((img_k, txt_k), dim=2),
+                             torch.cat((img_v, txt_v), dim=2),
+                             pe=pe, mask=attn_mask)
 
-        txt_attn, img_attn = attn[:, : txt.shape[1]], attn[:, txt.shape[1] :]
+            img_attn, txt_attn = attn[:, : img.shape[1]], attn[:, img.shape[1]:]
+        else:
+            # run actual attention
+            attn = attention(torch.cat((txt_q, img_q), dim=2),
+                             torch.cat((txt_k, img_k), dim=2),
+                             torch.cat((txt_v, img_v), dim=2),
+                             pe=pe, mask=attn_mask)
+
+            txt_attn, img_attn = attn[:, : txt.shape[1]], attn[:, txt.shape[1]:]
 
         # calculate the img bloks
         img = img + img_mod1.gate * self.img_attn.proj(img_attn)
@@ -217,7 +228,7 @@ class SingleStreamBlock(nn.Module):
         self.mlp_act = nn.GELU(approximate="tanh")
         self.modulation = Modulation(hidden_size, double=False, dtype=dtype, device=device, operations=operations)
 
-    def forward(self, x: Tensor, vec: Tensor, pe: Tensor) -> Tensor:
+    def forward(self, x: Tensor, vec: Tensor, pe: Tensor, attn_mask=None) -> Tensor:
         mod, _ = self.modulation(vec)
         x_mod = (1 + mod.scale) * self.pre_norm(x) + mod.shift
         qkv, mlp = torch.split(self.linear1(x_mod), [3 * self.hidden_size, self.mlp_hidden_dim], dim=-1)
@@ -226,7 +237,7 @@ class SingleStreamBlock(nn.Module):
         q, k = self.norm(q, k, v)
 
         # compute attention
-        attn = attention(q, k, v, pe=pe)
+        attn = attention(q, k, v, pe=pe, mask=attn_mask)
         # compute activation in mlp stream, cat again and run second linear layer
         output = self.linear2(torch.cat((attn, self.mlp_act(mlp)), 2))
         x += mod.gate * output

comfy/ldm/flux/math.py

@@ -1,14 +1,15 @@
 import torch
 from einops import rearrange
 from torch import Tensor
+
 from comfy.ldm.modules.attention import optimized_attention
 import comfy.model_management
 
-def attention(q: Tensor, k: Tensor, v: Tensor, pe: Tensor) -> Tensor:
+def attention(q: Tensor, k: Tensor, v: Tensor, pe: Tensor, mask=None) -> Tensor:
     q, k = apply_rope(q, k, pe)
 
     heads = q.shape[1]
-    x = optimized_attention(q, k, v, heads, skip_reshape=True)
+    x = optimized_attention(q, k, v, heads, skip_reshape=True, mask=mask)
     return x
 
 
@@ -33,3 +34,4 @@ def apply_rope(xq: Tensor, xk: Tensor, freqs_cis: Tensor):
     xq_out = freqs_cis[..., 0] * xq_[..., 0] + freqs_cis[..., 1] * xq_[..., 1]
     xk_out = freqs_cis[..., 0] * xk_[..., 0] + freqs_cis[..., 1] * xk_[..., 1]
     return xq_out.reshape(*xq.shape).type_as(xq), xk_out.reshape(*xk.shape).type_as(xk)
+

comfy/ldm/flux/model.py

@@ -4,6 +4,8 @@ from dataclasses import dataclass
 
 import torch
 from torch import Tensor, nn
+from einops import rearrange, repeat
+import comfy.ldm.common_dit
 
 from .layers import (
     DoubleStreamBlock,
@@ -14,9 +16,6 @@ from .layers import (
     timestep_embedding,
 )
 
-from einops import rearrange, repeat
-import comfy.ldm.common_dit
-
 @dataclass
 class FluxParams:
     in_channels: int
@@ -98,8 +97,9 @@ class Flux(nn.Module):
         timesteps: Tensor,
         y: Tensor,
         guidance: Tensor = None,
-        control=None,
+        control = None,
         transformer_options={},
+        attn_mask: Tensor = None,
     ) -> Tensor:
         patches_replace = transformer_options.get("patches_replace", {})
         if img.ndim != 3 or txt.ndim != 3:
@@ -124,14 +124,27 @@ class Flux(nn.Module):
             if ("double_block", i) in blocks_replace:
                 def block_wrap(args):
                     out = {}
-                    out["img"], out["txt"] = block(img=args["img"], txt=args["txt"], vec=args["vec"], pe=args["pe"])
+                    out["img"], out["txt"] = block(img=args["img"],
+                                                   txt=args["txt"],
+                                                   vec=args["vec"],
+                                                   pe=args["pe"],
+                                                   attn_mask=args.get("attn_mask"))
                     return out
 
-                out = blocks_replace[("double_block", i)]({"img": img, "txt": txt, "vec": vec, "pe": pe}, {"original_block": block_wrap})
+                out = blocks_replace[("double_block", i)]({"img": img,
+                                                           "txt": txt,
+                                                           "vec": vec,
+                                                           "pe": pe,
+                                                           "attn_mask": attn_mask},
+                                                          {"original_block": block_wrap})
                 txt = out["txt"]
                 img = out["img"]
             else:
-                img, txt = block(img=img, txt=txt, vec=vec, pe=pe)
+                img, txt = block(img=img,
+                                 txt=txt,
+                                 vec=vec,
+                                 pe=pe,
+                                 attn_mask=attn_mask)
 
             if control is not None: # Controlnet
                 control_i = control.get("input")
@@ -146,13 +159,20 @@ class Flux(nn.Module):
             if ("single_block", i) in blocks_replace:
                 def block_wrap(args):
                     out = {}
-                    out["img"] = block(args["img"], vec=args["vec"], pe=args["pe"])
+                    out["img"] = block(args["img"],
+                                       vec=args["vec"],
+                                       pe=args["pe"],
+                                       attn_mask=args.get("attn_mask"))
                     return out
 
-                out = blocks_replace[("single_block", i)]({"img": img, "vec": vec, "pe": pe}, {"original_block": block_wrap})
+                out = blocks_replace[("single_block", i)]({"img": img,
+                                                           "vec": vec,
+                                                           "pe": pe,
+                                                           "attn_mask": attn_mask}, 
+                                                          {"original_block": block_wrap})
                 img = out["img"]
             else:
-                img = block(img, vec=vec, pe=pe)
+                img = block(img, vec=vec, pe=pe, attn_mask=attn_mask)
 
             if control is not None: # Controlnet
                 control_o = control.get("output")
@@ -181,5 +201,5 @@ class Flux(nn.Module):
         img_ids = repeat(img_ids, "h w c -> b (h w) c", b=bs)
 
         txt_ids = torch.zeros((bs, context.shape[1], 3), device=x.device, dtype=x.dtype)
-        out = self.forward_orig(img, img_ids, context, txt_ids, timestep, y, guidance, control, transformer_options)
+        out = self.forward_orig(img, img_ids, context, txt_ids, timestep, y, guidance, control, transformer_options, attn_mask=kwargs.get("attention_mask", None))
         return rearrange(out, "b (h w) (c ph pw) -> b c (h ph) (w pw)", h=h_len, w=w_len, ph=2, pw=2)[:,:,:h,:w]

comfy/ldm/genmo/joint_model/asymm_models_joint.py

@@ -461,8 +461,6 @@ class AsymmDiTJoint(nn.Module):
         pH, pW = H // self.patch_size, W // self.patch_size
         x = self.embed_x(x)  # (B, N, D), where N = T * H * W / patch_size ** 2
         assert x.ndim == 3
-        B = x.size(0)
-
 
         pH, pW = H // self.patch_size, W // self.patch_size
         N = T * pH * pW

comfy/ldm/hunyuan_video/model.py

@@ -0,0 +1,330 @@
+#Based on Flux code because of weird hunyuan video code license.
+
+import torch
+import comfy.ldm.flux.layers
+import comfy.ldm.modules.diffusionmodules.mmdit
+from comfy.ldm.modules.attention import optimized_attention
+
+
+from dataclasses import dataclass
+from einops import repeat
+
+from torch import Tensor, nn
+
+from comfy.ldm.flux.layers import (
+    DoubleStreamBlock,
+    EmbedND,
+    LastLayer,
+    MLPEmbedder,
+    SingleStreamBlock,
+    timestep_embedding
+)
+
+import comfy.ldm.common_dit
+
+
+@dataclass
+class HunyuanVideoParams:
+    in_channels: int
+    out_channels: int
+    vec_in_dim: int
+    context_in_dim: int
+    hidden_size: int
+    mlp_ratio: float
+    num_heads: int
+    depth: int
+    depth_single_blocks: int
+    axes_dim: list
+    theta: int
+    patch_size: list
+    qkv_bias: bool
+    guidance_embed: bool
+
+
+class SelfAttentionRef(nn.Module):
+    def __init__(self, dim: int, qkv_bias: bool = False, dtype=None, device=None, operations=None):
+        super().__init__()
+        self.qkv = operations.Linear(dim, dim * 3, bias=qkv_bias, dtype=dtype, device=device)
+        self.proj = operations.Linear(dim, dim, dtype=dtype, device=device)
+
+
+class TokenRefinerBlock(nn.Module):
+    def __init__(
+        self,
+        hidden_size,
+        heads,
+        dtype=None,
+        device=None,
+        operations=None
+    ):
+        super().__init__()
+        self.heads = heads
+        mlp_hidden_dim = hidden_size * 4
+
+        self.adaLN_modulation = nn.Sequential(
+            nn.SiLU(),
+            operations.Linear(hidden_size, 2 * hidden_size, bias=True, dtype=dtype, device=device),
+        )
+
+        self.norm1 = operations.LayerNorm(hidden_size, elementwise_affine=True, eps=1e-6, dtype=dtype, device=device)
+        self.self_attn = SelfAttentionRef(hidden_size, True, dtype=dtype, device=device, operations=operations)
+
+        self.norm2 = operations.LayerNorm(hidden_size, elementwise_affine=True, eps=1e-6, dtype=dtype, device=device)
+
+        self.mlp = nn.Sequential(
+            operations.Linear(hidden_size, mlp_hidden_dim, bias=True, dtype=dtype, device=device),
+            nn.SiLU(),
+            operations.Linear(mlp_hidden_dim, hidden_size, bias=True, dtype=dtype, device=device),
+        )
+
+    def forward(self, x, c, mask):
+        mod1, mod2 = self.adaLN_modulation(c).chunk(2, dim=1)
+
+        norm_x = self.norm1(x)
+        qkv = self.self_attn.qkv(norm_x)
+        q, k, v = qkv.reshape(qkv.shape[0], qkv.shape[1], 3, self.heads, -1).permute(2, 0, 3, 1, 4)
+        attn = optimized_attention(q, k, v, self.heads, mask=mask, skip_reshape=True)
+
+        x = x + self.self_attn.proj(attn) * mod1.unsqueeze(1)
+        x = x + self.mlp(self.norm2(x)) * mod2.unsqueeze(1)
+        return x
+
+
+class IndividualTokenRefiner(nn.Module):
+    def __init__(
+        self,
+        hidden_size,
+        heads,
+        num_blocks,
+        dtype=None,
+        device=None,
+        operations=None
+    ):
+        super().__init__()
+        self.blocks = nn.ModuleList(
+            [
+                TokenRefinerBlock(
+                    hidden_size=hidden_size,
+                    heads=heads,
+                    dtype=dtype,
+                    device=device,
+                    operations=operations
+                )
+                for _ in range(num_blocks)
+            ]
+        )
+
+    def forward(self, x, c, mask):
+        m = None
+        if mask is not None:
+            m = mask.view(mask.shape[0], 1, 1, mask.shape[1]).repeat(1, 1, mask.shape[1], 1)
+            m = m + m.transpose(2, 3)
+
+        for block in self.blocks:
+            x = block(x, c, m)
+        return x
+
+
+
+class TokenRefiner(nn.Module):
+    def __init__(
+        self,
+        text_dim,
+        hidden_size,
+        heads,
+        num_blocks,
+        dtype=None,
+        device=None,
+        operations=None
+    ):
+        super().__init__()
+
+        self.input_embedder = operations.Linear(text_dim, hidden_size, bias=True, dtype=dtype, device=device)
+        self.t_embedder = MLPEmbedder(256, hidden_size, dtype=dtype, device=device, operations=operations)
+        self.c_embedder = MLPEmbedder(text_dim, hidden_size, dtype=dtype, device=device, operations=operations)
+        self.individual_token_refiner = IndividualTokenRefiner(hidden_size, heads, num_blocks, dtype=dtype, device=device, operations=operations)
+
+    def forward(
+        self,
+        x,
+        timesteps,
+        mask,
+    ):
+        t = self.t_embedder(timestep_embedding(timesteps, 256, time_factor=1.0).to(x.dtype))
+        # m = mask.float().unsqueeze(-1)
+        # c = (x.float() * m).sum(dim=1) / m.sum(dim=1) #TODO: the following works when the x.shape is the same length as the tokens but might break otherwise
+        c = x.sum(dim=1) / x.shape[1]
+
+        c = t + self.c_embedder(c.to(x.dtype))
+        x = self.input_embedder(x)
+        x = self.individual_token_refiner(x, c, mask)
+        return x
+
+class HunyuanVideo(nn.Module):
+    """
+    Transformer model for flow matching on sequences.
+    """
+
+    def __init__(self, image_model=None, final_layer=True, dtype=None, device=None, operations=None, **kwargs):
+        super().__init__()
+        self.dtype = dtype
+        params = HunyuanVideoParams(**kwargs)
+        self.params = params
+        self.patch_size = params.patch_size
+        self.in_channels = params.in_channels
+        self.out_channels = params.out_channels
+        if params.hidden_size % params.num_heads != 0:
+            raise ValueError(
+                f"Hidden size {params.hidden_size} must be divisible by num_heads {params.num_heads}"
+            )
+        pe_dim = params.hidden_size // params.num_heads
+        if sum(params.axes_dim) != pe_dim:
+            raise ValueError(f"Got {params.axes_dim} but expected positional dim {pe_dim}")
+        self.hidden_size = params.hidden_size
+        self.num_heads = params.num_heads
+        self.pe_embedder = EmbedND(dim=pe_dim, theta=params.theta, axes_dim=params.axes_dim)
+
+        self.img_in = comfy.ldm.modules.diffusionmodules.mmdit.PatchEmbed(None, self.patch_size, self.in_channels, self.hidden_size, conv3d=True, dtype=dtype, device=device, operations=operations)
+        self.time_in = MLPEmbedder(in_dim=256, hidden_dim=self.hidden_size, dtype=dtype, device=device, operations=operations)
+        self.vector_in = MLPEmbedder(params.vec_in_dim, self.hidden_size, dtype=dtype, device=device, operations=operations)
+        self.guidance_in = (
+            MLPEmbedder(in_dim=256, hidden_dim=self.hidden_size, dtype=dtype, device=device, operations=operations) if params.guidance_embed else nn.Identity()
+        )
+
+        self.txt_in = TokenRefiner(params.context_in_dim, self.hidden_size, self.num_heads, 2, dtype=dtype, device=device, operations=operations)
+
+        self.double_blocks = nn.ModuleList(
+            [
+                DoubleStreamBlock(
+                    self.hidden_size,
+                    self.num_heads,
+                    mlp_ratio=params.mlp_ratio,
+                    qkv_bias=params.qkv_bias,
+                    flipped_img_txt=True,
+                    dtype=dtype, device=device, operations=operations
+                )
+                for _ in range(params.depth)
+            ]
+        )
+
+        self.single_blocks = nn.ModuleList(
+            [
+                SingleStreamBlock(self.hidden_size, self.num_heads, mlp_ratio=params.mlp_ratio, dtype=dtype, device=device, operations=operations)
+                for _ in range(params.depth_single_blocks)
+            ]
+        )
+
+        if final_layer:
+            self.final_layer = LastLayer(self.hidden_size, self.patch_size[-1], self.out_channels, dtype=dtype, device=device, operations=operations)
+
+    def forward_orig(
+        self,
+        img: Tensor,
+        img_ids: Tensor,
+        txt: Tensor,
+        txt_ids: Tensor,
+        txt_mask: Tensor,
+        timesteps: Tensor,
+        y: Tensor,
+        guidance: Tensor = None,
+        control=None,
+        transformer_options={},
+    ) -> Tensor:
+        patches_replace = transformer_options.get("patches_replace", {})
+
+        initial_shape = list(img.shape)
+        # running on sequences img
+        img = self.img_in(img)
+        vec = self.time_in(timestep_embedding(timesteps, 256, time_factor=1.0).to(img.dtype))
+
+        vec = vec + self.vector_in(y[:, :self.params.vec_in_dim])
+
+        if self.params.guidance_embed:
+            if guidance is None:
+                raise ValueError("Didn't get guidance strength for guidance distilled model.")
+            vec = vec + self.guidance_in(timestep_embedding(guidance, 256).to(img.dtype))
+
+        if txt_mask is not None and not torch.is_floating_point(txt_mask):
+            txt_mask = (txt_mask - 1).to(img.dtype) * torch.finfo(img.dtype).max
+
+        txt = self.txt_in(txt, timesteps, txt_mask)
+
+        ids = torch.cat((img_ids, txt_ids), dim=1)
+        pe = self.pe_embedder(ids)
+
+        img_len = img.shape[1]
+        if txt_mask is not None:
+            attn_mask_len = img_len + txt.shape[1]
+            attn_mask = torch.zeros((1, 1, attn_mask_len), dtype=img.dtype, device=img.device)
+            attn_mask[:, 0, img_len:] = txt_mask
+        else:
+            attn_mask = None
+
+        blocks_replace = patches_replace.get("dit", {})
+        for i, block in enumerate(self.double_blocks):
+            if ("double_block", i) in blocks_replace:
+                def block_wrap(args):
+                    out = {}
+                    out["img"], out["txt"] = block(img=args["img"], txt=args["txt"], vec=args["vec"], pe=args["pe"], attn_mask=args["attention_mask"])
+                    return out
+
+                out = blocks_replace[("double_block", i)]({"img": img, "txt": txt, "vec": vec, "pe": pe, "attention_mask": attn_mask}, {"original_block": block_wrap})
+                txt = out["txt"]
+                img = out["img"]
+            else:
+                img, txt = block(img=img, txt=txt, vec=vec, pe=pe, attn_mask=attn_mask)
+
+            if control is not None: # Controlnet
+                control_i = control.get("input")
+                if i < len(control_i):
+                    add = control_i[i]
+                    if add is not None:
+                        img += add
+
+        img = torch.cat((img, txt), 1)
+
+        for i, block in enumerate(self.single_blocks):
+            if ("single_block", i) in blocks_replace:
+                def block_wrap(args):
+                    out = {}
+                    out["img"] = block(args["img"], vec=args["vec"], pe=args["pe"], attn_mask=args["attention_mask"])
+                    return out
+
+                out = blocks_replace[("single_block", i)]({"img": img, "vec": vec, "pe": pe, "attention_mask": attn_mask}, {"original_block": block_wrap})
+                img = out["img"]
+            else:
+                img = block(img, vec=vec, pe=pe, attn_mask=attn_mask)
+
+            if control is not None: # Controlnet
+                control_o = control.get("output")
+                if i < len(control_o):
+                    add = control_o[i]
+                    if add is not None:
+                        img[:, : img_len] += add
+
+        img = img[:, : img_len]
+
+        img = self.final_layer(img, vec)  # (N, T, patch_size ** 2 * out_channels)
+
+        shape = initial_shape[-3:]
+        for i in range(len(shape)):
+            shape[i] = shape[i] // self.patch_size[i]
+        img = img.reshape([img.shape[0]] + shape + [self.out_channels] + self.patch_size)
+        img = img.permute(0, 4, 1, 5, 2, 6, 3, 7)
+        img = img.reshape(initial_shape)
+        return img
+
+    def forward(self, x, timestep, context, y, guidance, attention_mask=None, control=None, transformer_options={}, **kwargs):
+        bs, c, t, h, w = x.shape
+        patch_size = self.patch_size
+        t_len = ((t + (patch_size[0] // 2)) // patch_size[0])
+        h_len = ((h + (patch_size[1] // 2)) // patch_size[1])
+        w_len = ((w + (patch_size[2] // 2)) // patch_size[2])
+        img_ids = torch.zeros((t_len, h_len, w_len, 3), device=x.device, dtype=x.dtype)
+        img_ids[:, :, :, 0] = img_ids[:, :, :, 0] + torch.linspace(0, t_len - 1, steps=t_len, device=x.device, dtype=x.dtype).reshape(-1, 1, 1)
+        img_ids[:, :, :, 1] = img_ids[:, :, :, 1] + torch.linspace(0, h_len - 1, steps=h_len, device=x.device, dtype=x.dtype).reshape(1, -1, 1)
+        img_ids[:, :, :, 2] = img_ids[:, :, :, 2] + torch.linspace(0, w_len - 1, steps=w_len, device=x.device, dtype=x.dtype).reshape(1, 1, -1)
+        img_ids = repeat(img_ids, "t h w c -> b (t h w) c", b=bs)
+        txt_ids = torch.zeros((bs, context.shape[1], 3), device=x.device, dtype=x.dtype)
+        out = self.forward_orig(x, img_ids, context, txt_ids, attention_mask, timestep, y, guidance, control, transformer_options)
+        return out

comfy/ldm/hydit/controlnet.py

@@ -164,9 +164,6 @@ class HunYuanControlNet(nn.Module):
             ),
         )
 
-        # Image embedding
-        num_patches = self.x_embedder.num_patches
-
         # HUnYuanDiT Blocks
         self.blocks = nn.ModuleList(
             [

comfy/ldm/hydit/models.py

@@ -248,9 +248,6 @@ class HunYuanDiT(nn.Module):
             operations.Linear(hidden_size * 4, hidden_size, bias=True, dtype=dtype, device=device),
         )
 
-        # Image embedding
-        num_patches = self.x_embedder.num_patches
-
         # HUnYuanDiT Blocks
         self.blocks = nn.ModuleList([
             HunYuanDiTBlock(hidden_size=hidden_size,

comfy/ldm/models/autoencoder.py

@@ -1,10 +1,12 @@
+import logging
+import math
 import torch
 from contextlib import contextmanager
 from typing import Any, Dict, Tuple, Union
 
 from comfy.ldm.modules.distributions.distributions import DiagonalGaussianDistribution
 
-from comfy.ldm.util import instantiate_from_config
+from comfy.ldm.util import get_obj_from_str, instantiate_from_config
 from comfy.ldm.modules.ema import LitEma
 import comfy.ops
 
@@ -52,7 +54,7 @@ class AbstractAutoencoder(torch.nn.Module):
 
         if self.use_ema:
             self.model_ema = LitEma(self, decay=ema_decay)
-            logpy.info(f"Keeping EMAs of {len(list(self.model_ema.buffers()))}.")
+            logging.info(f"Keeping EMAs of {len(list(self.model_ema.buffers()))}.")
 
     def get_input(self, batch) -> Any:
         raise NotImplementedError()
@@ -68,14 +70,14 @@ class AbstractAutoencoder(torch.nn.Module):
             self.model_ema.store(self.parameters())
             self.model_ema.copy_to(self)
             if context is not None:
-                logpy.info(f"{context}: Switched to EMA weights")
+                logging.info(f"{context}: Switched to EMA weights")
         try:
             yield None
         finally:
             if self.use_ema:
                 self.model_ema.restore(self.parameters())
                 if context is not None:
-                    logpy.info(f"{context}: Restored training weights")
+                    logging.info(f"{context}: Restored training weights")
 
     def encode(self, *args, **kwargs) -> torch.Tensor:
         raise NotImplementedError("encode()-method of abstract base class called")
@@ -84,7 +86,7 @@ class AbstractAutoencoder(torch.nn.Module):
         raise NotImplementedError("decode()-method of abstract base class called")
 
     def instantiate_optimizer_from_config(self, params, lr, cfg):
-        logpy.info(f"loading >>> {cfg['target']} <<< optimizer from config")
+        logging.info(f"loading >>> {cfg['target']} <<< optimizer from config")
         return get_obj_from_str(cfg["target"])(
             params, lr=lr, **cfg.get("params", dict())
         )
@@ -112,7 +114,7 @@ class AutoencodingEngine(AbstractAutoencoder):
 
         self.encoder: torch.nn.Module = instantiate_from_config(encoder_config)
         self.decoder: torch.nn.Module = instantiate_from_config(decoder_config)
-        self.regularization: AbstractRegularizer = instantiate_from_config(
+        self.regularization = instantiate_from_config(
             regularizer_config
         )
 
@@ -160,12 +162,19 @@ class AutoencodingEngineLegacy(AutoencodingEngine):
             },
             **kwargs,
         )
-        self.quant_conv = comfy.ops.disable_weight_init.Conv2d(
+
+        if ddconfig.get("conv3d", False):
+            conv_op = comfy.ops.disable_weight_init.Conv3d
+        else:
+            conv_op = comfy.ops.disable_weight_init.Conv2d
+
+        self.quant_conv = conv_op(
             (1 + ddconfig["double_z"]) * ddconfig["z_channels"],
             (1 + ddconfig["double_z"]) * embed_dim,
             1,
         )
-        self.post_quant_conv = comfy.ops.disable_weight_init.Conv2d(embed_dim, ddconfig["z_channels"], 1)
+
+        self.post_quant_conv = conv_op(embed_dim, ddconfig["z_channels"], 1)
         self.embed_dim = embed_dim
 
     def get_autoencoder_params(self) -> list:

comfy/ldm/modules/attention.py

@@ -157,8 +157,6 @@ def attention_sub_quad(query, key, value, heads, mask=None, attn_precision=None,
         b, _, dim_head = query.shape
         dim_head //= heads
 
-    scale = dim_head ** -0.5
-
     if skip_reshape:
         query = query.reshape(b * heads, -1, dim_head)
         value = value.reshape(b * heads, -1, dim_head)
@@ -177,9 +175,8 @@ def attention_sub_quad(query, key, value, heads, mask=None, attn_precision=None,
         bytes_per_token = torch.finfo(query.dtype).bits//8
     batch_x_heads, q_tokens, _ = query.shape
     _, _, k_tokens = key.shape
-    qk_matmul_size_bytes = batch_x_heads * bytes_per_token * q_tokens * k_tokens
 
-    mem_free_total, mem_free_torch = model_management.get_free_memory(query.device, True)
+    mem_free_total, _ = model_management.get_free_memory(query.device, True)
 
     kv_chunk_size_min = None
     kv_chunk_size = None
@@ -230,7 +227,6 @@ def attention_split(q, k, v, heads, mask=None, attn_precision=None, skip_reshape
 
     scale = dim_head ** -0.5
 
-    h = heads
     if skip_reshape:
          q, k, v = map(
             lambda t: t.reshape(b * heads, -1, dim_head),
@@ -344,12 +340,9 @@ except:
     pass
 
 def attention_xformers(q, k, v, heads, mask=None, attn_precision=None, skip_reshape=False):
-    if skip_reshape:
-        b, _, _, dim_head = q.shape
-    else:
-        b, _, dim_head = q.shape
-        dim_head //= heads
-
+    b = q.shape[0]
+    dim_head = q.shape[-1]
+    # check to make sure xformers isn't broken
     disabled_xformers = False
 
     if BROKEN_XFORMERS:
@@ -364,35 +357,44 @@ def attention_xformers(q, k, v, heads, mask=None, attn_precision=None, skip_resh
         return attention_pytorch(q, k, v, heads, mask, skip_reshape=skip_reshape)
 
     if skip_reshape:
-         q, k, v = map(
-            lambda t: t.reshape(b * heads, -1, dim_head),
+        # b h k d -> b k h d
+        q, k, v = map(
+            lambda t: t.permute(0, 2, 1, 3),
             (q, k, v),
         )
+    # actually do the reshaping
     else:
+        dim_head //= heads
         q, k, v = map(
             lambda t: t.reshape(b, -1, heads, dim_head),
             (q, k, v),
         )
 
     if mask is not None:
+        # add a singleton batch dimension
+        if mask.ndim == 2:
+            mask = mask.unsqueeze(0)
+        # add a singleton heads dimension
+        if mask.ndim == 3:
+            mask = mask.unsqueeze(1)
+        # pad to a multiple of 8
         pad = 8 - mask.shape[-1] % 8
-        mask_out = torch.empty([q.shape[0], q.shape[2], q.shape[1], mask.shape[-1] + pad], dtype=q.dtype, device=q.device)
+        # the xformers docs says that it's allowed to have a mask of shape (1, Nq, Nk)
+        # but when using separated heads, the shape has to be (B, H, Nq, Nk)
+        # in flux, this matrix ends up being over 1GB
+        # here, we create a mask with the same batch/head size as the input mask (potentially singleton or full)
+        mask_out = torch.empty([mask.shape[0], mask.shape[1], q.shape[1], mask.shape[-1] + pad], dtype=q.dtype, device=q.device)
+
         mask_out[..., :mask.shape[-1]] = mask
+        # doesn't this remove the padding again??
         mask = mask_out[..., :mask.shape[-1]]
+        mask = mask.expand(b, heads, -1, -1)
 
     out = xformers.ops.memory_efficient_attention(q, k, v, attn_bias=mask)
 
-    if skip_reshape:
-        out = (
-            out.unsqueeze(0)
-            .reshape(b, heads, -1, dim_head)
-            .permute(0, 2, 1, 3)
-            .reshape(b, -1, heads * dim_head)
-        )
-    else:
-        out = (
-            out.reshape(b, -1, heads * dim_head)
-        )
+    out = (
+        out.reshape(b, -1, heads * dim_head)
+    )
 
     return out
 
@@ -414,15 +416,34 @@ def attention_pytorch(q, k, v, heads, mask=None, attn_precision=None, skip_resha
             (q, k, v),
         )
 
-    if SDP_BATCH_LIMIT >= q.shape[0]:
+    if mask is not None:
+        # add a batch dimension if there isn't already one
+        if mask.ndim == 2:
+            mask = mask.unsqueeze(0)
+        # add a heads dimension if there isn't already one
+        if mask.ndim == 3:
+            mask = mask.unsqueeze(1)
+
+    if SDP_BATCH_LIMIT >= b:
         out = torch.nn.functional.scaled_dot_product_attention(q, k, v, attn_mask=mask, dropout_p=0.0, is_causal=False)
         out = (
             out.transpose(1, 2).reshape(b, -1, heads * dim_head)
         )
     else:
-        out = torch.empty((q.shape[0], q.shape[2], heads * dim_head), dtype=q.dtype, layout=q.layout, device=q.device)
-        for i in range(0, q.shape[0], SDP_BATCH_LIMIT):
-            out[i : i + SDP_BATCH_LIMIT] = torch.nn.functional.scaled_dot_product_attention(q[i : i + SDP_BATCH_LIMIT], k[i : i + SDP_BATCH_LIMIT], v[i : i + SDP_BATCH_LIMIT], attn_mask=mask, dropout_p=0.0, is_causal=False).transpose(1, 2).reshape(-1, q.shape[2], heads * dim_head)
+        out = torch.empty((b, q.shape[2], heads * dim_head), dtype=q.dtype, layout=q.layout, device=q.device)
+        for i in range(0, b, SDP_BATCH_LIMIT):
+            m = mask
+            if mask is not None:
+                if mask.shape[0] > 1:
+                    m = mask[i : i + SDP_BATCH_LIMIT]
+
+            out[i : i + SDP_BATCH_LIMIT] = torch.nn.functional.scaled_dot_product_attention(
+                q[i : i + SDP_BATCH_LIMIT],
+                k[i : i + SDP_BATCH_LIMIT],
+                v[i : i + SDP_BATCH_LIMIT],
+                attn_mask=m,
+                dropout_p=0.0, is_causal=False
+            ).transpose(1, 2).reshape(-1, q.shape[2], heads * dim_head)
     return out
 
 

comfy/ldm/modules/diffusionmodules/mmdit.py

@@ -1,3 +1,4 @@
+from functools import partial
 from typing import Dict, Optional, List
 
 import numpy as np
@@ -70,45 +71,33 @@ class PatchEmbed(nn.Module):
             strict_img_size: bool = True,
             dynamic_img_pad: bool = True,
             padding_mode='circular',
+            conv3d=False,
             dtype=None,
             device=None,
             operations=None,
     ):
         super().__init__()
-        self.patch_size = (patch_size, patch_size)
+        try:
+            len(patch_size)
+            self.patch_size = patch_size
+        except:
+            if conv3d:
+                self.patch_size = (patch_size, patch_size, patch_size)
+            else:
+                self.patch_size = (patch_size, patch_size)
         self.padding_mode = padding_mode
-        if img_size is not None:
-            self.img_size = (img_size, img_size)
-            self.grid_size = tuple([s // p for s, p in zip(self.img_size, self.patch_size)])
-            self.num_patches = self.grid_size[0] * self.grid_size[1]
-        else:
-            self.img_size = None
-            self.grid_size = None
-            self.num_patches = None
 
         # flatten spatial dim and transpose to channels last, kept for bwd compat
         self.flatten = flatten
         self.strict_img_size = strict_img_size
         self.dynamic_img_pad = dynamic_img_pad
-
-        self.proj = operations.Conv2d(in_chans, embed_dim, kernel_size=patch_size, stride=patch_size, bias=bias, dtype=dtype, device=device)
+        if conv3d:
+            self.proj = operations.Conv3d(in_chans, embed_dim, kernel_size=patch_size, stride=patch_size, bias=bias, dtype=dtype, device=device)
+        else:
+            self.proj = operations.Conv2d(in_chans, embed_dim, kernel_size=patch_size, stride=patch_size, bias=bias, dtype=dtype, device=device)
         self.norm = norm_layer(embed_dim) if norm_layer else nn.Identity()
 
     def forward(self, x):
-        # B, C, H, W = x.shape
-        # if self.img_size is not None:
-        #     if self.strict_img_size:
-        #         _assert(H == self.img_size[0], f"Input height ({H}) doesn't match model ({self.img_size[0]}).")
-        #         _assert(W == self.img_size[1], f"Input width ({W}) doesn't match model ({self.img_size[1]}).")
-        #     elif not self.dynamic_img_pad:
-        #         _assert(
-        #             H % self.patch_size[0] == 0,
-        #             f"Input height ({H}) should be divisible by patch size ({self.patch_size[0]})."
-        #         )
-        #         _assert(
-        #             W % self.patch_size[1] == 0,
-        #             f"Input width ({W}) should be divisible by patch size ({self.patch_size[1]})."
-        #         )
         if self.dynamic_img_pad:
             x = comfy.ldm.common_dit.pad_to_patch_size(x, self.patch_size, padding_mode=self.padding_mode)
         x = self.proj(x)

comfy/ldm/modules/diffusionmodules/model.py

@@ -43,51 +43,100 @@ def Normalize(in_channels, num_groups=32):
     return ops.GroupNorm(num_groups=num_groups, num_channels=in_channels, eps=1e-6, affine=True)
 
 
+class VideoConv3d(nn.Module):
+    def __init__(self, n_channels, out_channels, kernel_size, stride=1, dilation=1, padding_mode='replicate', padding=1, **kwargs):
+        super().__init__()
+
+        self.padding_mode = padding_mode
+        if padding != 0:
+            padding = (padding, padding, padding, padding, kernel_size - 1, 0)
+        else:
+            kwargs["padding"] = padding
+
+        self.padding = padding
+        self.conv = ops.Conv3d(n_channels, out_channels, kernel_size, stride=stride, dilation=dilation, **kwargs)
+
+    def forward(self, x):
+        if self.padding != 0:
+            x = torch.nn.functional.pad(x, self.padding, mode=self.padding_mode)
+        return self.conv(x)
+
+def interpolate_up(x, scale_factor):
+    try:
+        return torch.nn.functional.interpolate(x, scale_factor=scale_factor, mode="nearest")
+    except: #operation not implemented for bf16
+        orig_shape = list(x.shape)
+        out_shape = orig_shape[:2]
+        for i in range(len(orig_shape) - 2):
+            out_shape.append(round(orig_shape[i + 2] * scale_factor[i]))
+        out = torch.empty(out_shape, dtype=x.dtype, layout=x.layout, device=x.device)
+        split = 8
+        l = out.shape[1] // split
+        for i in range(0, out.shape[1], l):
+            out[:,i:i+l] = torch.nn.functional.interpolate(x[:,i:i+l].to(torch.float32), scale_factor=scale_factor, mode="nearest").to(x.dtype)
+        return out
+
 class Upsample(nn.Module):
-    def __init__(self, in_channels, with_conv):
+    def __init__(self, in_channels, with_conv, conv_op=ops.Conv2d, scale_factor=2.0):
         super().__init__()
         self.with_conv = with_conv
+        self.scale_factor = scale_factor
+
         if self.with_conv:
-            self.conv = ops.Conv2d(in_channels,
+            self.conv = conv_op(in_channels,
                                         in_channels,
                                         kernel_size=3,
                                         stride=1,
                                         padding=1)
 
     def forward(self, x):
-        try:
-            x = torch.nn.functional.interpolate(x, scale_factor=2.0, mode="nearest")
-        except: #operation not implemented for bf16
-            b, c, h, w = x.shape
-            out = torch.empty((b, c, h*2, w*2), dtype=x.dtype, layout=x.layout, device=x.device)
-            split = 8
-            l = out.shape[1] // split
-            for i in range(0, out.shape[1], l):
-                out[:,i:i+l] = torch.nn.functional.interpolate(x[:,i:i+l].to(torch.float32), scale_factor=2.0, mode="nearest").to(x.dtype)
-            del x
-            x = out
+        scale_factor = self.scale_factor
+        if not isinstance(scale_factor, tuple):
+            scale_factor = (scale_factor,) * (x.ndim - 2)
 
+        if x.ndim == 5 and scale_factor[0] > 1.0:
+            t = x.shape[2]
+            if t > 1:
+                a, b = x.split((1, t - 1), dim=2)
+                del x
+                b = interpolate_up(b, scale_factor)
+            else:
+                a = x
+
+            a = interpolate_up(a.squeeze(2), scale_factor=scale_factor[1:]).unsqueeze(2)
+            if t > 1:
+                x = torch.cat((a, b), dim=2)
+            else:
+                x = a
+        else:
+            x = interpolate_up(x, self.scale_factor)
         if self.with_conv:
             x = self.conv(x)
         return x
 
 
 class Downsample(nn.Module):
-    def __init__(self, in_channels, with_conv):
+    def __init__(self, in_channels, with_conv, stride=2, conv_op=ops.Conv2d):
         super().__init__()
         self.with_conv = with_conv
         if self.with_conv:
             # no asymmetric padding in torch conv, must do it ourselves
-            self.conv = ops.Conv2d(in_channels,
+            self.conv = conv_op(in_channels,
                                         in_channels,
                                         kernel_size=3,
-                                        stride=2,
+                                        stride=stride,
                                         padding=0)
 
     def forward(self, x):
         if self.with_conv:
-            pad = (0,1,0,1)
-            x = torch.nn.functional.pad(x, pad, mode="constant", value=0)
+            if x.ndim == 4:
+                pad = (0, 1, 0, 1)
+                mode = "constant"
+                x = torch.nn.functional.pad(x, pad, mode=mode, value=0)
+            elif x.ndim == 5:
+                pad = (1, 1, 1, 1, 2, 0)
+                mode = "replicate"
+                x = torch.nn.functional.pad(x, pad, mode=mode)
             x = self.conv(x)
         else:
             x = torch.nn.functional.avg_pool2d(x, kernel_size=2, stride=2)
@@ -96,7 +145,7 @@ class Downsample(nn.Module):
 
 class ResnetBlock(nn.Module):
     def __init__(self, *, in_channels, out_channels=None, conv_shortcut=False,
-                 dropout, temb_channels=512):
+                 dropout, temb_channels=512, conv_op=ops.Conv2d):
         super().__init__()
         self.in_channels = in_channels
         out_channels = in_channels if out_channels is None else out_channels
@@ -105,7 +154,7 @@ class ResnetBlock(nn.Module):
 
         self.swish = torch.nn.SiLU(inplace=True)
         self.norm1 = Normalize(in_channels)
-        self.conv1 = ops.Conv2d(in_channels,
+        self.conv1 = conv_op(in_channels,
                                      out_channels,
                                      kernel_size=3,
                                      stride=1,
@@ -115,20 +164,20 @@ class ResnetBlock(nn.Module):
                                              out_channels)
         self.norm2 = Normalize(out_channels)
         self.dropout = torch.nn.Dropout(dropout, inplace=True)
-        self.conv2 = ops.Conv2d(out_channels,
+        self.conv2 = conv_op(out_channels,
                                      out_channels,
                                      kernel_size=3,
                                      stride=1,
                                      padding=1)
         if self.in_channels != self.out_channels:
             if self.use_conv_shortcut:
-                self.conv_shortcut = ops.Conv2d(in_channels,
+                self.conv_shortcut = conv_op(in_channels,
                                                      out_channels,
                                                      kernel_size=3,
                                                      stride=1,
                                                      padding=1)
             else:
-                self.nin_shortcut = ops.Conv2d(in_channels,
+                self.nin_shortcut = conv_op(in_channels,
                                                     out_channels,
                                                     kernel_size=1,
                                                     stride=1,
@@ -162,7 +211,6 @@ def slice_attention(q, k, v):
 
     mem_free_total = model_management.get_free_memory(q.device)
 
-    gb = 1024 ** 3
     tensor_size = q.shape[0] * q.shape[1] * k.shape[2] * q.element_size()
     modifier = 3 if q.element_size() == 2 else 2.5
     mem_required = tensor_size * modifier
@@ -195,21 +243,25 @@ def slice_attention(q, k, v):
 
 def normal_attention(q, k, v):
     # compute attention
-    b,c,h,w = q.shape
+    orig_shape = q.shape
+    b = orig_shape[0]
+    c = orig_shape[1]
 
-    q = q.reshape(b,c,h*w)
-    q = q.permute(0,2,1)   # b,hw,c
-    k = k.reshape(b,c,h*w) # b,c,hw
-    v = v.reshape(b,c,h*w)
+    q = q.reshape(b, c, -1)
+    q = q.permute(0, 2, 1)   # b,hw,c
+    k = k.reshape(b, c, -1) # b,c,hw
+    v = v.reshape(b, c, -1)
 
     r1 = slice_attention(q, k, v)
-    h_ = r1.reshape(b,c,h,w)
+    h_ = r1.reshape(orig_shape)
     del r1
     return h_
 
 def xformers_attention(q, k, v):
     # compute attention
-    B, C, H, W = q.shape
+    orig_shape = q.shape
+    B = orig_shape[0]
+    C = orig_shape[1]
     q, k, v = map(
         lambda t: t.view(B, C, -1).transpose(1, 2).contiguous(),
         (q, k, v),
@@ -217,14 +269,16 @@ def xformers_attention(q, k, v):
 
     try:
         out = xformers.ops.memory_efficient_attention(q, k, v, attn_bias=None)
-        out = out.transpose(1, 2).reshape(B, C, H, W)
-    except NotImplementedError as e:
-        out = slice_attention(q.view(B, -1, C), k.view(B, -1, C).transpose(1, 2), v.view(B, -1, C).transpose(1, 2)).reshape(B, C, H, W)
+        out = out.transpose(1, 2).reshape(orig_shape)
+    except NotImplementedError:
+        out = slice_attention(q.view(B, -1, C), k.view(B, -1, C).transpose(1, 2), v.view(B, -1, C).transpose(1, 2)).reshape(orig_shape)
     return out
 
 def pytorch_attention(q, k, v):
     # compute attention
-    B, C, H, W = q.shape
+    orig_shape = q.shape
+    B = orig_shape[0]
+    C = orig_shape[1]
     q, k, v = map(
         lambda t: t.view(B, 1, C, -1).transpose(2, 3).contiguous(),
         (q, k, v),
@@ -232,35 +286,35 @@ def pytorch_attention(q, k, v):
 
     try:
         out = torch.nn.functional.scaled_dot_product_attention(q, k, v, attn_mask=None, dropout_p=0.0, is_causal=False)
-        out = out.transpose(2, 3).reshape(B, C, H, W)
-    except model_management.OOM_EXCEPTION as e:
+        out = out.transpose(2, 3).reshape(orig_shape)
+    except model_management.OOM_EXCEPTION:
         logging.warning("scaled_dot_product_attention OOMed: switched to slice attention")
-        out = slice_attention(q.view(B, -1, C), k.view(B, -1, C).transpose(1, 2), v.view(B, -1, C).transpose(1, 2)).reshape(B, C, H, W)
+        out = slice_attention(q.view(B, -1, C), k.view(B, -1, C).transpose(1, 2), v.view(B, -1, C).transpose(1, 2)).reshape(orig_shape)
     return out
 
 
 class AttnBlock(nn.Module):
-    def __init__(self, in_channels):
+    def __init__(self, in_channels, conv_op=ops.Conv2d):
         super().__init__()
         self.in_channels = in_channels
 
         self.norm = Normalize(in_channels)
-        self.q = ops.Conv2d(in_channels,
+        self.q = conv_op(in_channels,
                                  in_channels,
                                  kernel_size=1,
                                  stride=1,
                                  padding=0)
-        self.k = ops.Conv2d(in_channels,
+        self.k = conv_op(in_channels,
                                  in_channels,
                                  kernel_size=1,
                                  stride=1,
                                  padding=0)
-        self.v = ops.Conv2d(in_channels,
+        self.v = conv_op(in_channels,
                                  in_channels,
                                  kernel_size=1,
                                  stride=1,
                                  padding=0)
-        self.proj_out = ops.Conv2d(in_channels,
+        self.proj_out = conv_op(in_channels,
                                         in_channels,
                                         kernel_size=1,
                                         stride=1,
@@ -290,8 +344,8 @@ class AttnBlock(nn.Module):
         return x+h_
 
 
-def make_attn(in_channels, attn_type="vanilla", attn_kwargs=None):
-    return AttnBlock(in_channels)
+def make_attn(in_channels, attn_type="vanilla", attn_kwargs=None, conv_op=ops.Conv2d):
+    return AttnBlock(in_channels, conv_op=conv_op)
 
 
 class Model(nn.Module):
@@ -450,6 +504,7 @@ class Encoder(nn.Module):
     def __init__(self, *, ch, out_ch, ch_mult=(1,2,4,8), num_res_blocks,
                  attn_resolutions, dropout=0.0, resamp_with_conv=True, in_channels,
                  resolution, z_channels, double_z=True, use_linear_attn=False, attn_type="vanilla",
+                 conv3d=False, time_compress=None,
                  **ignore_kwargs):
         super().__init__()
         if use_linear_attn: attn_type = "linear"
@@ -460,8 +515,15 @@ class Encoder(nn.Module):
         self.resolution = resolution
         self.in_channels = in_channels
 
+        if conv3d:
+            conv_op = VideoConv3d
+            mid_attn_conv_op = ops.Conv3d
+        else:
+            conv_op = ops.Conv2d
+            mid_attn_conv_op = ops.Conv2d
+
         # downsampling
-        self.conv_in = ops.Conv2d(in_channels,
+        self.conv_in = conv_op(in_channels,
                                        self.ch,
                                        kernel_size=3,
                                        stride=1,
@@ -480,15 +542,20 @@ class Encoder(nn.Module):
                 block.append(ResnetBlock(in_channels=block_in,
                                          out_channels=block_out,
                                          temb_channels=self.temb_ch,
-                                         dropout=dropout))
+                                         dropout=dropout,
+                                         conv_op=conv_op))
                 block_in = block_out
                 if curr_res in attn_resolutions:
-                    attn.append(make_attn(block_in, attn_type=attn_type))
+                    attn.append(make_attn(block_in, attn_type=attn_type, conv_op=conv_op))
             down = nn.Module()
             down.block = block
             down.attn = attn
             if i_level != self.num_resolutions-1:
-                down.downsample = Downsample(block_in, resamp_with_conv)
+                stride = 2
+                if time_compress is not None:
+                    if (self.num_resolutions - 1 - i_level) > math.log2(time_compress):
+                        stride = (1, 2, 2)
+                down.downsample = Downsample(block_in, resamp_with_conv, stride=stride, conv_op=conv_op)
                 curr_res = curr_res // 2
             self.down.append(down)
 
@@ -497,16 +564,18 @@ class Encoder(nn.Module):
         self.mid.block_1 = ResnetBlock(in_channels=block_in,
                                        out_channels=block_in,
                                        temb_channels=self.temb_ch,
-                                       dropout=dropout)
-        self.mid.attn_1 = make_attn(block_in, attn_type=attn_type)
+                                       dropout=dropout,
+                                       conv_op=conv_op)
+        self.mid.attn_1 = make_attn(block_in, attn_type=attn_type, conv_op=mid_attn_conv_op)
         self.mid.block_2 = ResnetBlock(in_channels=block_in,
                                        out_channels=block_in,
                                        temb_channels=self.temb_ch,
-                                       dropout=dropout)
+                                       dropout=dropout,
+                                       conv_op=conv_op)
 
         # end
         self.norm_out = Normalize(block_in)
-        self.conv_out = ops.Conv2d(block_in,
+        self.conv_out = conv_op(block_in,
                                         2*z_channels if double_z else z_channels,
                                         kernel_size=3,
                                         stride=1,
@@ -544,9 +613,10 @@ class Decoder(nn.Module):
                  conv_out_op=ops.Conv2d,
                  resnet_op=ResnetBlock,
                  attn_op=AttnBlock,
+                 conv3d=False,
+                 time_compress=None,
                 **ignorekwargs):
         super().__init__()
-        if use_linear_attn: attn_type = "linear"
         self.ch = ch
         self.temb_ch = 0
         self.num_resolutions = len(ch_mult)
@@ -556,8 +626,15 @@ class Decoder(nn.Module):
         self.give_pre_end = give_pre_end
         self.tanh_out = tanh_out
 
-        # compute in_ch_mult, block_in and curr_res at lowest res
-        in_ch_mult = (1,)+tuple(ch_mult)
+        if conv3d:
+            conv_op = VideoConv3d
+            conv_out_op = VideoConv3d
+            mid_attn_conv_op = ops.Conv3d
+        else:
+            conv_op = ops.Conv2d
+            mid_attn_conv_op = ops.Conv2d
+
+        # compute block_in and curr_res at lowest res
         block_in = ch*ch_mult[self.num_resolutions-1]
         curr_res = resolution // 2**(self.num_resolutions-1)
         self.z_shape = (1,z_channels,curr_res,curr_res)
@@ -565,7 +642,7 @@ class Decoder(nn.Module):
             self.z_shape, np.prod(self.z_shape)))
 
         # z to block_in
-        self.conv_in = ops.Conv2d(z_channels,
+        self.conv_in = conv_op(z_channels,
                                        block_in,
                                        kernel_size=3,
                                        stride=1,
@@ -576,12 +653,14 @@ class Decoder(nn.Module):
         self.mid.block_1 = resnet_op(in_channels=block_in,
                                        out_channels=block_in,
                                        temb_channels=self.temb_ch,
-                                       dropout=dropout)
-        self.mid.attn_1 = attn_op(block_in)
+                                       dropout=dropout,
+                                       conv_op=conv_op)
+        self.mid.attn_1 = attn_op(block_in, conv_op=mid_attn_conv_op)
         self.mid.block_2 = resnet_op(in_channels=block_in,
                                        out_channels=block_in,
                                        temb_channels=self.temb_ch,
-                                       dropout=dropout)
+                                       dropout=dropout,
+                                       conv_op=conv_op)
 
         # upsampling
         self.up = nn.ModuleList()
@@ -593,15 +672,21 @@ class Decoder(nn.Module):
                 block.append(resnet_op(in_channels=block_in,
                                          out_channels=block_out,
                                          temb_channels=self.temb_ch,
-                                         dropout=dropout))
+                                         dropout=dropout,
+                                         conv_op=conv_op))
                 block_in = block_out
                 if curr_res in attn_resolutions:
-                    attn.append(attn_op(block_in))
+                    attn.append(attn_op(block_in, conv_op=conv_op))
             up = nn.Module()
             up.block = block
             up.attn = attn
             if i_level != 0:
-                up.upsample = Upsample(block_in, resamp_with_conv)
+                scale_factor = 2.0
+                if time_compress is not None:
+                    if i_level > math.log2(time_compress):
+                        scale_factor = (1.0, 2.0, 2.0)
+
+                up.upsample = Upsample(block_in, resamp_with_conv, conv_op=conv_op, scale_factor=scale_factor)
                 curr_res = curr_res * 2
             self.up.insert(0, up) # prepend to get consistent order
 

comfy/ldm/modules/sub_quadratic_attention.py

@@ -22,7 +22,6 @@ except ImportError:
 	from typing import Optional, NamedTuple, List
 	from typing_extensions import Protocol
 
-from torch import Tensor
 from typing import List
 
 from comfy import model_management
@@ -172,7 +171,7 @@ def _get_attention_scores_no_kv_chunking(
         del attn_scores
     except model_management.OOM_EXCEPTION:
         logging.warning("ran out of memory while running softmax in  _get_attention_scores_no_kv_chunking, trying slower in place softmax instead")
-        attn_scores -= attn_scores.max(dim=-1, keepdim=True).values
+        attn_scores -= attn_scores.max(dim=-1, keepdim=True).values # noqa: F821 attn_scores is not defined
         torch.exp(attn_scores, out=attn_scores)
         summed = torch.sum(attn_scores, dim=-1, keepdim=True)
         attn_scores /= summed

comfy/ldm/modules/temporal_ae.py

@@ -194,6 +194,7 @@ def make_time_attn(
     attn_kwargs=None,
     alpha: float = 0,
     merge_strategy: str = "learned",
+    conv_op=ops.Conv2d,
 ):
     return partialclass(
         AttnVideoBlock, in_channels, alpha=alpha, merge_strategy=merge_strategy

comfy/ldm/util.py

@@ -133,7 +133,6 @@ class AdamWwithEMAandWings(optim.Optimizer):
             exp_avgs = []
             exp_avg_sqs = []
             ema_params_with_grad = []
-            state_sums = []
             max_exp_avg_sqs = []
             state_steps = []
             amsgrad = group['amsgrad']

comfy/model_base.py

@@ -31,6 +31,7 @@ import comfy.ldm.audio.dit
 import comfy.ldm.audio.embedders
 import comfy.ldm.flux.model
 import comfy.ldm.lightricks.model
+import comfy.ldm.hunyuan_video.model
 
 import comfy.model_management
 import comfy.patcher_extension
@@ -427,7 +428,6 @@ class SVD_img2vid(BaseModel):
 
         latent_image = kwargs.get("concat_latent_image", None)
         noise = kwargs.get("noise", None)
-        device = kwargs["device"]
 
         if latent_image is None:
             latent_image = torch.zeros_like(noise)
@@ -687,6 +687,7 @@ class StableAudio1(BaseModel):
                 sd["{}{}".format(k, l)] = s[l]
         return sd
 
+
 class HunyuanDiT(BaseModel):
     def __init__(self, model_config, model_type=ModelType.V_PREDICTION, device=None):
         super().__init__(model_config, model_type, device=device, unet_model=comfy.ldm.hydit.models.HunYuanDiT)
@@ -711,8 +712,6 @@ class HunyuanDiT(BaseModel):
 
         width = kwargs.get("width", 768)
         height = kwargs.get("height", 768)
-        crop_w = kwargs.get("crop_w", 0)
-        crop_h = kwargs.get("crop_h", 0)
         target_width = kwargs.get("target_width", width)
         target_height = kwargs.get("target_height", height)
 
@@ -769,6 +768,16 @@ class Flux(BaseModel):
         cross_attn = kwargs.get("cross_attn", None)
         if cross_attn is not None:
             out['c_crossattn'] = comfy.conds.CONDRegular(cross_attn)
+        # upscale the attention mask, since now we 
+        attention_mask = kwargs.get("attention_mask", None)
+        if attention_mask is not None:
+            shape = kwargs["noise"].shape
+            mask_ref_size = kwargs["attention_mask_img_shape"]
+            # the model will pad to the patch size, and then divide
+            # essentially dividing and rounding up
+            (h_tok, w_tok) = (math.ceil(shape[2] / self.diffusion_model.patch_size), math.ceil(shape[3] / self.diffusion_model.patch_size))
+            attention_mask = utils.upscale_dit_mask(attention_mask, mask_ref_size, (h_tok, w_tok))
+            out['attention_mask'] = comfy.conds.CONDRegular(attention_mask)
         out['guidance'] = comfy.conds.CONDRegular(torch.FloatTensor([kwargs.get("guidance", 3.5)]))
         return out
 
@@ -810,3 +819,21 @@ class LTXV(BaseModel):
 
         out['frame_rate'] = comfy.conds.CONDConstant(kwargs.get("frame_rate", 25))
         return out
+
+class HunyuanVideo(BaseModel):
+    def __init__(self, model_config, model_type=ModelType.FLOW, device=None):
+        super().__init__(model_config, model_type, device=device, unet_model=comfy.ldm.hunyuan_video.model.HunyuanVideo)
+
+    def encode_adm(self, **kwargs):
+        return kwargs["pooled_output"]
+
+    def extra_conds(self, **kwargs):
+        out = super().extra_conds(**kwargs)
+        attention_mask = kwargs.get("attention_mask", None)
+        if attention_mask is not None:
+            out['attention_mask'] = comfy.conds.CONDRegular(attention_mask)
+        cross_attn = kwargs.get("cross_attn", None)
+        if cross_attn is not None:
+            out['c_crossattn'] = comfy.conds.CONDRegular(cross_attn)
+        out['guidance'] = comfy.conds.CONDRegular(torch.FloatTensor([kwargs.get("guidance", 6.0)]))
+        return out

comfy/model_detection.py

@@ -133,6 +133,26 @@ def detect_unet_config(state_dict, key_prefix):
             unet_config["image_model"] = "hydit1"
         return unet_config
 
+    if '{}txt_in.individual_token_refiner.blocks.0.norm1.weight'.format(key_prefix) in state_dict_keys: #Hunyuan Video
+        dit_config = {}
+        dit_config["image_model"] = "hunyuan_video"
+        dit_config["in_channels"] = 16
+        dit_config["patch_size"] = [1, 2, 2]
+        dit_config["out_channels"] = 16
+        dit_config["vec_in_dim"] = 768
+        dit_config["context_in_dim"] = 4096
+        dit_config["hidden_size"] = 3072
+        dit_config["mlp_ratio"] = 4.0
+        dit_config["num_heads"] = 24
+        dit_config["depth"] = count_blocks(state_dict_keys, '{}double_blocks.'.format(key_prefix) + '{}.')
+        dit_config["depth_single_blocks"] = count_blocks(state_dict_keys, '{}single_blocks.'.format(key_prefix) + '{}.')
+        dit_config["axes_dim"] = [16, 56, 56]
+        dit_config["theta"] = 256
+        dit_config["qkv_bias"] = True
+        guidance_keys = list(filter(lambda a: a.startswith("{}guidance_in.".format(key_prefix)), state_dict_keys))
+        dit_config["guidance_embed"] = len(guidance_keys) > 0
+        return dit_config
+
     if '{}double_blocks.0.img_attn.norm.key_norm.scale'.format(key_prefix) in state_dict_keys: #Flux
         dit_config = {}
         dit_config["image_model"] = "flux"
@@ -216,7 +236,6 @@ def detect_unet_config(state_dict, key_prefix):
 
     num_res_blocks = []
     channel_mult = []
-    attention_resolutions = []
     transformer_depth = []
     transformer_depth_output = []
     context_dim = None
@@ -388,7 +407,6 @@ def convert_config(unet_config):
             t_out += [d] * (res + 1)
             s *= 2
         transformer_depth = t_in
-        transformer_depth_output = t_out
         new_config["transformer_depth"] = t_in
         new_config["transformer_depth_output"] = t_out
         new_config["transformer_depth_middle"] = transformer_depth_middle

comfy/model_management.py

@@ -512,7 +512,7 @@ def load_models_gpu(models, memory_required=0, force_patch_weights=False, minimu
         if vram_set_state == VRAMState.NO_VRAM:
             lowvram_model_memory = 64 * 1024 * 1024
 
-        cur_loaded_model = loaded_model.model_load(lowvram_model_memory, force_patch_weights=force_patch_weights)
+        loaded_model.model_load(lowvram_model_memory, force_patch_weights=force_patch_weights)
         current_loaded_models.insert(0, loaded_model)
     return
 
@@ -581,7 +581,7 @@ def unet_offload_device():
 
 def unet_inital_load_device(parameters, dtype):
     torch_dev = get_torch_device()
-    if vram_state == VRAMState.HIGH_VRAM:
+    if vram_state == VRAMState.HIGH_VRAM or vram_state == VRAMState.SHARED:
         return torch_dev
 
     cpu_dev = torch.device("cpu")
@@ -695,7 +695,7 @@ def text_encoder_initial_device(load_device, offload_device, model_size=0):
         return offload_device
 
     if is_device_mps(load_device):
-        return offload_device
+        return load_device
 
     mem_l = get_free_memory(load_device)
     mem_o = get_free_memory(offload_device)

comfy/patcher_extension.py

@@ -113,7 +113,7 @@ class WrapperExecutor:
     def _create_next_executor(self) -> 'WrapperExecutor':
         new_idx = self.idx + 1
         if new_idx > len(self.wrappers):
-            raise Exception(f"Wrapper idx exceeded available wrappers; something went very wrong.")
+            raise Exception("Wrapper idx exceeded available wrappers; something went very wrong.")
         if self.class_obj is None:
             return WrapperExecutor.new_executor(self.original, self.wrappers, new_idx)
         return WrapperExecutor.new_class_executor(self.original, self.class_obj, self.wrappers, new_idx)

comfy/sampler_helpers.py

@@ -103,7 +103,6 @@ def cleanup_additional_models(models):
 
 
 def prepare_sampling(model: 'ModelPatcher', noise_shape, conds):
-    device = model.load_device
     real_model: 'BaseModel' = None
     models, inference_memory = get_additional_models(conds, model.model_dtype())
     models += model.get_nested_additional_models()  # TODO: does this require inference_memory update?

comfy/samplers.py

@@ -130,11 +130,6 @@ def can_concat_cond(c1, c2):
     return cond_equal_size(c1.conditioning, c2.conditioning)
 
 def cond_cat(c_list):
-    c_crossattn = []
-    c_concat = []
-    c_adm = []
-    crossattn_max_len = 0
-
     temp = {}
     for x in c_list:
         for k in x:
@@ -608,8 +603,6 @@ def pre_run_control(model, conds):
     for t in range(len(conds)):
         x = conds[t]
 
-        timestep_start = None
-        timestep_end = None
         percent_to_timestep_function = lambda a: s.percent_to_sigma(a)
         if 'control' in x:
             x['control'].pre_run(model, percent_to_timestep_function)

comfy/sd.py

@@ -31,6 +31,7 @@ import comfy.text_encoders.flux
 import comfy.text_encoders.long_clipl
 import comfy.text_encoders.genmo
 import comfy.text_encoders.lt
+import comfy.text_encoders.hunyuan_video
 
 import comfy.model_patcher
 import comfy.lora
@@ -306,12 +307,23 @@ class VAE:
                     self.upscale_ratio = 4
 
                 self.latent_channels = ddconfig['z_channels'] = sd["decoder.conv_in.weight"].shape[1]
-                if 'quant_conv.weight' in sd:
-                    self.first_stage_model = AutoencoderKL(ddconfig=ddconfig, embed_dim=4)
+                if 'post_quant_conv.weight' in sd:
+                    self.first_stage_model = AutoencoderKL(ddconfig=ddconfig, embed_dim=sd['post_quant_conv.weight'].shape[1])
                 else:
                     self.first_stage_model = AutoencodingEngine(regularizer_config={'target': "comfy.ldm.models.autoencoder.DiagonalGaussianRegularizer"},
                                                                 encoder_config={'target': "comfy.ldm.modules.diffusionmodules.model.Encoder", 'params': ddconfig},
                                                                 decoder_config={'target': "comfy.ldm.modules.diffusionmodules.model.Decoder", 'params': ddconfig})
+            elif "decoder.conv_in.conv.weight" in sd:
+                ddconfig = {'double_z': True, 'z_channels': 4, 'resolution': 256, 'in_channels': 3, 'out_ch': 3, 'ch': 128, 'ch_mult': [1, 2, 4, 4], 'num_res_blocks': 2, 'attn_resolutions': [], 'dropout': 0.0}
+                ddconfig["conv3d"] = True
+                ddconfig["time_compress"] = 4
+                self.upscale_ratio = (lambda a: max(0, a * 4 - 3), 8, 8)
+                self.latent_dim = 3
+                self.latent_channels = ddconfig['z_channels'] = sd["decoder.conv_in.conv.weight"].shape[1]
+                self.first_stage_model = AutoencoderKL(ddconfig=ddconfig, embed_dim=sd['post_quant_conv.weight'].shape[1])
+                self.memory_used_decode = lambda shape, dtype: (1500 * shape[2] * shape[3] * shape[4] * (4 * 8 * 8)) * model_management.dtype_size(dtype)
+                self.memory_used_encode = lambda shape, dtype: (900 * max(shape[2], 2) * shape[3] * shape[4]) * model_management.dtype_size(dtype)
+
             elif "decoder.layers.1.layers.0.beta" in sd:
                 self.first_stage_model = AudioOobleckVAE()
                 self.memory_used_encode = lambda shape, dtype: (1000 * shape[2]) * model_management.dtype_size(dtype)
@@ -435,7 +447,7 @@ class VAE:
                 if pixel_samples is None:
                     pixel_samples = torch.empty((samples_in.shape[0],) + tuple(out.shape[1:]), device=self.output_device)
                 pixel_samples[x:x+batch_number] = out
-        except model_management.OOM_EXCEPTION as e:
+        except model_management.OOM_EXCEPTION:
             logging.warning("Warning: Ran out of memory when regular VAE decoding, retrying with tiled VAE decoding.")
             dims = samples_in.ndim - 2
             if dims == 1:
@@ -490,7 +502,7 @@ class VAE:
                     samples = torch.empty((pixel_samples.shape[0],) + tuple(out.shape[1:]), device=self.output_device)
                 samples[x:x + batch_number] = out
 
-        except model_management.OOM_EXCEPTION as e:
+        except model_management.OOM_EXCEPTION:
             logging.warning("Warning: Ran out of memory when regular VAE encoding, retrying with tiled VAE encoding.")
             if len(pixel_samples.shape) == 3:
                 samples = self.encode_tiled_1d(pixel_samples)
@@ -544,6 +556,7 @@ class CLIPType(Enum):
     FLUX = 6
     MOCHI = 7
     LTXV = 8
+    HUNYUAN_VIDEO = 9
 
 def load_clip(ckpt_paths, embedding_directory=None, clip_type=CLIPType.STABLE_DIFFUSION, model_options={}):
     clip_data = []
@@ -559,6 +572,7 @@ class TEModel(Enum):
     T5_XXL = 4
     T5_XL = 5
     T5_BASE = 6
+    LLAMA3_8 = 7
 
 def detect_te_model(sd):
     if "text_model.encoder.layers.30.mlp.fc1.weight" in sd:
@@ -575,6 +589,8 @@ def detect_te_model(sd):
             return TEModel.T5_XL
     if "encoder.block.0.layer.0.SelfAttention.k.weight" in sd:
         return TEModel.T5_BASE
+    if "model.layers.0.post_attention_layernorm.weight" in sd:
+        return TEModel.LLAMA3_8
     return None
 
 
@@ -652,6 +668,9 @@ def load_text_encoder_state_dicts(state_dicts=[], embedding_directory=None, clip
         elif clip_type == CLIPType.FLUX:
             clip_target.clip = comfy.text_encoders.flux.flux_clip(**t5xxl_detect(clip_data))
             clip_target.tokenizer = comfy.text_encoders.flux.FluxTokenizer
+        elif clip_type == CLIPType.HUNYUAN_VIDEO:
+            clip_target.clip = comfy.text_encoders.hunyuan_video.hunyuan_video_clip() #TODO
+            clip_target.tokenizer = comfy.text_encoders.hunyuan_video.HunyuanVideoTokenizer
         else:
             clip_target.clip = sdxl_clip.SDXLClipModel
             clip_target.tokenizer = sdxl_clip.SDXLTokenizer
@@ -691,7 +710,6 @@ def load_checkpoint(config_path=None, ckpt_path=None, output_vae=True, output_cl
             config = yaml.safe_load(stream)
     model_config_params = config['model']['params']
     clip_config = model_config_params['cond_stage_config']
-    scale_factor = model_config_params['scale_factor']
 
     if "parameterization" in model_config_params:
         if model_config_params["parameterization"] == "v":

comfy/sd1_clip.py

@@ -336,7 +336,6 @@ def expand_directory_list(directories):
     return list(dirs)
 
 def bundled_embed(embed, prefix, suffix): #bundled embedding in lora format
-    i = 0
     out_list = []
     for k in embed:
         if k.startswith(prefix) and k.endswith(suffix):
@@ -392,7 +391,7 @@ def load_embed(embedding_name, embedding_directory, embedding_size, embed_key=No
                     embed_out = safe_load_embed_zip(embed_path)
             else:
                 embed = torch.load(embed_path, map_location="cpu")
-    except Exception as e:
+    except Exception:
         logging.warning("{}\n\nerror loading embedding, skipping loading: {}".format(traceback.format_exc(), embedding_name))
         return None
 
@@ -421,7 +420,7 @@ def load_embed(embedding_name, embedding_directory, embedding_size, embed_key=No
     return embed_out
 
 class SDTokenizer:
-    def __init__(self, tokenizer_path=None, max_length=77, pad_with_end=True, embedding_directory=None, embedding_size=768, embedding_key='clip_l', tokenizer_class=CLIPTokenizer, has_start_token=True, has_end_token=True, pad_to_max_length=True, min_length=None, pad_token=None, tokenizer_data={}):
+    def __init__(self, tokenizer_path=None, max_length=77, pad_with_end=True, embedding_directory=None, embedding_size=768, embedding_key='clip_l', tokenizer_class=CLIPTokenizer, has_start_token=True, has_end_token=True, pad_to_max_length=True, min_length=None, pad_token=None, end_token=None, tokenizer_data={}):
         if tokenizer_path is None:
             tokenizer_path = os.path.join(os.path.dirname(os.path.realpath(__file__)), "sd1_tokenizer")
         self.tokenizer = tokenizer_class.from_pretrained(tokenizer_path)
@@ -434,11 +433,16 @@ class SDTokenizer:
             self.tokens_start = 1
             self.start_token = empty[0]
             if has_end_token:
-                self.end_token = empty[1]
+                if end_token is not None:
+                    self.end_token = end_token
+                else:
+                    self.end_token = empty[1]
         else:
             self.tokens_start = 0
             self.start_token = None
-            if has_end_token:
+            if end_token is not None:
+                self.end_token = end_token
+            else:
                 self.end_token = empty[0]
 
         if pad_token is not None:

comfy/supported_models.py

@@ -12,6 +12,7 @@ import comfy.text_encoders.hydit
 import comfy.text_encoders.flux
 import comfy.text_encoders.genmo
 import comfy.text_encoders.lt
+import comfy.text_encoders.hunyuan_video
 
 from . import supported_models_base
 from . import latent_formats
@@ -224,7 +225,6 @@ class SDXL(supported_models_base.BASE):
 
     def process_clip_state_dict_for_saving(self, state_dict):
         replace_prefix = {}
-        keys_to_replace = {}
         state_dict_g = diffusers_convert.convert_text_enc_state_dict_v20(state_dict, "clip_g")
         for k in state_dict:
             if k.startswith("clip_l"):
@@ -527,7 +527,6 @@ class SD3(supported_models_base.BASE):
         clip_l = False
         clip_g = False
         t5 = False
-        dtype_t5 = None
         pref = self.text_encoder_key_prefix[0]
         if "{}clip_l.transformer.text_model.final_layer_norm.weight".format(pref) in state_dict:
             clip_l = True
@@ -740,6 +739,54 @@ class LTXV(supported_models_base.BASE):
         t5_detect = comfy.text_encoders.sd3_clip.t5_xxl_detect(state_dict, "{}t5xxl.transformer.".format(pref))
         return supported_models_base.ClipTarget(comfy.text_encoders.lt.LTXVT5Tokenizer, comfy.text_encoders.lt.ltxv_te(**t5_detect))
 
-models = [Stable_Zero123, SD15_instructpix2pix, SD15, SD20, SD21UnclipL, SD21UnclipH, SDXL_instructpix2pix, SDXLRefiner, SDXL, SSD1B, KOALA_700M, KOALA_1B, Segmind_Vega, SD_X4Upscaler, Stable_Cascade_C, Stable_Cascade_B, SV3D_u, SV3D_p, SD3, StableAudio, AuraFlow, HunyuanDiT, HunyuanDiT1, FluxInpaint, Flux, FluxSchnell, GenmoMochi, LTXV]
+class HunyuanVideo(supported_models_base.BASE):
+    unet_config = {
+        "image_model": "hunyuan_video",
+    }
+
+    sampling_settings = {
+        "shift": 7.0,
+    }
+
+    unet_extra_config = {}
+    latent_format = latent_formats.HunyuanVideo
+
+    memory_usage_factor = 2.0 #TODO
+
+    supported_inference_dtypes = [torch.bfloat16, torch.float16, torch.float32]
+
+    vae_key_prefix = ["vae."]
+    text_encoder_key_prefix = ["text_encoders."]
+
+    def get_model(self, state_dict, prefix="", device=None):
+        out = model_base.HunyuanVideo(self, device=device)
+        return out
+
+    def process_unet_state_dict(self, state_dict):
+        out_sd = {}
+        for k in list(state_dict.keys()):
+            key_out = k
+            key_out = key_out.replace("txt_in.t_embedder.mlp.0.", "txt_in.t_embedder.in_layer.").replace("txt_in.t_embedder.mlp.2.", "txt_in.t_embedder.out_layer.")
+            key_out = key_out.replace("txt_in.c_embedder.linear_1.", "txt_in.c_embedder.in_layer.").replace("txt_in.c_embedder.linear_2.", "txt_in.c_embedder.out_layer.")
+            key_out = key_out.replace("_mod.linear.", "_mod.lin.").replace("_attn_qkv.", "_attn.qkv.")
+            key_out = key_out.replace("mlp.fc1.", "mlp.0.").replace("mlp.fc2.", "mlp.2.")
+            key_out = key_out.replace("_attn_q_norm.weight", "_attn.norm.query_norm.scale").replace("_attn_k_norm.weight", "_attn.norm.key_norm.scale")
+            key_out = key_out.replace(".q_norm.weight", ".norm.query_norm.scale").replace(".k_norm.weight", ".norm.key_norm.scale")
+            key_out = key_out.replace("_attn_proj.", "_attn.proj.")
+            key_out = key_out.replace(".modulation.linear.", ".modulation.lin.")
+            key_out = key_out.replace("_in.mlp.2.", "_in.out_layer.").replace("_in.mlp.0.", "_in.in_layer.")
+            out_sd[key_out] = state_dict[k]
+        return out_sd
+
+    def process_unet_state_dict_for_saving(self, state_dict):
+        replace_prefix = {"": "model.model."}
+        return utils.state_dict_prefix_replace(state_dict, replace_prefix)
+
+    def clip_target(self, state_dict={}):
+        # pref = self.text_encoder_key_prefix[0]
+        # t5_detect = comfy.text_encoders.sd3_clip.t5_xxl_detect(state_dict, "{}t5xxl.transformer.".format(pref))
+        return supported_models_base.ClipTarget(comfy.text_encoders.hunyuan_video.HunyuanVideoTokenizer, comfy.text_encoders.hunyuan_video.hunyuan_video_clip()) #TODO
+
+models = [Stable_Zero123, SD15_instructpix2pix, SD15, SD20, SD21UnclipL, SD21UnclipH, SDXL_instructpix2pix, SDXLRefiner, SDXL, SSD1B, KOALA_700M, KOALA_1B, Segmind_Vega, SD_X4Upscaler, Stable_Cascade_C, Stable_Cascade_B, SV3D_u, SV3D_p, SD3, StableAudio, AuraFlow, HunyuanDiT, HunyuanDiT1, FluxInpaint, Flux, FluxSchnell, GenmoMochi, LTXV, HunyuanVideo]
 
 models += [SVD_img2vid]

comfy/text_encoders/hunyuan_video.py

@@ -0,0 +1,98 @@
+from comfy import sd1_clip
+import comfy.model_management
+import comfy.text_encoders.llama
+from transformers import LlamaTokenizerFast
+import torch
+import os
+
+
+class LLAMA3Tokenizer(sd1_clip.SDTokenizer):
+    def __init__(self, embedding_directory=None, tokenizer_data={}, min_length=256):
+        tokenizer_path = os.path.join(os.path.dirname(os.path.realpath(__file__)), "llama_tokenizer")
+        super().__init__(tokenizer_path, embedding_directory=embedding_directory, pad_with_end=False, embedding_size=4096, embedding_key='llama', tokenizer_class=LlamaTokenizerFast, has_start_token=True, has_end_token=True, pad_to_max_length=False, max_length=99999999, pad_token=128258, end_token=128009, min_length=min_length)
+
+class LLAMAModel(sd1_clip.SDClipModel):
+    def __init__(self, device="cpu", layer="hidden", layer_idx=-3, dtype=None, attention_mask=True, model_options={}):
+        llama_scaled_fp8 = model_options.get("llama_scaled_fp8", None)
+        if llama_scaled_fp8 is not None:
+            model_options = model_options.copy()
+            model_options["scaled_fp8"] = llama_scaled_fp8
+
+        super().__init__(device=device, layer=layer, layer_idx=layer_idx, textmodel_json_config={}, dtype=dtype, special_tokens={"start": 128000, "pad": 128258}, layer_norm_hidden_state=False, model_class=comfy.text_encoders.llama.Llama2, enable_attention_masks=attention_mask, return_attention_masks=attention_mask, model_options=model_options)
+
+
+class HunyuanVideoTokenizer:
+    def __init__(self, embedding_directory=None, tokenizer_data={}):
+        clip_l_tokenizer_class = tokenizer_data.get("clip_l_tokenizer_class", sd1_clip.SDTokenizer)
+        self.clip_l = clip_l_tokenizer_class(embedding_directory=embedding_directory)
+        self.llama_template = """<|start_header_id|>system<|end_header_id|>
+
+Describe the video by detailing the following aspects: 1. The main content and theme of the video.2. The color, shape, size, texture, quantity, text, and spatial relationships of the objects.3. Actions, events, behaviors temporal relationships, physical movement changes of the objects.4. background environment, light, style and atmosphere.5. camera angles, movements, and transitions used in the video:<|eot_id|><|start_header_id|>user<|end_header_id|>"""  # 93 tokens
+        self.llama = LLAMA3Tokenizer(embedding_directory=embedding_directory, min_length=1)
+
+    def tokenize_with_weights(self, text:str, return_word_ids=False):
+        out = {}
+        out["l"] = self.clip_l.tokenize_with_weights(text, return_word_ids)
+
+        llama_text = "{}{}".format(self.llama_template, text)
+        out["llama"] = self.llama.tokenize_with_weights(llama_text, return_word_ids)
+        return out
+
+    def untokenize(self, token_weight_pair):
+        return self.clip_l.untokenize(token_weight_pair)
+
+    def state_dict(self):
+        return {}
+
+
+class HunyuanVideoClipModel(torch.nn.Module):
+    def __init__(self, dtype_llama=None, device="cpu", dtype=None, model_options={}):
+        super().__init__()
+        dtype_llama = comfy.model_management.pick_weight_dtype(dtype_llama, dtype, device)
+        clip_l_class = model_options.get("clip_l_class", sd1_clip.SDClipModel)
+        self.clip_l = clip_l_class(device=device, dtype=dtype, return_projected_pooled=False, model_options=model_options)
+        self.llama = LLAMAModel(device=device, dtype=dtype_llama, model_options=model_options)
+        self.dtypes = set([dtype, dtype_llama])
+
+    def set_clip_options(self, options):
+        self.clip_l.set_clip_options(options)
+        self.llama.set_clip_options(options)
+
+    def reset_clip_options(self):
+        self.clip_l.reset_clip_options()
+        self.llama.reset_clip_options()
+
+    def encode_token_weights(self, token_weight_pairs):
+        token_weight_pairs_l = token_weight_pairs["l"]
+        token_weight_pairs_llama = token_weight_pairs["llama"]
+
+        llama_out, llama_pooled, llama_extra_out = self.llama.encode_token_weights(token_weight_pairs_llama)
+
+        template_end = 0
+        for i, v in enumerate(token_weight_pairs_llama[0]):
+            if v[0] == 128007:  # <|end_header_id|>
+                template_end = i
+
+        llama_out = llama_out[:, template_end:]
+        llama_extra_out["attention_mask"] = llama_extra_out["attention_mask"][:, template_end:]
+        if llama_extra_out["attention_mask"].sum() == torch.numel(llama_extra_out["attention_mask"]):
+            llama_extra_out.pop("attention_mask")  # attention mask is useless if no masked elements
+
+        l_out, l_pooled = self.clip_l.encode_token_weights(token_weight_pairs_l)
+        return llama_out, l_pooled, llama_extra_out
+
+    def load_sd(self, sd):
+        if "text_model.encoder.layers.1.mlp.fc1.weight" in sd:
+            return self.clip_l.load_sd(sd)
+        else:
+            return self.llama.load_sd(sd)
+
+
+def hunyuan_video_clip(dtype_llama=None, llama_scaled_fp8=None):
+    class HunyuanVideoClipModel_(HunyuanVideoClipModel):
+        def __init__(self, device="cpu", dtype=None, model_options={}):
+            if llama_scaled_fp8 is not None and "llama_scaled_fp8" not in model_options:
+                model_options = model_options.copy()
+                model_options["llama_scaled_fp8"] = llama_scaled_fp8
+            super().__init__(dtype_llama=dtype_llama, device=device, dtype=dtype, model_options=model_options)
+    return HunyuanVideoClipModel_

comfy/text_encoders/llama.py

@@ -0,0 +1,221 @@
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+from dataclasses import dataclass
+from typing import Optional, Any
+
+from comfy.ldm.modules.attention import optimized_attention
+import comfy.model_management
+import comfy.ldm.common_dit
+
+import comfy.model_management
+
+@dataclass
+class Llama2Config:
+    vocab_size: int = 128320
+    hidden_size: int = 4096
+    intermediate_size: int = 14336
+    num_hidden_layers: int = 32
+    num_attention_heads: int = 32
+    num_key_value_heads: int = 8
+    max_position_embeddings: int = 8192
+    rms_norm_eps: float = 1e-5
+    rope_theta: float = 500000.0
+
+class RMSNorm(nn.Module):
+    def __init__(self, dim: int, eps: float = 1e-5, device=None, dtype=None):
+        super().__init__()
+        self.eps = eps
+        self.weight = nn.Parameter(torch.empty(dim, device=device, dtype=dtype))
+
+    def forward(self, x: torch.Tensor):
+        return comfy.ldm.common_dit.rms_norm(x, self.weight, self.eps)
+
+
+def rotate_half(x):
+    """Rotates half the hidden dims of the input."""
+    x1 = x[..., : x.shape[-1] // 2]
+    x2 = x[..., x.shape[-1] // 2 :]
+    return torch.cat((-x2, x1), dim=-1)
+
+
+def precompute_freqs_cis(head_dim, seq_len, theta, device=None):
+    theta_numerator = torch.arange(0, head_dim, 2, device=device).float()
+    inv_freq = 1.0 / (theta ** (theta_numerator / head_dim))
+
+    position_ids = torch.arange(0, seq_len, device=device).unsqueeze(0)
+
+    inv_freq_expanded = inv_freq[None, :, None].float().expand(position_ids.shape[0], -1, 1)
+    position_ids_expanded = position_ids[:, None, :].float()
+    freqs = (inv_freq_expanded.float() @ position_ids_expanded.float()).transpose(1, 2)
+    emb = torch.cat((freqs, freqs), dim=-1)
+    cos = emb.cos()
+    sin = emb.sin()
+    return (cos, sin)
+
+
+def apply_rope(xq, xk, freqs_cis):
+    cos = freqs_cis[0].unsqueeze(1)
+    sin = freqs_cis[1].unsqueeze(1)
+    q_embed = (xq * cos) + (rotate_half(xq) * sin)
+    k_embed = (xk * cos) + (rotate_half(xk) * sin)
+    return q_embed, k_embed
+
+
+class Attention(nn.Module):
+    def __init__(self, config: Llama2Config, device=None, dtype=None, ops: Any = None):
+        super().__init__()
+        self.num_heads = config.num_attention_heads
+        self.num_kv_heads = config.num_key_value_heads
+        self.hidden_size = config.hidden_size
+        self.head_dim = self.hidden_size // self.num_heads
+
+        ops = ops or nn
+        self.q_proj = ops.Linear(config.hidden_size, config.hidden_size, bias=False, device=device, dtype=dtype)
+        self.k_proj = ops.Linear(config.hidden_size, self.num_kv_heads * self.head_dim, bias=False, device=device, dtype=dtype)
+        self.v_proj = ops.Linear(config.hidden_size, self.num_kv_heads * self.head_dim, bias=False, device=device, dtype=dtype)
+        self.o_proj = ops.Linear(config.hidden_size, config.hidden_size, bias=False, device=device, dtype=dtype)
+
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        attention_mask: Optional[torch.Tensor] = None,
+        freqs_cis: Optional[torch.Tensor] = None,
+    ):
+        batch_size, seq_length, _ = hidden_states.shape
+
+        xq = self.q_proj(hidden_states)
+        xk = self.k_proj(hidden_states)
+        xv = self.v_proj(hidden_states)
+
+        xq = xq.view(batch_size, seq_length, self.num_heads, self.head_dim).transpose(1, 2)
+        xk = xk.view(batch_size, seq_length, self.num_kv_heads, self.head_dim).transpose(1, 2)
+        xv = xv.view(batch_size, seq_length, self.num_kv_heads, self.head_dim).transpose(1, 2)
+
+        xq, xk = apply_rope(xq, xk, freqs_cis=freqs_cis)
+
+        xk = xk.repeat_interleave(self.num_heads // self.num_kv_heads, dim=1)
+        xv = xv.repeat_interleave(self.num_heads // self.num_kv_heads, dim=1)
+
+        output = optimized_attention(xq, xk, xv, self.num_heads, mask=attention_mask, skip_reshape=True)
+        return self.o_proj(output)
+
+class MLP(nn.Module):
+    def __init__(self, config: Llama2Config, device=None, dtype=None, ops: Any = None):
+        super().__init__()
+        ops = ops or nn
+        self.gate_proj = ops.Linear(config.hidden_size, config.intermediate_size, bias=False, device=device, dtype=dtype)
+        self.up_proj = ops.Linear(config.hidden_size, config.intermediate_size, bias=False, device=device, dtype=dtype)
+        self.down_proj = ops.Linear(config.intermediate_size, config.hidden_size, bias=False, device=device, dtype=dtype)
+
+    def forward(self, x):
+        return self.down_proj(F.silu(self.gate_proj(x)) * self.up_proj(x))
+
+class TransformerBlock(nn.Module):
+    def __init__(self, config: Llama2Config, device=None, dtype=None, ops: Any = None):
+        super().__init__()
+        self.self_attn = Attention(config, device=device, dtype=dtype, ops=ops)
+        self.mlp = MLP(config, device=device, dtype=dtype, ops=ops)
+        self.input_layernorm = RMSNorm(config.hidden_size, eps=config.rms_norm_eps, device=device, dtype=dtype)
+        self.post_attention_layernorm = RMSNorm(config.hidden_size, eps=config.rms_norm_eps, device=device, dtype=dtype)
+
+    def forward(
+        self,
+        x: torch.Tensor,
+        attention_mask: Optional[torch.Tensor] = None,
+        freqs_cis: Optional[torch.Tensor] = None,
+    ):
+        # Self Attention
+        residual = x
+        x = self.input_layernorm(x)
+        x = self.self_attn(
+            hidden_states=x,
+            attention_mask=attention_mask,
+            freqs_cis=freqs_cis,
+        )
+        x = residual + x
+
+        # MLP
+        residual = x
+        x = self.post_attention_layernorm(x)
+        x = self.mlp(x)
+        x = residual + x
+
+        return x
+
+class Llama2_(nn.Module):
+    def __init__(self, config, device=None, dtype=None, ops=None):
+        super().__init__()
+        self.config = config
+        self.vocab_size = config.vocab_size
+
+        self.embed_tokens = ops.Embedding(
+            config.vocab_size,
+            config.hidden_size,
+            device=device,
+            dtype=dtype
+        )
+        self.layers = nn.ModuleList([
+            TransformerBlock(config, device=device, dtype=dtype, ops=ops)
+            for _ in range(config.num_hidden_layers)
+        ])
+        self.norm = RMSNorm(config.hidden_size, eps=config.rms_norm_eps, device=device, dtype=dtype)
+        # self.lm_head = ops.Linear(config.hidden_size, config.vocab_size, bias=False, device=device, dtype=dtype)
+
+    def forward(self, x, attention_mask=None, intermediate_output=None, final_layer_norm_intermediate=True, dtype=None):
+        x = self.embed_tokens(x, out_dtype=dtype)
+
+        freqs_cis = precompute_freqs_cis(self.config.hidden_size // self.config.num_attention_heads,
+                                         x.shape[1],
+                                         self.config.rope_theta,
+                                         device=x.device)
+
+        mask = None
+        if attention_mask is not None:
+            mask = 1.0 - attention_mask.to(x.dtype).reshape((attention_mask.shape[0], 1, -1, attention_mask.shape[-1])).expand(attention_mask.shape[0], 1, attention_mask.shape[-1], attention_mask.shape[-1])
+            mask = mask.masked_fill(mask.to(torch.bool), float("-inf"))
+
+        causal_mask = torch.empty(x.shape[1], x.shape[1], dtype=x.dtype, device=x.device).fill_(float("-inf")).triu_(1)
+        if mask is not None:
+            mask += causal_mask
+        else:
+            mask = causal_mask
+
+        intermediate = None
+        if intermediate_output is not None:
+            if intermediate_output < 0:
+                intermediate_output = len(self.layers) + intermediate_output
+
+        for i, layer in enumerate(self.layers):
+            x = layer(
+                x=x,
+                attention_mask=mask,
+                freqs_cis=freqs_cis,
+            )
+            if i == intermediate_output:
+                intermediate = x.clone()
+
+        x = self.norm(x)
+        if intermediate is not None and final_layer_norm_intermediate:
+            intermediate = self.norm(intermediate)
+
+        return x, intermediate
+
+
+class Llama2(torch.nn.Module):
+    def __init__(self, config_dict, dtype, device, operations):
+        super().__init__()
+        config = Llama2Config(**config_dict)
+        self.num_layers = config.num_hidden_layers
+
+        self.model = Llama2_(config, device=device, dtype=dtype, ops=operations)
+        self.dtype = dtype
+
+    def get_input_embeddings(self):
+        return self.model.embed_tokens
+
+    def set_input_embeddings(self, embeddings):
+        self.model.embed_tokens = embeddings
+
+    def forward(self, input_ids, *args, **kwargs):
+        return self.model(input_ids, *args, **kwargs)

comfy/text_encoders/t5.py

@@ -172,7 +172,6 @@ class T5LayerSelfAttention(torch.nn.Module):
         # self.dropout = nn.Dropout(config.dropout_rate)
 
     def forward(self, x, mask=None, past_bias=None, optimized_attention=None):
-        normed_hidden_states = self.layer_norm(x)
         output, past_bias = self.SelfAttention(self.layer_norm(x), mask=mask, past_bias=past_bias, optimized_attention=optimized_attention)
         # x = x + self.dropout(attention_output)
         x += output

comfy/utils.py

@@ -26,6 +26,8 @@ import numpy as np
 from PIL import Image
 import logging
 import itertools
+from torch.nn.functional import interpolate
+from einops import rearrange
 
 def load_torch_file(ckpt, safe_load=False, device=None):
     if device is None:
@@ -873,5 +875,46 @@ def reshape_mask(input_mask, output_shape):
     mask = torch.nn.functional.interpolate(input_mask, size=output_shape[2:], mode=scale_mode)
     if mask.shape[1] < output_shape[1]:
         mask = mask.repeat((1, output_shape[1]) + (1,) * dims)[:,:output_shape[1]]
-    mask = comfy.utils.repeat_to_batch_size(mask, output_shape[0])
+    mask = repeat_to_batch_size(mask, output_shape[0])
     return mask
+
+def upscale_dit_mask(mask: torch.Tensor, img_size_in, img_size_out):
+        hi, wi = img_size_in
+        ho, wo = img_size_out
+        # if it's already the correct size, no need to do anything
+        if (hi, wi) == (ho, wo):
+            return mask
+        if mask.ndim == 2:
+            mask = mask.unsqueeze(0)
+        if mask.ndim != 3:
+            raise ValueError(f"Got a mask of shape {list(mask.shape)}, expected [b, q, k] or [q, k]")
+        txt_tokens = mask.shape[1] - (hi * wi)
+        # quadrants of the mask
+        txt_to_txt = mask[:, :txt_tokens, :txt_tokens]
+        txt_to_img = mask[:, :txt_tokens, txt_tokens:]
+        img_to_img = mask[:, txt_tokens:, txt_tokens:]
+        img_to_txt = mask[:, txt_tokens:, :txt_tokens]
+
+        # convert to 1d x 2d, interpolate, then back to 1d x 1d
+        txt_to_img = rearrange  (txt_to_img, "b t (h w) -> b t h w", h=hi, w=wi)
+        txt_to_img = interpolate(txt_to_img, size=img_size_out, mode="bilinear")
+        txt_to_img = rearrange  (txt_to_img, "b t h w -> b t (h w)")
+        # this one is hard because we have to do it twice
+        # convert to 1d x 2d, interpolate, then to 2d x 1d, interpolate, then 1d x 1d
+        img_to_img = rearrange  (img_to_img, "b hw (h w) -> b hw h w", h=hi, w=wi)
+        img_to_img = interpolate(img_to_img, size=img_size_out, mode="bilinear")
+        img_to_img = rearrange  (img_to_img, "b (hk wk) hq wq -> b (hq wq) hk wk", hk=hi, wk=wi)
+        img_to_img = interpolate(img_to_img, size=img_size_out, mode="bilinear")
+        img_to_img = rearrange  (img_to_img, "b (hq wq) hk wk -> b (hk wk) (hq wq)", hq=ho, wq=wo)
+        # convert to 2d x 1d, interpolate, then back to 1d x 1d
+        img_to_txt = rearrange  (img_to_txt, "b (h w) t -> b t h w", h=hi, w=wi)
+        img_to_txt = interpolate(img_to_txt, size=img_size_out, mode="bilinear")
+        img_to_txt = rearrange  (img_to_txt, "b t h w -> b (h w) t")
+
+        # reassemble the mask from blocks
+        out = torch.cat([
+            torch.cat([txt_to_txt, txt_to_img], dim=2),
+            torch.cat([img_to_txt, img_to_img], dim=2)],
+            dim=1
+        )
+        return out

comfy_extras/nodes_clip_sdxl.py

@@ -22,14 +22,15 @@ class CLIPTextEncodeSDXL:
     @classmethod
     def INPUT_TYPES(s):
         return {"required": {
+            "clip": ("CLIP", ),
             "width": ("INT", {"default": 1024.0, "min": 0, "max": MAX_RESOLUTION}),
             "height": ("INT", {"default": 1024.0, "min": 0, "max": MAX_RESOLUTION}),
             "crop_w": ("INT", {"default": 0, "min": 0, "max": MAX_RESOLUTION}),
             "crop_h": ("INT", {"default": 0, "min": 0, "max": MAX_RESOLUTION}),
             "target_width": ("INT", {"default": 1024.0, "min": 0, "max": MAX_RESOLUTION}),
             "target_height": ("INT", {"default": 1024.0, "min": 0, "max": MAX_RESOLUTION}),
-            "text_g": ("STRING", {"multiline": True, "dynamicPrompts": True}), "clip": ("CLIP", ),
-            "text_l": ("STRING", {"multiline": True, "dynamicPrompts": True}), "clip": ("CLIP", ),
+            "text_g": ("STRING", {"multiline": True, "dynamicPrompts": True}),
+            "text_l": ("STRING", {"multiline": True, "dynamicPrompts": True}),
             }}
     RETURN_TYPES = ("CONDITIONING",)
     FUNCTION = "encode"

comfy_extras/nodes_hunyuan.py

@@ -1,3 +1,8 @@
+import nodes
+import torch
+import comfy.model_management
+
+
 class CLIPTextEncodeHunyuanDiT:
     @classmethod
     def INPUT_TYPES(s):
@@ -17,7 +22,23 @@ class CLIPTextEncodeHunyuanDiT:
 
         return (clip.encode_from_tokens_scheduled(tokens), )
 
+class EmptyHunyuanLatentVideo:
+    @classmethod
+    def INPUT_TYPES(s):
+        return {"required": { "width": ("INT", {"default": 848, "min": 16, "max": nodes.MAX_RESOLUTION, "step": 16}),
+                              "height": ("INT", {"default": 480, "min": 16, "max": nodes.MAX_RESOLUTION, "step": 16}),
+                              "length": ("INT", {"default": 25, "min": 1, "max": nodes.MAX_RESOLUTION, "step": 4}),
+                              "batch_size": ("INT", {"default": 1, "min": 1, "max": 4096})}}
+    RETURN_TYPES = ("LATENT",)
+    FUNCTION = "generate"
+
+    CATEGORY = "latent/video"
+
+    def generate(self, width, height, length, batch_size=1):
+        latent = torch.zeros([batch_size, 16, ((length - 1) // 4) + 1, height // 8, width // 8], device=comfy.model_management.intermediate_device())
+        return ({"samples":latent}, )
 
 NODE_CLASS_MAPPINGS = {
     "CLIPTextEncodeHunyuanDiT": CLIPTextEncodeHunyuanDiT,
+    "EmptyHunyuanLatentVideo": EmptyHunyuanLatentVideo,
 }

comfy_extras/nodes_hypertile.py

@@ -35,8 +35,6 @@ class HyperTile:
     CATEGORY = "model_patches/unet"
 
     def patch(self, model, tile_size, swap_size, max_depth, scale_depth):
-        model_channels = model.model.model_config.unet_config["model_channels"]
-
         latent_tile_size = max(32, tile_size) // 8
         self.temp = None
 

comfy_extras/nodes_load_3d.py

@@ -0,0 +1,95 @@
+import nodes
+import folder_paths
+import os
+
+def normalize_path(path):
+    return path.replace('\\', '/')
+
+class Load3D():
+    @classmethod
+    def INPUT_TYPES(s):
+        input_dir = os.path.join(folder_paths.get_input_directory(), "3d")
+
+        os.makedirs(input_dir, exist_ok=True)
+
+        files = [normalize_path(os.path.join("3d", f)) for f in os.listdir(input_dir) if f.endswith(('.gltf', '.glb', '.obj', '.mtl', '.fbx', '.stl'))]
+
+        return {"required": {
+            "model_file": (sorted(files), {"file_upload": True}),
+            "image": ("LOAD_3D", {}),
+            "width": ("INT", {"default": 1024, "min": 1, "max": 4096, "step": 1}),
+            "height": ("INT", {"default": 1024, "min": 1, "max": 4096, "step": 1}),
+            "show_grid": ([True, False],),
+            "camera_type": (["perspective", "orthographic"],),
+            "view": (["front", "right", "top", "isometric"],),
+            "material": (["original", "normal", "wireframe", "depth"],),
+            "bg_color": ("STRING", {"default": "#000000", "multiline": False}),
+            "light_intensity": ("INT", {"default": 10, "min": 1, "max": 20, "step": 1}),
+            "up_direction": (["original", "-x", "+x", "-y", "+y", "-z", "+z"],),
+        }}
+
+    RETURN_TYPES = ("IMAGE", "MASK", "STRING")
+    RETURN_NAMES = ("image", "mask", "mesh_path")
+
+    FUNCTION = "process"
+
+    CATEGORY = "3d"
+
+    def process(self, model_file, image, **kwargs):
+        imagepath = folder_paths.get_annotated_filepath(image)
+
+        load_image_node = nodes.LoadImage()
+
+        output_image, output_mask = load_image_node.load_image(image=imagepath)
+
+        return output_image, output_mask, model_file,
+
+class Load3DAnimation():
+    @classmethod
+    def INPUT_TYPES(s):
+        input_dir = os.path.join(folder_paths.get_input_directory(), "3d")
+
+        os.makedirs(input_dir, exist_ok=True)
+
+        files = [normalize_path(os.path.join("3d", f)) for f in os.listdir(input_dir) if f.endswith(('.gltf', '.glb', '.fbx'))]
+
+        return {"required": {
+            "model_file": (sorted(files), {"file_upload": True}),
+            "image": ("LOAD_3D_ANIMATION", {}),
+            "width": ("INT", {"default": 1024, "min": 1, "max": 4096, "step": 1}),
+            "height": ("INT", {"default": 1024, "min": 1, "max": 4096, "step": 1}),
+            "show_grid": ([True, False],),
+            "camera_type": (["perspective", "orthographic"],),
+            "view": (["front", "right", "top", "isometric"],),
+            "material": (["original", "normal", "wireframe", "depth"],),
+            "bg_color": ("STRING", {"default": "#000000", "multiline": False}),
+            "light_intensity": ("INT", {"default": 10, "min": 1, "max": 20, "step": 1}),
+            "up_direction": (["original", "-x", "+x", "-y", "+y", "-z", "+z"],),
+            "animation_speed": (["0.1", "0.5", "1", "1.5", "2"], {"default": "1"}),
+        }}
+
+    RETURN_TYPES = ("IMAGE", "MASK", "STRING")
+    RETURN_NAMES = ("image", "mask", "mesh_path")
+
+    FUNCTION = "process"
+
+    CATEGORY = "3d"
+
+    def process(self, model_file, image, **kwargs):
+        imagepath = folder_paths.get_annotated_filepath(image)
+
+        load_image_node = nodes.LoadImage()
+
+        output_image, output_mask = load_image_node.load_image(image=imagepath)
+
+        return output_image, output_mask, model_file,
+
+NODE_CLASS_MAPPINGS = {
+    "Load3D": Load3D,
+    "Load3DAnimation": Load3DAnimation
+}
+
+NODE_DISPLAY_NAME_MAPPINGS = {
+    "Load3D": "Load 3D",
+    "Load3DAnimation": "Load 3D - Animation"
+}

comfy_extras/nodes_model_advanced.py

@@ -240,7 +240,6 @@ class ModelSamplingContinuousV:
     def patch(self, model, sampling, sigma_max, sigma_min):
         m = model.clone()
 
-        latent_format = None
         sigma_data = 1.0
         if sampling == "v_prediction":
             sampling_type = comfy.model_sampling.V_PREDICTION

comfy_extras/nodes_photomaker.py

@@ -16,6 +16,7 @@ VISION_CONFIG_DICT = {
     "patch_size": 14,
     "projection_dim": 768,
     "hidden_act": "quick_gelu",
+    "model_type": "clip_vision_model",
 }
 
 class MLP(nn.Module):

execution.py

@@ -144,11 +144,16 @@ def _map_node_over_list(obj, input_data_all, func, allow_interrupt=False, execut
         return {k: v[i if len(v) > i else -1] for k, v in d.items()}
     
     results = []
-    def process_inputs(inputs, index=None):
+    def process_inputs(inputs, index=None, input_is_list=False):
         if allow_interrupt:
             nodes.before_node_execution()
         execution_block = None
         for k, v in inputs.items():
+            if input_is_list:
+                for e in v:
+                    if isinstance(e, ExecutionBlocker):
+                        v = e
+                        break
             if isinstance(v, ExecutionBlocker):
                 execution_block = execution_block_cb(v) if execution_block_cb else v
                 break
@@ -160,7 +165,7 @@ def _map_node_over_list(obj, input_data_all, func, allow_interrupt=False, execut
             results.append(execution_block)
 
     if input_is_list:
-        process_inputs(input_data_all, 0)
+        process_inputs(input_data_all, 0, input_is_list=input_is_list)
     elif max_len_input == 0:
         process_inputs({})
     else: 
@@ -760,7 +765,7 @@ def validate_prompt(prompt):
         if 'class_type' not in prompt[x]:
             error = {
                 "type": "invalid_prompt",
-                "message": f"Cannot execute because a node is missing the class_type property.",
+                "message": "Cannot execute because a node is missing the class_type property.",
                 "details": f"Node ID '#{x}'",
                 "extra_info": {}
             }

fix_torch.py

@@ -22,7 +22,7 @@ def fix_pytorch_libomp():
             if b"libomp140.x86_64.dll" not in contents:
                 break
         try:
-            mydll = ctypes.cdll.LoadLibrary(test_file)
-        except FileNotFoundError as e:
+            ctypes.cdll.LoadLibrary(test_file)
+        except FileNotFoundError:
             logging.warning("Detected pytorch version with libomp issue, patching.")
             shutil.copyfile(os.path.join(lib_folder, "libiomp5md.dll"), dest)

folder_paths.py

@@ -200,10 +200,17 @@ def add_model_folder_path(folder_name: str, full_folder_path: str, is_default: b
     global folder_names_and_paths
     folder_name = map_legacy(folder_name)
     if folder_name in folder_names_and_paths:
-        if is_default:
-            folder_names_and_paths[folder_name][0].insert(0, full_folder_path)
+        paths, _exts = folder_names_and_paths[folder_name]
+        if full_folder_path in paths:
+            if is_default and paths[0] != full_folder_path:
+                # If the path to the folder is not the first in the list, move it to the beginning.
+                paths.remove(full_folder_path)
+                paths.insert(0, full_folder_path)
         else:
-            folder_names_and_paths[folder_name][0].append(full_folder_path)
+            if is_default:
+                paths.insert(0, full_folder_path)
+            else:
+                paths.append(full_folder_path)
     else:
         folder_names_and_paths[folder_name] = ([full_folder_path], set())
 

main.py

@@ -7,6 +7,9 @@ import folder_paths
 import time
 from comfy.cli_args import args
 from app.logger import setup_logger
+import itertools
+import utils.extra_config
+import logging
 
 if __name__ == "__main__":
     #NOTE: These do not do anything on core ComfyUI which should already have no communication with the internet, they are for custom nodes.
@@ -16,6 +19,40 @@ if __name__ == "__main__":
 
 setup_logger(log_level=args.verbose)
 
+def apply_custom_paths():
+    # extra model paths
+    extra_model_paths_config_path = os.path.join(os.path.dirname(os.path.realpath(__file__)), "extra_model_paths.yaml")
+    if os.path.isfile(extra_model_paths_config_path):
+        utils.extra_config.load_extra_path_config(extra_model_paths_config_path)
+
+    if args.extra_model_paths_config:
+        for config_path in itertools.chain(*args.extra_model_paths_config):
+            utils.extra_config.load_extra_path_config(config_path)
+
+    # --output-directory, --input-directory, --user-directory
+    if args.output_directory:
+        output_dir = os.path.abspath(args.output_directory)
+        logging.info(f"Setting output directory to: {output_dir}")
+        folder_paths.set_output_directory(output_dir)
+
+    # These are the default folders that checkpoints, clip and vae models will be saved to when using CheckpointSave, etc.. nodes
+    folder_paths.add_model_folder_path("checkpoints", os.path.join(folder_paths.get_output_directory(), "checkpoints"))
+    folder_paths.add_model_folder_path("clip", os.path.join(folder_paths.get_output_directory(), "clip"))
+    folder_paths.add_model_folder_path("vae", os.path.join(folder_paths.get_output_directory(), "vae"))
+    folder_paths.add_model_folder_path("diffusion_models",
+                                       os.path.join(folder_paths.get_output_directory(), "diffusion_models"))
+    folder_paths.add_model_folder_path("loras", os.path.join(folder_paths.get_output_directory(), "loras"))
+
+    if args.input_directory:
+        input_dir = os.path.abspath(args.input_directory)
+        logging.info(f"Setting input directory to: {input_dir}")
+        folder_paths.set_input_directory(input_dir)
+
+    if args.user_directory:
+        user_dir = os.path.abspath(args.user_directory)
+        logging.info(f"Setting user directory to: {user_dir}")
+        folder_paths.set_user_directory(user_dir)
+
 
 def execute_prestartup_script():
     def execute_script(script_path):
@@ -57,18 +94,16 @@ def execute_prestartup_script():
             print("{:6.1f} seconds{}:".format(n[0], import_message), n[1])
         print()
 
+apply_custom_paths()
 execute_prestartup_script()
 
 
 # Main code
 import asyncio
-import itertools
 import shutil
 import threading
 import gc
 
-import logging
-import utils.extra_config
 
 if os.name == "nt":
     logging.getLogger("xformers").addFilter(lambda record: 'A matching Triton is not available' not in record.getMessage())
@@ -112,6 +147,7 @@ def cuda_malloc_warning():
             logging.warning("\nWARNING: this card most likely does not support cuda-malloc, if you get \"CUDA error\" please run ComfyUI with: --disable-cuda-malloc\n")
 
 def prompt_worker(q, server):
+    current_time: float = 0.0
     e = execution.PromptExecutor(server, lru_size=args.cache_lru)
     last_gc_collect = 0
     need_gc = False
@@ -208,14 +244,6 @@ if __name__ == "__main__":
     server = server.PromptServer(loop)
     q = execution.PromptQueue(server)
 
-    extra_model_paths_config_path = os.path.join(os.path.dirname(os.path.realpath(__file__)), "extra_model_paths.yaml")
-    if os.path.isfile(extra_model_paths_config_path):
-        utils.extra_config.load_extra_path_config(extra_model_paths_config_path)
-
-    if args.extra_model_paths_config:
-        for config_path in itertools.chain(*args.extra_model_paths_config):
-            utils.extra_config.load_extra_path_config(config_path)
-
     nodes.init_extra_nodes(init_custom_nodes=not args.disable_all_custom_nodes)
 
     cuda_malloc_warning()
@@ -225,28 +253,6 @@ if __name__ == "__main__":
 
     threading.Thread(target=prompt_worker, daemon=True, args=(q, server,)).start()
 
-    if args.output_directory:
-        output_dir = os.path.abspath(args.output_directory)
-        logging.info(f"Setting output directory to: {output_dir}")
-        folder_paths.set_output_directory(output_dir)
-
-    #These are the default folders that checkpoints, clip and vae models will be saved to when using CheckpointSave, etc.. nodes
-    folder_paths.add_model_folder_path("checkpoints", os.path.join(folder_paths.get_output_directory(), "checkpoints"))
-    folder_paths.add_model_folder_path("clip", os.path.join(folder_paths.get_output_directory(), "clip"))
-    folder_paths.add_model_folder_path("vae", os.path.join(folder_paths.get_output_directory(), "vae"))
-    folder_paths.add_model_folder_path("diffusion_models", os.path.join(folder_paths.get_output_directory(), "diffusion_models"))
-    folder_paths.add_model_folder_path("loras", os.path.join(folder_paths.get_output_directory(), "loras"))
-
-    if args.input_directory:
-        input_dir = os.path.abspath(args.input_directory)
-        logging.info(f"Setting input directory to: {input_dir}")
-        folder_paths.set_input_directory(input_dir)
-    
-    if args.user_directory:
-        user_dir = os.path.abspath(args.user_directory)
-        logging.info(f"Setting user directory to: {user_dir}")
-        folder_paths.set_user_directory(user_dir)
-
     if args.quick_test_for_ci:
         exit(0)
 

nodes.py

@@ -929,7 +929,7 @@ class DualCLIPLoader:
     def INPUT_TYPES(s):
         return {"required": { "clip_name1": (folder_paths.get_filename_list("text_encoders"), ),
                               "clip_name2": (folder_paths.get_filename_list("text_encoders"), ),
-                              "type": (["sdxl", "sd3", "flux"], ),
+                              "type": (["sdxl", "sd3", "flux", "hunyuan_video"], ),
                              }}
     RETURN_TYPES = ("CLIP",)
     FUNCTION = "load_clip"
@@ -947,6 +947,8 @@ class DualCLIPLoader:
             clip_type = comfy.sd.CLIPType.SD3
         elif type == "flux":
             clip_type = comfy.sd.CLIPType.FLUX
+        elif type == "hunyuan_video":
+            clip_type = comfy.sd.CLIPType.HUNYUAN_VIDEO
 
         clip = comfy.sd.load_clip(ckpt_paths=[clip_path1, clip_path2], embedding_directory=folder_paths.get_folder_paths("embeddings"), clip_type=clip_type)
         return (clip,)
@@ -1008,23 +1010,58 @@ class StyleModelApply:
                              "style_model": ("STYLE_MODEL", ),
                              "clip_vision_output": ("CLIP_VISION_OUTPUT", ),
                              "strength": ("FLOAT", {"default": 1.0, "min": 0.0, "max": 10.0, "step": 0.001}),
-                             "strength_type": (["multiply"], ),
+                             "strength_type": (["multiply", "attn_bias"], ),
                              }}
     RETURN_TYPES = ("CONDITIONING",)
     FUNCTION = "apply_stylemodel"
 
     CATEGORY = "conditioning/style_model"
 
-    def apply_stylemodel(self, clip_vision_output, style_model, conditioning, strength, strength_type):
+    def apply_stylemodel(self, conditioning, style_model, clip_vision_output, strength, strength_type):
         cond = style_model.get_cond(clip_vision_output).flatten(start_dim=0, end_dim=1).unsqueeze(dim=0)
         if strength_type == "multiply":
             cond *= strength
 
-        c = []
+        n = cond.shape[1]
+        c_out = []
         for t in conditioning:
-            n = [torch.cat((t[0], cond), dim=1), t[1].copy()]
-            c.append(n)
-        return (c, )
+            (txt, keys) = t
+            keys = keys.copy()
+            if strength_type == "attn_bias" and strength != 1.0:
+                # math.log raises an error if the argument is zero
+                # torch.log returns -inf, which is what we want
+                attn_bias = torch.log(torch.Tensor([strength]))
+                # get the size of the mask image
+                mask_ref_size = keys.get("attention_mask_img_shape", (1, 1))
+                n_ref = mask_ref_size[0] * mask_ref_size[1]
+                n_txt = txt.shape[1]
+                # grab the existing mask
+                mask = keys.get("attention_mask", None)
+                # create a default mask if it doesn't exist
+                if mask is None:
+                    mask = torch.zeros((txt.shape[0], n_txt + n_ref, n_txt + n_ref), dtype=torch.float16)
+                # convert the mask dtype, because it might be boolean
+                # we want it to be interpreted as a bias
+                if mask.dtype == torch.bool:
+                    # log(True) = log(1) = 0
+                    # log(False) = log(0) = -inf
+                    mask = torch.log(mask.to(dtype=torch.float16))
+                # now we make the mask bigger to add space for our new tokens
+                new_mask = torch.zeros((txt.shape[0], n_txt + n + n_ref, n_txt + n + n_ref), dtype=torch.float16)
+                # copy over the old mask, in quandrants
+                new_mask[:, :n_txt, :n_txt] = mask[:, :n_txt, :n_txt]
+                new_mask[:, :n_txt, n_txt+n:] = mask[:, :n_txt, n_txt:]
+                new_mask[:, n_txt+n:, :n_txt] = mask[:, n_txt:, :n_txt]
+                new_mask[:, n_txt+n:, n_txt+n:] = mask[:, n_txt:, n_txt:]
+                # now fill in the attention bias to our redux tokens
+                new_mask[:, :n_txt, n_txt:n_txt+n] = attn_bias
+                new_mask[:, n_txt+n:, n_txt:n_txt+n] = attn_bias
+                keys["attention_mask"] = new_mask.to(txt.device)
+                keys["attention_mask_img_shape"] = mask_ref_size
+
+            c_out.append([torch.cat((txt, cond), dim=1), keys])
+
+        return (c_out,)
 
 class unCLIPConditioning:
     @classmethod
@@ -2150,6 +2187,7 @@ def init_builtin_extra_nodes():
         "nodes_mahiro.py",
         "nodes_lt.py",
         "nodes_hooks.py",
+        "nodes_load_3d.py",
     ]
 
     import_failed = []

notebooks/comfyui_colab.ipynb

@@ -237,11 +237,7 @@
    "source": [
     "!npm install -g localtunnel\n",
     "\n",
-    "import subprocess\n",
     "import threading\n",
-    "import time\n",
-    "import socket\n",
-    "import urllib.request\n",
     "\n",
     "def iframe_thread(port):\n",
     "  while True:\n",
@@ -288,8 +284,6 @@
    "outputs": [],
    "source": [
     "import threading\n",
-    "import time\n",
-    "import socket\n",
     "def iframe_thread(port):\n",
     "  while True:\n",
     "      time.sleep(0.5)\n",

ruff.toml

@@ -4,5 +4,7 @@ lint.ignore = ["ALL"]
 # Enable specific rules
 lint.select = [
     "S307",  # suspicious-eval-usage
-    "F401",  # unused-import
+    # The "F" series in Ruff stands for "Pyflakes" rules, which catch various Python syntax errors and undefined names.
+    # See all rules here: https://docs.astral.sh/ruff/rules/#pyflakes-f
+    "F",
 ]

server.py

@@ -460,7 +460,21 @@ class PromptServer():
                             return web.Response(body=alpha_buffer.read(), content_type='image/png',
                                                 headers={"Content-Disposition": f"filename=\"{filename}\""})
                     else:
-                        return web.FileResponse(file, headers={"Content-Disposition": f"filename=\"{filename}\""})
+                        # Get content type from mimetype, defaulting to 'application/octet-stream'
+                        content_type = mimetypes.guess_type(filename)[0] or 'application/octet-stream'
+
+                        # For security, force certain extensions to download instead of display
+                        file_extension = os.path.splitext(filename)[1].lower()
+                        if file_extension in {'.html', '.htm', '.js', '.css'}:
+                            content_type = 'application/octet-stream'  # Forces download
+
+                        return web.FileResponse(
+                            file,
+                            headers={
+                                "Content-Disposition": f"filename=\"{filename}\"",
+                                "Content-Type": content_type
+                            }
+                        )
 
             return web.Response(status=404)
 
@@ -563,7 +577,7 @@ class PromptServer():
                 for x in nodes.NODE_CLASS_MAPPINGS:
                     try:
                         out[x] = node_info(x)
-                    except Exception as e:
+                    except Exception:
                         logging.error(f"[ERROR] An error occurred while retrieving information for the '{x}' node.")
                         logging.error(traceback.format_exc())
                 return web.json_response(out)
@@ -584,7 +598,7 @@ class PromptServer():
             return web.json_response(self.prompt_queue.get_history(max_items=max_items))
 
         @routes.get("/history/{prompt_id}")
-        async def get_history(request):
+        async def get_history_prompt_id(request):
             prompt_id = request.match_info.get("prompt_id", None)
             return web.json_response(self.prompt_queue.get_history(prompt_id=prompt_id))
 
@@ -599,8 +613,6 @@ class PromptServer():
         @routes.post("/prompt")
         async def post_prompt(request):
             logging.info("got prompt")
-            resp_code = 200
-            out_string = ""
             json_data =  await request.json()
             json_data = self.trigger_on_prompt(json_data)
 
@@ -832,8 +844,8 @@ class PromptServer():
         for handler in self.on_prompt_handlers:
             try:
                 json_data = handler(json_data)
-            except Exception as e:
-                logging.warning(f"[ERROR] An error occurred during the on_prompt_handler processing")
+            except Exception:
+                logging.warning("[ERROR] An error occurred during the on_prompt_handler processing")
                 logging.warning(traceback.format_exc())
 
         return json_data

tests-unit/comfy_test/folder_path_test.py

@@ -7,6 +7,14 @@ from unittest.mock import patch
 
 import folder_paths
 
+@pytest.fixture()
+def clear_folder_paths():
+    # Clear the global dictionary before each test to ensure isolation
+    original = folder_paths.folder_names_and_paths.copy()
+    folder_paths.folder_names_and_paths.clear()
+    yield
+    folder_paths.folder_names_and_paths = original
+
 @pytest.fixture
 def temp_dir():
     with tempfile.TemporaryDirectory() as tmpdirname:
@@ -30,9 +38,33 @@ def test_get_annotated_filepath():
     assert folder_paths.get_annotated_filepath("test.txt", default_dir) == os.path.join(default_dir, "test.txt")
     assert folder_paths.get_annotated_filepath("test.txt [output]") == os.path.join(folder_paths.get_output_directory(), "test.txt")
 
-def test_add_model_folder_path():
-    folder_paths.add_model_folder_path("test_folder", "/test/path")
-    assert "/test/path" in folder_paths.get_folder_paths("test_folder")
+def test_add_model_folder_path_append(clear_folder_paths):
+    folder_paths.add_model_folder_path("test_folder", "/default/path", is_default=True)
+    folder_paths.add_model_folder_path("test_folder", "/test/path", is_default=False)
+    assert folder_paths.get_folder_paths("test_folder") == ["/default/path", "/test/path"]
+
+
+def test_add_model_folder_path_insert(clear_folder_paths):
+    folder_paths.add_model_folder_path("test_folder", "/test/path", is_default=False)
+    folder_paths.add_model_folder_path("test_folder", "/default/path", is_default=True)
+    assert folder_paths.get_folder_paths("test_folder") == ["/default/path", "/test/path"]
+
+
+def test_add_model_folder_path_re_add_existing_default(clear_folder_paths):
+    folder_paths.add_model_folder_path("test_folder", "/test/path", is_default=False)
+    folder_paths.add_model_folder_path("test_folder", "/old_default/path", is_default=True)
+    assert folder_paths.get_folder_paths("test_folder") == ["/old_default/path", "/test/path"]
+    folder_paths.add_model_folder_path("test_folder", "/test/path", is_default=True)
+    assert folder_paths.get_folder_paths("test_folder") == ["/test/path", "/old_default/path"]
+
+
+def test_add_model_folder_path_re_add_existing_non_default(clear_folder_paths):
+    folder_paths.add_model_folder_path("test_folder", "/test/path", is_default=False)
+    folder_paths.add_model_folder_path("test_folder", "/default/path", is_default=True)
+    assert folder_paths.get_folder_paths("test_folder") == ["/default/path", "/test/path"]
+    folder_paths.add_model_folder_path("test_folder", "/test/path", is_default=False)
+    assert folder_paths.get_folder_paths("test_folder") == ["/default/path", "/test/path"]
+
 
 def test_recursive_search(temp_dir):
     os.makedirs(os.path.join(temp_dir, "subdir"))

tests/inference/testing_nodes/testing-pack/util.py

@@ -259,7 +259,7 @@ class TestForLoopOpen:
         graph = GraphBuilder()
         if "initial_value0" in kwargs:
             remaining = kwargs["initial_value0"]
-        while_open = graph.node("TestWhileLoopOpen", condition=remaining, initial_value0=remaining, **{(f"initial_value{i}"): kwargs.get(f"initial_value{i}", None) for i in range(1, NUM_FLOW_SOCKETS)})
+        graph.node("TestWhileLoopOpen", condition=remaining, initial_value0=remaining, **{(f"initial_value{i}"): kwargs.get(f"initial_value{i}", None) for i in range(1, NUM_FLOW_SOCKETS)})
         outputs = [kwargs.get(f"initial_value{i}", None) for i in range(1, NUM_FLOW_SOCKETS)]
         return {
             "result": tuple(["stub", remaining] + outputs),

web/assets/DownloadGitView-DyhrHmlh.js

@@ -0,0 +1,58 @@
+var __defProp = Object.defineProperty;
+var __name = (target, value) => __defProp(target, "name", { value, configurable: true });
+import { a as defineComponent, f as openBlock, g as createElementBlock, A as createBaseVNode, a8 as toDisplayString, h as createVNode, z as unref, D as script, bU as useRouter } from "./index-CSl7lfOs.js";
+const _hoisted_1 = { class: "font-sans w-screen h-screen mx-0 grid place-items-center justify-center items-center text-neutral-900 bg-neutral-300 pointer-events-auto" };
+const _hoisted_2 = { class: "col-start-1 h-screen row-start-1 place-content-center mx-auto overflow-y-auto" };
+const _hoisted_3 = { class: "max-w-screen-sm flex flex-col gap-8 p-8 bg-[url('/assets/images/Git-Logo-White.svg')] bg-no-repeat bg-right-top bg-origin-padding" };
+const _hoisted_4 = { class: "mt-24 text-4xl font-bold text-red-500" };
+const _hoisted_5 = { class: "space-y-4" };
+const _hoisted_6 = { class: "text-xl" };
+const _hoisted_7 = { class: "text-xl" };
+const _hoisted_8 = { class: "text-m" };
+const _hoisted_9 = { class: "flex gap-4 flex-row-reverse" };
+const _sfc_main = /* @__PURE__ */ defineComponent({
+  __name: "DownloadGitView",
+  setup(__props) {
+    const openGitDownloads = /* @__PURE__ */ __name(() => {
+      window.open("https://git-scm.com/downloads/", "_blank");
+    }, "openGitDownloads");
+    const skipGit = /* @__PURE__ */ __name(() => {
+      console.warn("pushing");
+      const router = useRouter();
+      router.push("install");
+    }, "skipGit");
+    return (_ctx, _cache) => {
+      return openBlock(), createElementBlock("div", _hoisted_1, [
+        createBaseVNode("div", _hoisted_2, [
+          createBaseVNode("div", _hoisted_3, [
+            createBaseVNode("h1", _hoisted_4, toDisplayString(_ctx.$t("downloadGit.title")), 1),
+            createBaseVNode("div", _hoisted_5, [
+              createBaseVNode("p", _hoisted_6, toDisplayString(_ctx.$t("downloadGit.message")), 1),
+              createBaseVNode("p", _hoisted_7, toDisplayString(_ctx.$t("downloadGit.instructions")), 1),
+              createBaseVNode("p", _hoisted_8, toDisplayString(_ctx.$t("downloadGit.warning")), 1)
+            ]),
+            createBaseVNode("div", _hoisted_9, [
+              createVNode(unref(script), {
+                label: _ctx.$t("downloadGit.gitWebsite"),
+                icon: "pi pi-external-link",
+                "icon-pos": "right",
+                onClick: openGitDownloads,
+                severity: "primary"
+              }, null, 8, ["label"]),
+              createVNode(unref(script), {
+                label: _ctx.$t("downloadGit.skip"),
+                icon: "pi pi-exclamation-triangle",
+                onClick: skipGit,
+                severity: "secondary"
+              }, null, 8, ["label"])
+            ])
+          ])
+        ])
+      ]);
+    };
+  }
+});
+export {
+  _sfc_main as default
+};
+//# sourceMappingURL=DownloadGitView-DyhrHmlh.js.map

web/assets/DownloadGitView-DyhrHmlh.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"DownloadGitView-DyhrHmlh.js","sources":["../../src/views/DownloadGitView.vue"],"sourcesContent":["<template>\n  <div\n    class=\"font-sans w-screen h-screen mx-0 grid place-items-center justify-center items-center text-neutral-900 bg-neutral-300 pointer-events-auto\"\n  >\n    <div\n      class=\"col-start-1 h-screen row-start-1 place-content-center mx-auto overflow-y-auto\"\n    >\n      <div\n        class=\"max-w-screen-sm flex flex-col gap-8 p-8 bg-[url('/assets/images/Git-Logo-White.svg')] bg-no-repeat bg-right-top bg-origin-padding\"\n      >\n        <!-- Header -->\n        <h1 class=\"mt-24 text-4xl font-bold text-red-500\">\n          {{ $t('downloadGit.title') }}\n        </h1>\n\n        <!-- Message -->\n        <div class=\"space-y-4\">\n          <p class=\"text-xl\">\n            {{ $t('downloadGit.message') }}\n          </p>\n          <p class=\"text-xl\">\n            {{ $t('downloadGit.instructions') }}\n          </p>\n          <p class=\"text-m\">\n            {{ $t('downloadGit.warning') }}\n          </p>\n        </div>\n\n        <!-- Actions -->\n        <div class=\"flex gap-4 flex-row-reverse\">\n          <Button\n            :label=\"$t('downloadGit.gitWebsite')\"\n            icon=\"pi pi-external-link\"\n            icon-pos=\"right\"\n            @click=\"openGitDownloads\"\n            severity=\"primary\"\n          />\n          <Button\n            :label=\"$t('downloadGit.skip')\"\n            icon=\"pi pi-exclamation-triangle\"\n            @click=\"skipGit\"\n            severity=\"secondary\"\n          />\n        </div>\n      </div>\n    </div>\n  </div>\n</template>\n\n<script setup lang=\"ts\">\nimport Button from 'primevue/button'\nimport { useRouter } from 'vue-router'\n\nconst openGitDownloads = () => {\n  window.open('https://git-scm.com/downloads/', '_blank')\n}\n\nconst skipGit = () => {\n  console.warn('pushing')\n  const router = useRouter()\n  router.push('install')\n}\n</script>\n"],"names":[],"mappings":";;;;;;;;;;;;;;;AAqDA,UAAM,mBAAmB,6BAAM;AACtB,aAAA,KAAK,kCAAkC,QAAQ;AAAA,IAAA,GAD/B;AAIzB,UAAM,UAAU,6BAAM;AACpB,cAAQ,KAAK,SAAS;AACtB,YAAM,SAAS;AACf,aAAO,KAAK,SAAS;AAAA,IAAA,GAHP;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;"}

web/assets/ExtensionPanel-DgaZovwi.js

@@ -1,8 +1,8 @@
 var __defProp = Object.defineProperty;
 var __name = (target, value) => __defProp(target, "name", { value, configurable: true });
-import { d as defineComponent, r as ref, c6 as FilterMatchMode, ca as useExtensionStore, u as useSettingStore, o as onMounted, q as computed, g as openBlock, x as createBlock, y as withCtx, i as createVNode, c7 as SearchBox, z as unref, bT as script, A as createBaseVNode, h as createElementBlock, O as renderList, a6 as toDisplayString, aw as createTextVNode, N as Fragment, D as script$1, j as createCommentVNode, bV as script$3, c8 as _sfc_main$1 } from "./index-CoOvI8ZH.js";
-import { s as script$2, a as script$4 } from "./index-DK6Kev7f.js";
-import "./index-D4DWQPPQ.js";
+import { a as defineComponent, r as ref, ck as FilterMatchMode, co as useExtensionStore, u as useSettingStore, o as onMounted, q as computed, f as openBlock, x as createBlock, y as withCtx, h as createVNode, cl as SearchBox, z as unref, bW as script, A as createBaseVNode, g as createElementBlock, Q as renderList, a8 as toDisplayString, ay as createTextVNode, P as Fragment, D as script$1, i as createCommentVNode, c5 as script$3, cm as _sfc_main$1 } from "./index-CSl7lfOs.js";
+import { s as script$2, a as script$4 } from "./index-CgmI-OoW.js";
+import "./index-aSkd2KAK.js";
 const _hoisted_1 = { class: "flex justify-end" };
 const _sfc_main = /* @__PURE__ */ defineComponent({
   __name: "ExtensionPanel",
@@ -47,7 +47,7 @@ const _sfc_main = /* @__PURE__ */ defineComponent({
           createVNode(SearchBox, {
             modelValue: filters.value["global"].value,
             "onUpdate:modelValue": _cache[0] || (_cache[0] = ($event) => filters.value["global"].value = $event),
-            placeholder: _ctx.$t("searchExtensions") + "..."
+            placeholder: _ctx.$t("g.searchExtensions") + "..."
           }, null, 8, ["modelValue", "placeholder"]),
           hasChanges.value ? (openBlock(), createBlock(unref(script), {
             key: 0,
@@ -67,7 +67,7 @@ const _sfc_main = /* @__PURE__ */ defineComponent({
               ]),
               createBaseVNode("div", _hoisted_1, [
                 createVNode(unref(script$1), {
-                  label: _ctx.$t("reloadToApplyChanges"),
+                  label: _ctx.$t("g.reloadToApplyChanges"),
                   onClick: applyChanges,
                   outlined: "",
                   severity: "danger"
@@ -87,7 +87,7 @@ const _sfc_main = /* @__PURE__ */ defineComponent({
             default: withCtx(() => [
               createVNode(unref(script$2), {
                 field: "name",
-                header: _ctx.$t("extensionName"),
+                header: _ctx.$t("g.extensionName"),
                 sortable: ""
               }, null, 8, ["header"]),
               createVNode(unref(script$2), { pt: {
@@ -114,4 +114,4 @@ const _sfc_main = /* @__PURE__ */ defineComponent({
 export {
   _sfc_main as default
 };
-//# sourceMappingURL=ExtensionPanel-DsD42OtO.js.map
+//# sourceMappingURL=ExtensionPanel-DgaZovwi.js.map

web/assets/ExtensionPanel-DgaZovwi.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"ExtensionPanel-DgaZovwi.js","sources":["../../src/components/dialog/content/setting/ExtensionPanel.vue"],"sourcesContent":["<template>\n  <PanelTemplate value=\"Extension\" class=\"extension-panel\">\n    <template #header>\n      <SearchBox\n        v-model=\"filters['global'].value\"\n        :placeholder=\"$t('g.searchExtensions') + '...'\"\n      />\n      <Message v-if=\"hasChanges\" severity=\"info\" pt:text=\"w-full\">\n        <ul>\n          <li v-for=\"ext in changedExtensions\" :key=\"ext.name\">\n            <span>\n              {{ extensionStore.isExtensionEnabled(ext.name) ? '[-]' : '[+]' }}\n            </span>\n            {{ ext.name }}\n          </li>\n        </ul>\n        <div class=\"flex justify-end\">\n          <Button\n            :label=\"$t('g.reloadToApplyChanges')\"\n            @click=\"applyChanges\"\n            outlined\n            severity=\"danger\"\n          />\n        </div>\n      </Message>\n    </template>\n    <DataTable\n      :value=\"extensionStore.extensions\"\n      stripedRows\n      size=\"small\"\n      :filters=\"filters\"\n    >\n      <Column field=\"name\" :header=\"$t('g.extensionName')\" sortable></Column>\n      <Column\n        :pt=\"{\n          bodyCell: 'flex items-center justify-end'\n        }\"\n      >\n        <template #body=\"slotProps\">\n          <ToggleSwitch\n            v-model=\"editingEnabledExtensions[slotProps.data.name]\"\n            @change=\"updateExtensionStatus\"\n          />\n        </template>\n      </Column>\n    </DataTable>\n  </PanelTemplate>\n</template>\n\n<script setup lang=\"ts\">\nimport { ref, computed, onMounted } from 'vue'\nimport { useExtensionStore } from '@/stores/extensionStore'\nimport { useSettingStore } from '@/stores/settingStore'\nimport DataTable from 'primevue/datatable'\nimport Column from 'primevue/column'\nimport ToggleSwitch from 'primevue/toggleswitch'\nimport Button from 'primevue/button'\nimport Message from 'primevue/message'\nimport { FilterMatchMode } from '@primevue/core/api'\nimport PanelTemplate from './PanelTemplate.vue'\nimport SearchBox from '@/components/common/SearchBox.vue'\n\nconst filters = ref({\n  global: { value: '', matchMode: FilterMatchMode.CONTAINS }\n})\n\nconst extensionStore = useExtensionStore()\nconst settingStore = useSettingStore()\n\nconst editingEnabledExtensions = ref<Record<string, boolean>>({})\n\nonMounted(() => {\n  extensionStore.extensions.forEach((ext) => {\n    editingEnabledExtensions.value[ext.name] =\n      extensionStore.isExtensionEnabled(ext.name)\n  })\n})\n\nconst changedExtensions = computed(() => {\n  return extensionStore.extensions.filter(\n    (ext) =>\n      editingEnabledExtensions.value[ext.name] !==\n      extensionStore.isExtensionEnabled(ext.name)\n  )\n})\n\nconst hasChanges = computed(() => {\n  return changedExtensions.value.length > 0\n})\n\nconst updateExtensionStatus = () => {\n  const editingDisabledExtensionNames = Object.entries(\n    editingEnabledExtensions.value\n  )\n    .filter(([_, enabled]) => !enabled)\n    .map(([name]) => name)\n\n  settingStore.set('Comfy.Extension.Disabled', [\n    ...extensionStore.inactiveDisabledExtensionNames,\n    ...editingDisabledExtensionNames\n  ])\n}\n\nconst applyChanges = () => {\n  // Refresh the page to apply changes\n  window.location.reload()\n}\n</script>\n"],"names":[],"mappings":";;;;;;;;;AA8DA,UAAM,UAAU,IAAI;AAAA,MAClB,QAAQ,EAAE,OAAO,IAAI,WAAW,gBAAgB,SAAS;AAAA,IAAA,CAC1D;AAED,UAAM,iBAAiB;AACvB,UAAM,eAAe;AAEf,UAAA,2BAA2B,IAA6B,CAAA,CAAE;AAEhE,cAAU,MAAM;AACC,qBAAA,WAAW,QAAQ,CAAC,QAAQ;AACzC,iCAAyB,MAAM,IAAI,IAAI,IACrC,eAAe,mBAAmB,IAAI,IAAI;AAAA,MAAA,CAC7C;AAAA,IAAA,CACF;AAEK,UAAA,oBAAoB,SAAS,MAAM;AACvC,aAAO,eAAe,WAAW;AAAA,QAC/B,CAAC,QACC,yBAAyB,MAAM,IAAI,IAAI,MACvC,eAAe,mBAAmB,IAAI,IAAI;AAAA,MAAA;AAAA,IAC9C,CACD;AAEK,UAAA,aAAa,SAAS,MAAM;AACzB,aAAA,kBAAkB,MAAM,SAAS;AAAA,IAAA,CACzC;AAED,UAAM,wBAAwB,6BAAM;AAClC,YAAM,gCAAgC,OAAO;AAAA,QAC3C,yBAAyB;AAAA,MAExB,EAAA,OAAO,CAAC,CAAC,GAAG,OAAO,MAAM,CAAC,OAAO,EACjC,IAAI,CAAC,CAAC,IAAI,MAAM,IAAI;AAEvB,mBAAa,IAAI,4BAA4B;AAAA,QAC3C,GAAG,eAAe;AAAA,QAClB,GAAG;AAAA,MAAA,CACJ;AAAA,IAAA,GAV2B;AAa9B,UAAM,eAAe,6BAAM;AAEzB,aAAO,SAAS;IAAO,GAFJ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;"}

web/assets/ExtensionPanel-DsD42OtO.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"ExtensionPanel-DsD42OtO.js","sources":["../../src/components/dialog/content/setting/ExtensionPanel.vue"],"sourcesContent":["<template>\n  <PanelTemplate value=\"Extension\" class=\"extension-panel\">\n    <template #header>\n      <SearchBox\n        v-model=\"filters['global'].value\"\n        :placeholder=\"$t('searchExtensions') + '...'\"\n      />\n      <Message v-if=\"hasChanges\" severity=\"info\" pt:text=\"w-full\">\n        <ul>\n          <li v-for=\"ext in changedExtensions\" :key=\"ext.name\">\n            <span>\n              {{ extensionStore.isExtensionEnabled(ext.name) ? '[-]' : '[+]' }}\n            </span>\n            {{ ext.name }}\n          </li>\n        </ul>\n        <div class=\"flex justify-end\">\n          <Button\n            :label=\"$t('reloadToApplyChanges')\"\n            @click=\"applyChanges\"\n            outlined\n            severity=\"danger\"\n          />\n        </div>\n      </Message>\n    </template>\n    <DataTable\n      :value=\"extensionStore.extensions\"\n      stripedRows\n      size=\"small\"\n      :filters=\"filters\"\n    >\n      <Column field=\"name\" :header=\"$t('extensionName')\" sortable></Column>\n      <Column\n        :pt=\"{\n          bodyCell: 'flex items-center justify-end'\n        }\"\n      >\n        <template #body=\"slotProps\">\n          <ToggleSwitch\n            v-model=\"editingEnabledExtensions[slotProps.data.name]\"\n            @change=\"updateExtensionStatus\"\n          />\n        </template>\n      </Column>\n    </DataTable>\n  </PanelTemplate>\n</template>\n\n<script setup lang=\"ts\">\nimport { ref, computed, onMounted } from 'vue'\nimport { useExtensionStore } from '@/stores/extensionStore'\nimport { useSettingStore } from '@/stores/settingStore'\nimport DataTable from 'primevue/datatable'\nimport Column from 'primevue/column'\nimport ToggleSwitch from 'primevue/toggleswitch'\nimport Button from 'primevue/button'\nimport Message from 'primevue/message'\nimport { FilterMatchMode } from '@primevue/core/api'\nimport PanelTemplate from './PanelTemplate.vue'\nimport SearchBox from '@/components/common/SearchBox.vue'\n\nconst filters = ref({\n  global: { value: '', matchMode: FilterMatchMode.CONTAINS }\n})\n\nconst extensionStore = useExtensionStore()\nconst settingStore = useSettingStore()\n\nconst editingEnabledExtensions = ref<Record<string, boolean>>({})\n\nonMounted(() => {\n  extensionStore.extensions.forEach((ext) => {\n    editingEnabledExtensions.value[ext.name] =\n      extensionStore.isExtensionEnabled(ext.name)\n  })\n})\n\nconst changedExtensions = computed(() => {\n  return extensionStore.extensions.filter(\n    (ext) =>\n      editingEnabledExtensions.value[ext.name] !==\n      extensionStore.isExtensionEnabled(ext.name)\n  )\n})\n\nconst hasChanges = computed(() => {\n  return changedExtensions.value.length > 0\n})\n\nconst updateExtensionStatus = () => {\n  const editingDisabledExtensionNames = Object.entries(\n    editingEnabledExtensions.value\n  )\n    .filter(([_, enabled]) => !enabled)\n    .map(([name]) => name)\n\n  settingStore.set('Comfy.Extension.Disabled', [\n    ...extensionStore.inactiveDisabledExtensionNames,\n    ...editingDisabledExtensionNames\n  ])\n}\n\nconst applyChanges = () => {\n  // Refresh the page to apply changes\n  window.location.reload()\n}\n</script>\n"],"names":[],"mappings":";;;;;;;;;AA8DA,UAAM,UAAU,IAAI;AAAA,MAClB,QAAQ,EAAE,OAAO,IAAI,WAAW,gBAAgB,SAAS;AAAA,IAAA,CAC1D;AAED,UAAM,iBAAiB;AACvB,UAAM,eAAe;AAEf,UAAA,2BAA2B,IAA6B,CAAA,CAAE;AAEhE,cAAU,MAAM;AACC,qBAAA,WAAW,QAAQ,CAAC,QAAQ;AACzC,iCAAyB,MAAM,IAAI,IAAI,IACrC,eAAe,mBAAmB,IAAI,IAAI;AAAA,MAAA,CAC7C;AAAA,IAAA,CACF;AAEK,UAAA,oBAAoB,SAAS,MAAM;AACvC,aAAO,eAAe,WAAW;AAAA,QAC/B,CAAC,QACC,yBAAyB,MAAM,IAAI,IAAI,MACvC,eAAe,mBAAmB,IAAI,IAAI;AAAA,MAAA;AAAA,IAC9C,CACD;AAEK,UAAA,aAAa,SAAS,MAAM;AACzB,aAAA,kBAAkB,MAAM,SAAS;AAAA,IAAA,CACzC;AAED,UAAM,wBAAwB,6BAAM;AAClC,YAAM,gCAAgC,OAAO;AAAA,QAC3C,yBAAyB;AAAA,MAExB,EAAA,OAAO,CAAC,CAAC,GAAG,OAAO,MAAM,CAAC,OAAO,EACjC,IAAI,CAAC,CAAC,IAAI,MAAM,IAAI;AAEvB,mBAAa,IAAI,4BAA4B;AAAA,QAC3C,GAAG,eAAe;AAAA,QAClB,GAAG;AAAA,MAAA,CACJ;AAAA,IAAA,GAV2B;AAa9B,UAAM,eAAe,6BAAM;AAEzB,aAAO,SAAS;IAAO,GAFJ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;"}

web/assets/GraphView-B3TpSwhZ.css

@@ -45,7 +45,7 @@
   --sidebar-icon-size: 1rem;
 }
 
-.side-tool-bar-container[data-v-e0812a25] {
+.side-tool-bar-container[data-v-7851c166] {
   display: flex;
   flex-direction: column;
   align-items: center;
@@ -55,10 +55,11 @@
   width: var(--sidebar-width);
   height: 100%;
 
-  background-color: var(--comfy-menu-bg);
+  background-color: var(--comfy-menu-secondary-bg);
   color: var(--fg-color);
+  box-shadow: var(--bar-shadow);
 }
-.side-tool-bar-end[data-v-e0812a25] {
+.side-tool-bar-end[data-v-7851c166] {
   align-self: flex-end;
   margin-top: auto;
 }
@@ -97,7 +98,7 @@
   z-index: 999;
 }
 
-[data-v-37f672ab] .highlight {
+[data-v-d7cc0bce] .highlight {
   background-color: var(--p-primary-color);
   color: var(--p-primary-contrast-color);
   font-weight: bold;
@@ -124,7 +125,7 @@
   align-items: flex-start !important;
 }
 
-.node-tooltip[data-v-c2e0098f] {
+.node-tooltip[data-v-9ecc8adc] {
   background: var(--comfy-input-bg);
   border-radius: 5px;
   box-shadow: 0 0 5px rgba(0, 0, 0, 0.4);
@@ -152,31 +153,30 @@
   border-radius: 0;
 }
 
-.comfy-menu-hamburger[data-v-2ddd26e8] {
+.comfy-menu-hamburger[data-v-962c4073] {
   pointer-events: auto;
   position: fixed;
   z-index: 9999;
 }
 
-[data-v-783f8efe] .p-togglebutton::before {
+[data-v-4cb762cb] .p-togglebutton::before {
   display: none
 }
-[data-v-783f8efe] .p-togglebutton {
+[data-v-4cb762cb] .p-togglebutton {
   position: relative;
   flex-shrink: 0;
   border-radius: 0px;
   background-color: transparent;
-  padding-left: 0.5rem;
-  padding-right: 0.5rem
+  padding: 0px
 }
-[data-v-783f8efe] .p-togglebutton.p-togglebutton-checked {
+[data-v-4cb762cb] .p-togglebutton.p-togglebutton-checked {
   border-bottom-width: 2px;
   border-bottom-color: var(--p-button-text-primary-color)
 }
-[data-v-783f8efe] .p-togglebutton-checked .close-button,[data-v-783f8efe] .p-togglebutton:hover .close-button {
+[data-v-4cb762cb] .p-togglebutton-checked .close-button,[data-v-4cb762cb] .p-togglebutton:hover .close-button {
   visibility: visible
 }
-.status-indicator[data-v-783f8efe] {
+.status-indicator[data-v-4cb762cb] {
   position: absolute;
   font-weight: 700;
   font-size: 1.5rem;
@@ -184,22 +184,22 @@
   left: 50%;
   transform: translate(-50%, -50%)
 }
-[data-v-783f8efe] .p-togglebutton:hover .status-indicator {
+[data-v-4cb762cb] .p-togglebutton:hover .status-indicator {
   display: none
 }
-[data-v-783f8efe] .p-togglebutton .close-button {
+[data-v-4cb762cb] .p-togglebutton .close-button {
   visibility: hidden
 }
 
-.top-menubar[data-v-9646ca0a] .p-menubar-item-link svg {
+.top-menubar[data-v-a2b12676] .p-menubar-item-link svg {
   display: none;
 }
-[data-v-9646ca0a] .p-menubar-submenu.dropdown-direction-up {
+[data-v-a2b12676] .p-menubar-submenu.dropdown-direction-up {
   top: auto;
   bottom: 100%;
   flex-direction: column-reverse;
 }
-.keybinding-tag[data-v-9646ca0a] {
+.keybinding-tag[data-v-a2b12676] {
   background: var(--p-content-hover-background);
   border-color: var(--p-content-border-color);
   border-style: solid;
@@ -210,7 +210,7 @@
   border-bottom-left-radius: 0;
 }
 
-.comfyui-queue-button[data-v-95bc9be0] .p-splitbutton-dropdown {
+.comfyui-queue-button[data-v-d3897845] .p-splitbutton-dropdown {
   border-top-right-radius: 0;
   border-bottom-right-radius: 0;
 }
@@ -238,10 +238,11 @@
   display: none;
 }
 
-.comfyui-menu[data-v-d84a704d] {
+.comfyui-menu[data-v-d792da31] {
   width: 100vw;
   background: var(--comfy-menu-bg);
   color: var(--fg-color);
+  box-shadow: var(--bar-shadow);
   font-family: Arial, Helvetica, sans-serif;
   font-size: 0.8em;
   box-sizing: border-box;
@@ -250,13 +251,16 @@
   grid-column: 1/-1;
   max-height: 90vh;
 }
-.comfyui-menu.dropzone[data-v-d84a704d] {
+.comfyui-menu.dropzone[data-v-d792da31] {
   background: var(--p-highlight-background);
 }
-.comfyui-menu.dropzone-active[data-v-d84a704d] {
+.comfyui-menu.dropzone-active[data-v-d792da31] {
   background: var(--p-highlight-background-focus);
 }
-.comfyui-logo[data-v-d84a704d] {
+[data-v-d792da31] .p-menubar-item-label {
+  line-height: revert;
+}
+.comfyui-logo[data-v-d792da31] {
   font-size: 1.2em;
   -webkit-user-select: none;
      -moz-user-select: none;

web/assets/InstallView-8N2LdZUx.css

@@ -1,4 +1,4 @@
 
-[data-v-53e62b05] .p-steppanel {
+[data-v-7ef01cf2] .p-steppanel {
     background-color: transparent
 }

web/assets/KeybindingPanel-BlOA8Yhu.js

@@ -1,8 +1,8 @@
 var __defProp = Object.defineProperty;
 var __name = (target, value) => __defProp(target, "name", { value, configurable: true });
-import { d as defineComponent, q as computed, g as openBlock, h as createElementBlock, N as Fragment, O as renderList, i as createVNode, y as withCtx, aw as createTextVNode, a6 as toDisplayString, z as unref, aA as script, j as createCommentVNode, r as ref, c6 as FilterMatchMode, M as useKeybindingStore, F as useCommandStore, aJ as watchEffect, bg as useToast, t as resolveDirective, x as createBlock, c7 as SearchBox, A as createBaseVNode, D as script$2, ao as script$4, bk as withModifiers, bT as script$5, aH as script$6, v as withDirectives, c8 as _sfc_main$2, P as pushScopeId, Q as popScopeId, c1 as KeyComboImpl, c9 as KeybindingImpl, _ as _export_sfc } from "./index-CoOvI8ZH.js";
-import { s as script$1, a as script$3 } from "./index-DK6Kev7f.js";
-import "./index-D4DWQPPQ.js";
+import { a as defineComponent, q as computed, f as openBlock, g as createElementBlock, P as Fragment, Q as renderList, h as createVNode, y as withCtx, ay as createTextVNode, a8 as toDisplayString, z as unref, aC as script, i as createCommentVNode, r as ref, ck as FilterMatchMode, O as useKeybindingStore, F as useCommandStore, I as useI18n, aS as normalizeI18nKey, aL as watchEffect, bn as useToast, t as resolveDirective, x as createBlock, cl as SearchBox, A as createBaseVNode, D as script$2, aq as script$4, br as withModifiers, bW as script$5, aI as script$6, v as withDirectives, cm as _sfc_main$2, R as pushScopeId, U as popScopeId, ce as KeyComboImpl, cn as KeybindingImpl, _ as _export_sfc } from "./index-CSl7lfOs.js";
+import { s as script$1, a as script$3 } from "./index-CgmI-OoW.js";
+import "./index-aSkd2KAK.js";
 const _hoisted_1$1 = {
   key: 0,
   class: "px-2"
@@ -35,7 +35,7 @@ const _sfc_main$1 = /* @__PURE__ */ defineComponent({
     };
   }
 });
-const _withScopeId = /* @__PURE__ */ __name((n) => (pushScopeId("data-v-9d7e362e"), n = n(), popScopeId(), n), "_withScopeId");
+const _withScopeId = /* @__PURE__ */ __name((n) => (pushScopeId("data-v-c20ad403"), n = n(), popScopeId(), n), "_withScopeId");
 const _hoisted_1 = { class: "actions invisible flex flex-row" };
 const _hoisted_2 = ["title"];
 const _hoisted_3 = { key: 1 };
@@ -47,9 +47,11 @@ const _sfc_main = /* @__PURE__ */ defineComponent({
     });
     const keybindingStore = useKeybindingStore();
     const commandStore = useCommandStore();
+    const { t } = useI18n();
     const commandsData = computed(() => {
       return Object.values(commandStore.commands).map((command) => ({
         id: command.id,
+        label: t(`commands.${normalizeI18nKey(command.id)}.label`, command.label),
         keybinding: keybindingStore.getKeybindingByCommandId(command.id)
       }));
     });
@@ -140,7 +142,7 @@ const _sfc_main = /* @__PURE__ */ defineComponent({
           createVNode(SearchBox, {
             modelValue: filters.value["global"].value,
             "onUpdate:modelValue": _cache[0] || (_cache[0] = ($event) => filters.value["global"].value = $event),
-            placeholder: _ctx.$t("searchKeybindings") + "..."
+            placeholder: _ctx.$t("g.searchKeybindings") + "..."
           }, null, 8, ["modelValue", "placeholder"])
         ]),
         default: withCtx(() => [
@@ -188,7 +190,7 @@ const _sfc_main = /* @__PURE__ */ defineComponent({
                   createBaseVNode("div", {
                     class: "overflow-hidden text-ellipsis whitespace-nowrap",
                     title: slotProps.data.id
-                  }, toDisplayString(slotProps.data.id), 9, _hoisted_2)
+                  }, toDisplayString(slotProps.data.label), 9, _hoisted_2)
                 ]),
                 _: 1
               }),
@@ -257,14 +259,14 @@ const _sfc_main = /* @__PURE__ */ defineComponent({
           }, 8, ["visible", "header"]),
           withDirectives(createVNode(unref(script$2), {
             class: "mt-4",
-            label: _ctx.$t("reset"),
+            label: _ctx.$t("g.reset"),
             icon: "pi pi-trash",
             severity: "danger",
             fluid: "",
             text: "",
             onClick: resetKeybindings
           }, null, 8, ["label"]), [
-            [_directive_tooltip, _ctx.$t("resetKeybindingsTooltip")]
+            [_directive_tooltip, _ctx.$t("g.resetKeybindingsTooltip")]
           ])
         ]),
         _: 1
@@ -272,8 +274,8 @@ const _sfc_main = /* @__PURE__ */ defineComponent({
     };
   }
 });
-const KeybindingPanel = /* @__PURE__ */ _export_sfc(_sfc_main, [["__scopeId", "data-v-9d7e362e"]]);
+const KeybindingPanel = /* @__PURE__ */ _export_sfc(_sfc_main, [["__scopeId", "data-v-c20ad403"]]);
 export {
   KeybindingPanel as default
 };
-//# sourceMappingURL=KeybindingPanel-lcJrxHwZ.js.map
+//# sourceMappingURL=KeybindingPanel-BlOA8Yhu.js.map

web/assets/KeybindingPanel-C-7KE-Kw.css

@@ -1,8 +0,0 @@
-
-[data-v-9d7e362e] .p-datatable-tbody > tr > td {
-  padding: 0.25rem;
-  min-height: 2rem
-}
-[data-v-9d7e362e] .p-datatable-row-selected .actions,[data-v-9d7e362e] .p-datatable-selectable-row:hover .actions {
-  visibility: visible
-}

web/assets/KeybindingPanel-C3wT8hYZ.css

@@ -0,0 +1,8 @@
+
+[data-v-c20ad403] .p-datatable-tbody > tr > td {
+  padding: 0.25rem;
+  min-height: 2rem
+}
+[data-v-c20ad403] .p-datatable-row-selected .actions,[data-v-c20ad403] .p-datatable-selectable-row:hover .actions {
+  visibility: visible
+}

web/assets/NotSupportedView-Dhitj9aO.js

@@ -0,0 +1,82 @@
+var __defProp = Object.defineProperty;
+var __name = (target, value) => __defProp(target, "name", { value, configurable: true });
+import { a as defineComponent, bU as useRouter, t as resolveDirective, f as openBlock, g as createElementBlock, A as createBaseVNode, a8 as toDisplayString, h as createVNode, z as unref, D as script, v as withDirectives } from "./index-CSl7lfOs.js";
+const _imports_0 = "" + new URL("images/sad_girl.png", import.meta.url).href;
+const _hoisted_1 = { class: "font-sans w-screen h-screen flex items-center m-0 text-neutral-900 bg-neutral-300 pointer-events-auto" };
+const _hoisted_2 = { class: "flex-grow flex items-center justify-center" };
+const _hoisted_3 = { class: "flex flex-col gap-8 p-8" };
+const _hoisted_4 = { class: "text-4xl font-bold text-red-500" };
+const _hoisted_5 = { class: "space-y-4" };
+const _hoisted_6 = { class: "text-xl" };
+const _hoisted_7 = { class: "list-disc list-inside space-y-1 text-neutral-800" };
+const _hoisted_8 = { class: "flex gap-4" };
+const _hoisted_9 = /* @__PURE__ */ createBaseVNode("div", { class: "h-screen flex-grow-0" }, [
+  /* @__PURE__ */ createBaseVNode("img", {
+    src: _imports_0,
+    alt: "Sad girl illustration",
+    class: "h-full object-cover"
+  })
+], -1);
+const _sfc_main = /* @__PURE__ */ defineComponent({
+  __name: "NotSupportedView",
+  setup(__props) {
+    const openDocs = /* @__PURE__ */ __name(() => {
+      window.open(
+        "https://github.com/Comfy-Org/desktop#currently-supported-platforms",
+        "_blank"
+      );
+    }, "openDocs");
+    const reportIssue = /* @__PURE__ */ __name(() => {
+      window.open("https://forum.comfy.org/c/v1-feedback/", "_blank");
+    }, "reportIssue");
+    const router = useRouter();
+    const continueToInstall = /* @__PURE__ */ __name(() => {
+      router.push("/install");
+    }, "continueToInstall");
+    return (_ctx, _cache) => {
+      const _directive_tooltip = resolveDirective("tooltip");
+      return openBlock(), createElementBlock("div", _hoisted_1, [
+        createBaseVNode("div", _hoisted_2, [
+          createBaseVNode("div", _hoisted_3, [
+            createBaseVNode("h1", _hoisted_4, toDisplayString(_ctx.$t("notSupported.title")), 1),
+            createBaseVNode("div", _hoisted_5, [
+              createBaseVNode("p", _hoisted_6, toDisplayString(_ctx.$t("notSupported.message")), 1),
+              createBaseVNode("ul", _hoisted_7, [
+                createBaseVNode("li", null, toDisplayString(_ctx.$t("notSupported.supportedDevices.macos")), 1),
+                createBaseVNode("li", null, toDisplayString(_ctx.$t("notSupported.supportedDevices.windows")), 1)
+              ])
+            ]),
+            createBaseVNode("div", _hoisted_8, [
+              createVNode(unref(script), {
+                label: _ctx.$t("notSupported.learnMore"),
+                icon: "pi pi-github",
+                onClick: openDocs,
+                severity: "secondary"
+              }, null, 8, ["label"]),
+              createVNode(unref(script), {
+                label: _ctx.$t("notSupported.reportIssue"),
+                icon: "pi pi-flag",
+                onClick: reportIssue,
+                severity: "secondary"
+              }, null, 8, ["label"]),
+              withDirectives(createVNode(unref(script), {
+                label: _ctx.$t("notSupported.continue"),
+                icon: "pi pi-arrow-right",
+                iconPos: "right",
+                onClick: continueToInstall,
+                severity: "danger"
+              }, null, 8, ["label"]), [
+                [_directive_tooltip, _ctx.$t("notSupported.continueTooltip")]
+              ])
+            ])
+          ])
+        ]),
+        _hoisted_9
+      ]);
+    };
+  }
+});
+export {
+  _sfc_main as default
+};
+//# sourceMappingURL=NotSupportedView-Dhitj9aO.js.map

web/assets/NotSupportedView-Dhitj9aO.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"NotSupportedView-Dhitj9aO.js","sources":["../../../../../../../assets/images/sad_girl.png","../../src/views/NotSupportedView.vue"],"sourcesContent":["export default \"__VITE_PUBLIC_ASSET__b82952e7__\"","<template>\n  <div\n    class=\"font-sans w-screen h-screen flex items-center m-0 text-neutral-900 bg-neutral-300 pointer-events-auto\"\n  >\n    <div class=\"flex-grow flex items-center justify-center\">\n      <div class=\"flex flex-col gap-8 p-8\">\n        <!-- Header -->\n        <h1 class=\"text-4xl font-bold text-red-500\">\n          {{ $t('notSupported.title') }}\n        </h1>\n\n        <!-- Message -->\n        <div class=\"space-y-4\">\n          <p class=\"text-xl\">\n            {{ $t('notSupported.message') }}\n          </p>\n          <ul class=\"list-disc list-inside space-y-1 text-neutral-800\">\n            <li>{{ $t('notSupported.supportedDevices.macos') }}</li>\n            <li>{{ $t('notSupported.supportedDevices.windows') }}</li>\n          </ul>\n        </div>\n\n        <!-- Actions -->\n        <div class=\"flex gap-4\">\n          <Button\n            :label=\"$t('notSupported.learnMore')\"\n            icon=\"pi pi-github\"\n            @click=\"openDocs\"\n            severity=\"secondary\"\n          />\n          <Button\n            :label=\"$t('notSupported.reportIssue')\"\n            icon=\"pi pi-flag\"\n            @click=\"reportIssue\"\n            severity=\"secondary\"\n          />\n          <Button\n            :label=\"$t('notSupported.continue')\"\n            icon=\"pi pi-arrow-right\"\n            iconPos=\"right\"\n            @click=\"continueToInstall\"\n            severity=\"danger\"\n            v-tooltip=\"$t('notSupported.continueTooltip')\"\n          />\n        </div>\n      </div>\n    </div>\n\n    <!-- Right side image -->\n    <div class=\"h-screen flex-grow-0\">\n      <img\n        src=\"/assets/images/sad_girl.png\"\n        alt=\"Sad girl illustration\"\n        class=\"h-full object-cover\"\n      />\n    </div>\n  </div>\n</template>\n\n<script setup lang=\"ts\">\nimport Button from 'primevue/button'\nimport { useRouter } from 'vue-router'\n\nconst openDocs = () => {\n  window.open(\n    'https://github.com/Comfy-Org/desktop#currently-supported-platforms',\n    '_blank'\n  )\n}\n\nconst reportIssue = () => {\n  window.open('https://forum.comfy.org/c/v1-feedback/', '_blank')\n}\n\nconst router = useRouter()\nconst continueToInstall = () => {\n  router.push('/install')\n}\n</script>\n"],"names":[],"mappings":";;;AAAA,MAAe,aAAA,KAAA,IAAA,IAAA,uBAAA,YAAA,GAAA,EAAA;;;;;;;;;;;;;;;;;;;AC+Df,UAAM,WAAW,6BAAM;AACd,aAAA;AAAA,QACL;AAAA,QACA;AAAA,MAAA;AAAA,IACF,GAJe;AAOjB,UAAM,cAAc,6BAAM;AACjB,aAAA,KAAK,0CAA0C,QAAQ;AAAA,IAAA,GAD5C;AAIpB,UAAM,SAAS;AACf,UAAM,oBAAoB,6BAAM;AAC9B,aAAO,KAAK,UAAU;AAAA,IAAA,GADE;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;"}

web/assets/ServerConfigPanel-6N6BTSXC.js

@@ -1,7 +1,7 @@
 var __defProp = Object.defineProperty;
 var __name = (target, value) => __defProp(target, "name", { value, configurable: true });
-import { A as createBaseVNode, g as openBlock, h as createElementBlock, aU as markRaw, d as defineComponent, u as useSettingStore, bw as storeToRefs, w as watch, cy as useCopyToClipboard, x as createBlock, y as withCtx, z as unref, bT as script, a6 as toDisplayString, O as renderList, N as Fragment, i as createVNode, D as script$1, j as createCommentVNode, bI as script$2, cz as formatCamelCase, cA as FormItem, c8 as _sfc_main$1, bN as electronAPI } from "./index-CoOvI8ZH.js";
-import { u as useServerConfigStore } from "./serverConfigStore-cctR8PGG.js";
+import { A as createBaseVNode, f as openBlock, g as createElementBlock, aZ as markRaw, a as defineComponent, u as useSettingStore, aK as storeToRefs, w as watch, cL as useCopyToClipboard, I as useI18n, x as createBlock, y as withCtx, z as unref, bW as script, a8 as toDisplayString, Q as renderList, P as Fragment, h as createVNode, D as script$1, i as createCommentVNode, bN as script$2, cM as FormItem, cm as _sfc_main$1, bZ as electronAPI } from "./index-CSl7lfOs.js";
+import { u as useServerConfigStore } from "./serverConfigStore-D4vD2qBB.js";
 const _hoisted_1$1 = {
   viewBox: "0 0 24 24",
   width: "1.2em",
@@ -54,6 +54,14 @@ const _sfc_main = /* @__PURE__ */ defineComponent({
     const copyCommandLineArgs = /* @__PURE__ */ __name(async () => {
       await copyToClipboard(commandLineArgs.value);
     }, "copyCommandLineArgs");
+    const { t } = useI18n();
+    const translateItem = /* @__PURE__ */ __name((item) => {
+      return {
+        ...item,
+        name: t(`serverConfigItems.${item.id}.name`, item.name),
+        tooltip: item.tooltip ? t(`serverConfigItems.${item.id}.tooltip`, item.tooltip) : void 0
+      };
+    }, "translateItem");
     return (_ctx, _cache) => {
       const _component_i_lucide58terminal = __unplugin_components_0;
       return openBlock(), createBlock(_sfc_main$1, {
@@ -119,14 +127,14 @@ const _sfc_main = /* @__PURE__ */ defineComponent({
           (openBlock(true), createElementBlock(Fragment, null, renderList(Object.entries(unref(serverConfigsByCategory)), ([label, items], i) => {
             return openBlock(), createElementBlock("div", { key: label }, [
               i > 0 ? (openBlock(), createBlock(unref(script$2), { key: 0 })) : createCommentVNode("", true),
-              createBaseVNode("h3", null, toDisplayString(unref(formatCamelCase)(label)), 1),
+              createBaseVNode("h3", null, toDisplayString(_ctx.$t(`serverConfigCategories.${label}`, label)), 1),
               (openBlock(true), createElementBlock(Fragment, null, renderList(items, (item) => {
                 return openBlock(), createElementBlock("div", {
                   key: item.name,
                   class: "flex items-center mb-4"
                 }, [
                   createVNode(FormItem, {
-                    item,
+                    item: translateItem(item),
                     formValue: item.value,
                     "onUpdate:formValue": /* @__PURE__ */ __name(($event) => item.value = $event, "onUpdate:formValue"),
                     id: item.id,
@@ -147,4 +155,4 @@ const _sfc_main = /* @__PURE__ */ defineComponent({
 export {
   _sfc_main as default
 };
-//# sourceMappingURL=ServerConfigPanel-x68ubY-c.js.map
+//# sourceMappingURL=ServerConfigPanel-6N6BTSXC.js.map

web/assets/ServerConfigPanel-6N6BTSXC.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"ServerConfigPanel-6N6BTSXC.js","sources":["../../src/components/dialog/content/setting/ServerConfigPanel.vue"],"sourcesContent":["<template>\n  <PanelTemplate value=\"Server-Config\" class=\"server-config-panel\">\n    <template #header>\n      <div class=\"flex flex-col gap-2\">\n        <Message\n          v-if=\"modifiedConfigs.length > 0\"\n          severity=\"info\"\n          pt:text=\"w-full\"\n        >\n          <p>\n            {{ $t('serverConfig.modifiedConfigs') }}\n          </p>\n          <ul>\n            <li v-for=\"config in modifiedConfigs\" :key=\"config.id\">\n              {{ config.name }}: {{ config.initialValue }} → {{ config.value }}\n            </li>\n          </ul>\n          <div class=\"flex justify-end gap-2\">\n            <Button\n              :label=\"$t('serverConfig.revertChanges')\"\n              @click=\"revertChanges\"\n              outlined\n            />\n            <Button\n              :label=\"$t('serverConfig.restart')\"\n              @click=\"restartApp\"\n              outlined\n              severity=\"danger\"\n            />\n          </div>\n        </Message>\n        <Message v-if=\"commandLineArgs\" severity=\"secondary\" pt:text=\"w-full\">\n          <template #icon>\n            <i-lucide:terminal class=\"text-xl font-bold\" />\n          </template>\n          <div class=\"flex items-center justify-between\">\n            <p>{{ commandLineArgs }}</p>\n            <Button\n              icon=\"pi pi-clipboard\"\n              @click=\"copyCommandLineArgs\"\n              severity=\"secondary\"\n              text\n            />\n          </div>\n        </Message>\n      </div>\n    </template>\n    <div\n      v-for=\"([label, items], i) in Object.entries(serverConfigsByCategory)\"\n      :key=\"label\"\n    >\n      <Divider v-if=\"i > 0\" />\n      <h3>{{ $t(`serverConfigCategories.${label}`, label) }}</h3>\n      <div\n        v-for=\"item in items\"\n        :key=\"item.name\"\n        class=\"flex items-center mb-4\"\n      >\n        <FormItem\n          :item=\"translateItem(item)\"\n          v-model:formValue=\"item.value\"\n          :id=\"item.id\"\n          :labelClass=\"{\n            'text-highlight': item.initialValue !== item.value\n          }\"\n        />\n      </div>\n    </div>\n  </PanelTemplate>\n</template>\n\n<script setup lang=\"ts\">\nimport Button from 'primevue/button'\nimport Message from 'primevue/message'\nimport Divider from 'primevue/divider'\nimport FormItem from '@/components/common/FormItem.vue'\nimport PanelTemplate from './PanelTemplate.vue'\nimport { useServerConfigStore } from '@/stores/serverConfigStore'\nimport { storeToRefs } from 'pinia'\nimport { electronAPI } from '@/utils/envUtil'\nimport { useSettingStore } from '@/stores/settingStore'\nimport { watch } from 'vue'\nimport { useCopyToClipboard } from '@/hooks/clipboardHooks'\nimport type { FormItem as FormItemType } from '@/types/settingTypes'\nimport type { ServerConfig } from '@/constants/serverConfig'\nimport { useI18n } from 'vue-i18n'\n\nconst settingStore = useSettingStore()\nconst serverConfigStore = useServerConfigStore()\nconst {\n  serverConfigsByCategory,\n  serverConfigValues,\n  launchArgs,\n  commandLineArgs,\n  modifiedConfigs\n} = storeToRefs(serverConfigStore)\n\nconst revertChanges = () => {\n  serverConfigStore.revertChanges()\n}\n\nconst restartApp = () => {\n  electronAPI().restartApp()\n}\n\nwatch(launchArgs, (newVal) => {\n  settingStore.set('Comfy.Server.LaunchArgs', newVal)\n})\n\nwatch(serverConfigValues, (newVal) => {\n  settingStore.set('Comfy.Server.ServerConfigValues', newVal)\n})\n\nconst { copyToClipboard } = useCopyToClipboard()\nconst copyCommandLineArgs = async () => {\n  await copyToClipboard(commandLineArgs.value)\n}\n\nconst { t } = useI18n()\nconst translateItem = (item: ServerConfig<any>): FormItemType => {\n  return {\n    ...item,\n    name: t(`serverConfigItems.${item.id}.name`, item.name),\n    tooltip: item.tooltip\n      ? t(`serverConfigItems.${item.id}.tooltip`, item.tooltip)\n      : undefined\n  }\n}\n</script>\n"],"names":[],"mappings":";;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AAuFA,UAAM,eAAe;AACrB,UAAM,oBAAoB;AACpB,UAAA;AAAA,MACJ;AAAA,MACA;AAAA,MACA;AAAA,MACA;AAAA,MACA;AAAA,IAAA,IACE,YAAY,iBAAiB;AAEjC,UAAM,gBAAgB,6BAAM;AAC1B,wBAAkB,cAAc;AAAA,IAAA,GADZ;AAItB,UAAM,aAAa,6BAAM;AACvB,kBAAA,EAAc;IAAW,GADR;AAIb,UAAA,YAAY,CAAC,WAAW;AACf,mBAAA,IAAI,2BAA2B,MAAM;AAAA,IAAA,CACnD;AAEK,UAAA,oBAAoB,CAAC,WAAW;AACvB,mBAAA,IAAI,mCAAmC,MAAM;AAAA,IAAA,CAC3D;AAEK,UAAA,EAAE,oBAAoB;AAC5B,UAAM,sBAAsB,mCAAY;AAChC,YAAA,gBAAgB,gBAAgB,KAAK;AAAA,IAAA,GADjB;AAItB,UAAA,EAAE,MAAM;AACR,UAAA,gBAAgB,wBAAC,SAA0C;AACxD,aAAA;AAAA,QACL,GAAG;AAAA,QACH,MAAM,EAAE,qBAAqB,KAAK,EAAE,SAAS,KAAK,IAAI;AAAA,QACtD,SAAS,KAAK,UACV,EAAE,qBAAqB,KAAK,EAAE,YAAY,KAAK,OAAO,IACtD;AAAA,MAAA;AAAA,IACN,GAPoB;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;"}

web/assets/ServerConfigPanel-x68ubY-c.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"ServerConfigPanel-x68ubY-c.js","sources":["../../src/components/dialog/content/setting/ServerConfigPanel.vue"],"sourcesContent":["<template>\n  <PanelTemplate value=\"Server-Config\" class=\"server-config-panel\">\n    <template #header>\n      <div class=\"flex flex-col gap-2\">\n        <Message\n          v-if=\"modifiedConfigs.length > 0\"\n          severity=\"info\"\n          pt:text=\"w-full\"\n        >\n          <p>\n            {{ $t('serverConfig.modifiedConfigs') }}\n          </p>\n          <ul>\n            <li v-for=\"config in modifiedConfigs\" :key=\"config.id\">\n              {{ config.name }}: {{ config.initialValue }} → {{ config.value }}\n            </li>\n          </ul>\n          <div class=\"flex justify-end gap-2\">\n            <Button\n              :label=\"$t('serverConfig.revertChanges')\"\n              @click=\"revertChanges\"\n              outlined\n            />\n            <Button\n              :label=\"$t('serverConfig.restart')\"\n              @click=\"restartApp\"\n              outlined\n              severity=\"danger\"\n            />\n          </div>\n        </Message>\n        <Message v-if=\"commandLineArgs\" severity=\"secondary\" pt:text=\"w-full\">\n          <template #icon>\n            <i-lucide:terminal class=\"text-xl font-bold\" />\n          </template>\n          <div class=\"flex items-center justify-between\">\n            <p>{{ commandLineArgs }}</p>\n            <Button\n              icon=\"pi pi-clipboard\"\n              @click=\"copyCommandLineArgs\"\n              severity=\"secondary\"\n              text\n            />\n          </div>\n        </Message>\n      </div>\n    </template>\n    <div\n      v-for=\"([label, items], i) in Object.entries(serverConfigsByCategory)\"\n      :key=\"label\"\n    >\n      <Divider v-if=\"i > 0\" />\n      <h3>{{ formatCamelCase(label) }}</h3>\n      <div\n        v-for=\"item in items\"\n        :key=\"item.name\"\n        class=\"flex items-center mb-4\"\n      >\n        <FormItem\n          :item=\"item\"\n          v-model:formValue=\"item.value\"\n          :id=\"item.id\"\n          :labelClass=\"{\n            'text-highlight': item.initialValue !== item.value\n          }\"\n        />\n      </div>\n    </div>\n  </PanelTemplate>\n</template>\n\n<script setup lang=\"ts\">\nimport Button from 'primevue/button'\nimport Message from 'primevue/message'\nimport Divider from 'primevue/divider'\nimport FormItem from '@/components/common/FormItem.vue'\nimport PanelTemplate from './PanelTemplate.vue'\nimport { formatCamelCase } from '@/utils/formatUtil'\nimport { useServerConfigStore } from '@/stores/serverConfigStore'\nimport { storeToRefs } from 'pinia'\nimport { electronAPI } from '@/utils/envUtil'\nimport { useSettingStore } from '@/stores/settingStore'\nimport { watch } from 'vue'\nimport { useCopyToClipboard } from '@/hooks/clipboardHooks'\n\nconst settingStore = useSettingStore()\nconst serverConfigStore = useServerConfigStore()\nconst {\n  serverConfigsByCategory,\n  serverConfigValues,\n  launchArgs,\n  commandLineArgs,\n  modifiedConfigs\n} = storeToRefs(serverConfigStore)\n\nconst revertChanges = () => {\n  serverConfigStore.revertChanges()\n}\n\nconst restartApp = () => {\n  electronAPI().restartApp()\n}\n\nwatch(launchArgs, (newVal) => {\n  settingStore.set('Comfy.Server.LaunchArgs', newVal)\n})\n\nwatch(serverConfigValues, (newVal) => {\n  settingStore.set('Comfy.Server.ServerConfigValues', newVal)\n})\n\nconst { copyToClipboard } = useCopyToClipboard()\nconst copyCommandLineArgs = async () => {\n  await copyToClipboard(commandLineArgs.value)\n}\n</script>\n"],"names":[],"mappings":";;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AAqFA,UAAM,eAAe;AACrB,UAAM,oBAAoB;AACpB,UAAA;AAAA,MACJ;AAAA,MACA;AAAA,MACA;AAAA,MACA;AAAA,MACA;AAAA,IAAA,IACE,YAAY,iBAAiB;AAEjC,UAAM,gBAAgB,6BAAM;AAC1B,wBAAkB,cAAc;AAAA,IAAA,GADZ;AAItB,UAAM,aAAa,6BAAM;AACvB,kBAAA,EAAc;IAAW,GADR;AAIb,UAAA,YAAY,CAAC,WAAW;AACf,mBAAA,IAAI,2BAA2B,MAAM;AAAA,IAAA,CACnD;AAEK,UAAA,oBAAoB,CAAC,WAAW;AACvB,mBAAA,IAAI,mCAAmC,MAAM;AAAA,IAAA,CAC3D;AAEK,UAAA,EAAE,oBAAoB;AAC5B,UAAM,sBAAsB,mCAAY;AAChC,YAAA,gBAAgB,gBAAgB,KAAK;AAAA,IAAA,GADjB;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;"}

web/assets/ServerStartView-BHqjjHcl.css

@@ -1,5 +1,5 @@
 
-[data-v-f5429be7] .xterm-helper-textarea {
+[data-v-c0d3157e] .xterm-helper-textarea {
   /* Hide this as it moves all over when uv is running */
   display: none;
 }

web/assets/ServerStartView-BykYRkoj.js

@@ -1,15 +1,15 @@
 var __defProp = Object.defineProperty;
 var __name = (target, value) => __defProp(target, "name", { value, configurable: true });
-import { d as defineComponent, aD as useI18n, r as ref, o as onMounted, g as openBlock, h as createElementBlock, A as createBaseVNode, aw as createTextVNode, a6 as toDisplayString, z as unref, j as createCommentVNode, i as createVNode, D as script, bM as BaseTerminal, P as pushScopeId, Q as popScopeId, bN as electronAPI, _ as _export_sfc } from "./index-CoOvI8ZH.js";
-import { P as ProgressStatus } from "./index-BppSBmxJ.js";
-const _withScopeId = /* @__PURE__ */ __name((n) => (pushScopeId("data-v-f5429be7"), n = n(), popScopeId(), n), "_withScopeId");
+import { a as defineComponent, I as useI18n, r as ref, bX as ProgressStatus, o as onMounted, f as openBlock, g as createElementBlock, A as createBaseVNode, ay as createTextVNode, a8 as toDisplayString, z as unref, i as createCommentVNode, h as createVNode, D as script, x as createBlock, v as withDirectives, ad as vShow, bY as BaseTerminal, R as pushScopeId, U as popScopeId, bZ as electronAPI, _ as _export_sfc } from "./index-CSl7lfOs.js";
+const _withScopeId = /* @__PURE__ */ __name((n) => (pushScopeId("data-v-c0d3157e"), n = n(), popScopeId(), n), "_withScopeId");
 const _hoisted_1 = { class: "font-sans flex flex-col justify-center items-center h-screen m-0 text-neutral-300 bg-neutral-900 dark-theme pointer-events-auto" };
 const _hoisted_2 = { class: "text-2xl font-bold" };
 const _hoisted_3 = { key: 0 };
 const _hoisted_4 = {
   key: 0,
-  class: "flex items-center my-4 gap-2"
+  class: "flex flex-col items-center gap-4"
 };
+const _hoisted_5 = { class: "flex items-center my-4 gap-2" };
 const _sfc_main = /* @__PURE__ */ defineComponent({
   __name: "ServerStartView",
   setup(__props) {
@@ -18,9 +18,11 @@ const _sfc_main = /* @__PURE__ */ defineComponent({
     const status = ref(ProgressStatus.INITIAL_STATE);
     const electronVersion = ref("");
     let xterm;
+    const terminalVisible = ref(true);
     const updateProgress = /* @__PURE__ */ __name(({ status: newStatus }) => {
       status.value = newStatus;
-      xterm?.clear();
+      if (newStatus === ProgressStatus.ERROR) terminalVisible.value = false;
+      else xterm?.clear();
     }, "updateProgress");
     const terminalCreated = /* @__PURE__ */ __name(({ terminal, useAutoSize }, root) => {
       xterm = terminal;
@@ -49,31 +51,42 @@ const _sfc_main = /* @__PURE__ */ defineComponent({
           status.value === unref(ProgressStatus).ERROR ? (openBlock(), createElementBlock("span", _hoisted_3, " v" + toDisplayString(electronVersion.value), 1)) : createCommentVNode("", true)
         ]),
         status.value === unref(ProgressStatus).ERROR ? (openBlock(), createElementBlock("div", _hoisted_4, [
-          createVNode(unref(script), {
-            icon: "pi pi-flag",
+          createBaseVNode("div", _hoisted_5, [
+            createVNode(unref(script), {
+              icon: "pi pi-flag",
+              severity: "secondary",
+              label: unref(t)("serverStart.reportIssue"),
+              onClick: reportIssue
+            }, null, 8, ["label"]),
+            createVNode(unref(script), {
+              icon: "pi pi-file",
+              severity: "secondary",
+              label: unref(t)("serverStart.openLogs"),
+              onClick: openLogs
+            }, null, 8, ["label"]),
+            createVNode(unref(script), {
+              icon: "pi pi-refresh",
+              label: unref(t)("serverStart.reinstall"),
+              onClick: reinstall
+            }, null, 8, ["label"])
+          ]),
+          !terminalVisible.value ? (openBlock(), createBlock(unref(script), {
+            key: 0,
+            icon: "pi pi-search",
             severity: "secondary",
-            label: unref(t)("serverStart.reportIssue"),
-            onClick: reportIssue
-          }, null, 8, ["label"]),
-          createVNode(unref(script), {
-            icon: "pi pi-file",
-            severity: "secondary",
-            label: unref(t)("serverStart.openLogs"),
-            onClick: openLogs
-          }, null, 8, ["label"]),
-          createVNode(unref(script), {
-            icon: "pi pi-refresh",
-            label: unref(t)("serverStart.reinstall"),
-            onClick: reinstall
-          }, null, 8, ["label"])
+            label: unref(t)("serverStart.showTerminal"),
+            onClick: _cache[0] || (_cache[0] = ($event) => terminalVisible.value = true)
+          }, null, 8, ["label"])) : createCommentVNode("", true)
         ])) : createCommentVNode("", true),
-        createVNode(BaseTerminal, { onCreated: terminalCreated })
+        withDirectives(createVNode(BaseTerminal, { onCreated: terminalCreated }, null, 512), [
+          [vShow, terminalVisible.value]
+        ])
       ]);
     };
   }
 });
-const ServerStartView = /* @__PURE__ */ _export_sfc(_sfc_main, [["__scopeId", "data-v-f5429be7"]]);
+const ServerStartView = /* @__PURE__ */ _export_sfc(_sfc_main, [["__scopeId", "data-v-c0d3157e"]]);
 export {
   ServerStartView as default
 };
-//# sourceMappingURL=ServerStartView-CqRVtr1h.js.map
+//# sourceMappingURL=ServerStartView-BykYRkoj.js.map

web/assets/ServerStartView-BykYRkoj.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"ServerStartView-BykYRkoj.js","sources":["../../src/views/ServerStartView.vue"],"sourcesContent":["<template>\n  <div\n    class=\"font-sans flex flex-col justify-center items-center h-screen m-0 text-neutral-300 bg-neutral-900 dark-theme pointer-events-auto\"\n  >\n    <h2 class=\"text-2xl font-bold\">\n      {{ t(`serverStart.process.${status}`) }}\n      <span v-if=\"status === ProgressStatus.ERROR\">\n        v{{ electronVersion }}\n      </span>\n    </h2>\n    <div\n      v-if=\"status === ProgressStatus.ERROR\"\n      class=\"flex flex-col items-center gap-4\"\n    >\n      <div class=\"flex items-center my-4 gap-2\">\n        <Button\n          icon=\"pi pi-flag\"\n          severity=\"secondary\"\n          :label=\"t('serverStart.reportIssue')\"\n          @click=\"reportIssue\"\n        />\n        <Button\n          icon=\"pi pi-file\"\n          severity=\"secondary\"\n          :label=\"t('serverStart.openLogs')\"\n          @click=\"openLogs\"\n        />\n        <Button\n          icon=\"pi pi-refresh\"\n          :label=\"t('serverStart.reinstall')\"\n          @click=\"reinstall\"\n        />\n      </div>\n      <Button\n        v-if=\"!terminalVisible\"\n        icon=\"pi pi-search\"\n        severity=\"secondary\"\n        :label=\"t('serverStart.showTerminal')\"\n        @click=\"terminalVisible = true\"\n      />\n    </div>\n    <BaseTerminal v-show=\"terminalVisible\" @created=\"terminalCreated\" />\n  </div>\n</template>\n\n<script setup lang=\"ts\">\nimport Button from 'primevue/button'\nimport { ref, onMounted, Ref } from 'vue'\nimport BaseTerminal from '@/components/bottomPanel/tabs/terminal/BaseTerminal.vue'\nimport { ProgressStatus } from '@comfyorg/comfyui-electron-types'\nimport { electronAPI } from '@/utils/envUtil'\nimport type { useTerminal } from '@/hooks/bottomPanelTabs/useTerminal'\nimport { Terminal } from '@xterm/xterm'\nimport { useI18n } from 'vue-i18n'\n\nconst electron = electronAPI()\nconst { t } = useI18n()\n\nconst status = ref<ProgressStatus>(ProgressStatus.INITIAL_STATE)\nconst electronVersion = ref<string>('')\nlet xterm: Terminal | undefined\n\nconst terminalVisible = ref(true)\n\nconst updateProgress = ({ status: newStatus }: { status: ProgressStatus }) => {\n  status.value = newStatus\n\n  // Make critical error screen more obvious.\n  if (newStatus === ProgressStatus.ERROR) terminalVisible.value = false\n  else xterm?.clear()\n}\n\nconst terminalCreated = (\n  { terminal, useAutoSize }: ReturnType<typeof useTerminal>,\n  root: Ref<HTMLElement>\n) => {\n  xterm = terminal\n\n  useAutoSize(root, true, true)\n  electron.onLogMessage((message: string) => {\n    terminal.write(message)\n  })\n\n  terminal.options.cursorBlink = false\n  terminal.options.disableStdin = true\n  terminal.options.cursorInactiveStyle = 'block'\n}\n\nconst reinstall = () => electron.reinstall()\nconst reportIssue = () => {\n  window.open('https://forum.comfy.org/c/v1-feedback/', '_blank')\n}\nconst openLogs = () => electron.openLogsFolder()\n\nonMounted(async () => {\n  electron.sendReady()\n  electron.onProgressUpdate(updateProgress)\n  electronVersion.value = await electron.getElectronVersion()\n})\n</script>\n\n<style scoped>\n:deep(.xterm-helper-textarea) {\n  /* Hide this as it moves all over when uv is running */\n  display: none;\n}\n</style>\n"],"names":[],"mappings":";;;;;;;;;;;;;;;AAuDA,UAAM,WAAW;AACX,UAAA,EAAE,MAAM;AAER,UAAA,SAAS,IAAoB,eAAe,aAAa;AACzD,UAAA,kBAAkB,IAAY,EAAE;AAClC,QAAA;AAEE,UAAA,kBAAkB,IAAI,IAAI;AAEhC,UAAM,iBAAiB,wBAAC,EAAE,QAAQ,gBAA4C;AAC5E,aAAO,QAAQ;AAGf,UAAI,cAAc,eAAe,MAAO,iBAAgB,QAAQ;AAAA,kBACpD,MAAM;AAAA,IAAA,GALG;AAQvB,UAAM,kBAAkB,wBACtB,EAAE,UAAU,YAAA,GACZ,SACG;AACK,cAAA;AAEI,kBAAA,MAAM,MAAM,IAAI;AACnB,eAAA,aAAa,CAAC,YAAoB;AACzC,iBAAS,MAAM,OAAO;AAAA,MAAA,CACvB;AAED,eAAS,QAAQ,cAAc;AAC/B,eAAS,QAAQ,eAAe;AAChC,eAAS,QAAQ,sBAAsB;AAAA,IAAA,GAbjB;AAgBlB,UAAA,YAAY,6BAAM,SAAS,aAAf;AAClB,UAAM,cAAc,6BAAM;AACjB,aAAA,KAAK,0CAA0C,QAAQ;AAAA,IAAA,GAD5C;AAGd,UAAA,WAAW,6BAAM,SAAS,kBAAf;AAEjB,cAAU,YAAY;AACpB,eAAS,UAAU;AACnB,eAAS,iBAAiB,cAAc;AACxB,sBAAA,QAAQ,MAAM,SAAS,mBAAmB;AAAA,IAAA,CAC3D;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;"}

web/assets/ServerStartView-CqRVtr1h.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"ServerStartView-CqRVtr1h.js","sources":["../../src/views/ServerStartView.vue"],"sourcesContent":["<template>\n  <div\n    class=\"font-sans flex flex-col justify-center items-center h-screen m-0 text-neutral-300 bg-neutral-900 dark-theme pointer-events-auto\"\n  >\n    <h2 class=\"text-2xl font-bold\">\n      {{ t(`serverStart.process.${status}`) }}\n      <span v-if=\"status === ProgressStatus.ERROR\">\n        v{{ electronVersion }}\n      </span>\n    </h2>\n    <div\n      v-if=\"status === ProgressStatus.ERROR\"\n      class=\"flex items-center my-4 gap-2\"\n    >\n      <Button\n        icon=\"pi pi-flag\"\n        severity=\"secondary\"\n        :label=\"t('serverStart.reportIssue')\"\n        @click=\"reportIssue\"\n      />\n      <Button\n        icon=\"pi pi-file\"\n        severity=\"secondary\"\n        :label=\"t('serverStart.openLogs')\"\n        @click=\"openLogs\"\n      />\n      <Button\n        icon=\"pi pi-refresh\"\n        :label=\"t('serverStart.reinstall')\"\n        @click=\"reinstall\"\n      />\n    </div>\n    <BaseTerminal @created=\"terminalCreated\" />\n  </div>\n</template>\n\n<script setup lang=\"ts\">\nimport Button from 'primevue/button'\nimport { ref, onMounted, Ref } from 'vue'\nimport BaseTerminal from '@/components/bottomPanel/tabs/terminal/BaseTerminal.vue'\nimport { ProgressStatus } from '@comfyorg/comfyui-electron-types'\nimport { electronAPI } from '@/utils/envUtil'\nimport type { useTerminal } from '@/hooks/bottomPanelTabs/useTerminal'\nimport { Terminal } from '@xterm/xterm'\nimport { useI18n } from 'vue-i18n'\n\nconst electron = electronAPI()\nconst { t } = useI18n()\n\nconst status = ref<ProgressStatus>(ProgressStatus.INITIAL_STATE)\nconst electronVersion = ref<string>('')\nlet xterm: Terminal | undefined\n\nconst updateProgress = ({ status: newStatus }: { status: ProgressStatus }) => {\n  status.value = newStatus\n  xterm?.clear()\n}\n\nconst terminalCreated = (\n  { terminal, useAutoSize }: ReturnType<typeof useTerminal>,\n  root: Ref<HTMLElement>\n) => {\n  xterm = terminal\n\n  useAutoSize(root, true, true)\n  electron.onLogMessage((message: string) => {\n    terminal.write(message)\n  })\n\n  terminal.options.cursorBlink = false\n  terminal.options.disableStdin = true\n  terminal.options.cursorInactiveStyle = 'block'\n}\n\nconst reinstall = () => electron.reinstall()\nconst reportIssue = () => {\n  window.open('https://forum.comfy.org/c/v1-feedback/', '_blank')\n}\nconst openLogs = () => electron.openLogsFolder()\n\nonMounted(async () => {\n  electron.sendReady()\n  electron.onProgressUpdate(updateProgress)\n  electronVersion.value = await electron.getElectronVersion()\n})\n</script>\n\n<style scoped>\n:deep(.xterm-helper-textarea) {\n  /* Hide this as it moves all over when uv is running */\n  display: none;\n}\n</style>\n"],"names":[],"mappings":";;;;;;;;;;;;;;;AA8CA,UAAM,WAAW;AACX,UAAA,EAAE,MAAM;AAER,UAAA,SAAS,IAAoB,eAAe,aAAa;AACzD,UAAA,kBAAkB,IAAY,EAAE;AAClC,QAAA;AAEJ,UAAM,iBAAiB,wBAAC,EAAE,QAAQ,gBAA4C;AAC5E,aAAO,QAAQ;AACf,aAAO,MAAM;AAAA,IAAA,GAFQ;AAKvB,UAAM,kBAAkB,wBACtB,EAAE,UAAU,YAAA,GACZ,SACG;AACK,cAAA;AAEI,kBAAA,MAAM,MAAM,IAAI;AACnB,eAAA,aAAa,CAAC,YAAoB;AACzC,iBAAS,MAAM,OAAO;AAAA,MAAA,CACvB;AAED,eAAS,QAAQ,cAAc;AAC/B,eAAS,QAAQ,eAAe;AAChC,eAAS,QAAQ,sBAAsB;AAAA,IAAA,GAbjB;AAgBlB,UAAA,YAAY,6BAAM,SAAS,aAAf;AAClB,UAAM,cAAc,6BAAM;AACjB,aAAA,KAAK,0CAA0C,QAAQ;AAAA,IAAA,GAD5C;AAGd,UAAA,WAAW,6BAAM,SAAS,kBAAf;AAEjB,cAAU,YAAY;AACpB,eAAS,UAAU;AACnB,eAAS,iBAAiB,cAAc;AACxB,sBAAA,QAAQ,MAAM,SAAS,mBAAmB;AAAA,IAAA,CAC3D;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;"}

web/assets/UserSelectView-DMDUPUPX.js

@@ -0,0 +1,98 @@
+var __defProp = Object.defineProperty;
+var __name = (target, value) => __defProp(target, "name", { value, configurable: true });
+import { a as defineComponent, J as useUserStore, bU as useRouter, r as ref, q as computed, o as onMounted, f as openBlock, g as createElementBlock, A as createBaseVNode, a8 as toDisplayString, h as createVNode, z as unref, aq as script, bN as script$1, bV as script$2, x as createBlock, y as withCtx, ay as createTextVNode, bW as script$3, i as createCommentVNode, D as script$4 } from "./index-CSl7lfOs.js";
+const _hoisted_1 = {
+  id: "comfy-user-selection",
+  class: "font-sans flex flex-col items-center h-screen m-0 text-neutral-300 bg-neutral-900 dark-theme pointer-events-auto"
+};
+const _hoisted_2 = { class: "mt-[5vh] 2xl:mt-[20vh] min-w-84 relative rounded-lg bg-[var(--comfy-menu-bg)] p-5 px-10 shadow-lg" };
+const _hoisted_3 = /* @__PURE__ */ createBaseVNode("h1", { class: "my-2.5 mb-7 font-normal" }, "ComfyUI", -1);
+const _hoisted_4 = { class: "flex w-full flex-col items-center" };
+const _hoisted_5 = { class: "flex w-full flex-col gap-2" };
+const _hoisted_6 = { for: "new-user-input" };
+const _hoisted_7 = { class: "flex w-full flex-col gap-2" };
+const _hoisted_8 = { for: "existing-user-select" };
+const _hoisted_9 = { class: "mt-5" };
+const _sfc_main = /* @__PURE__ */ defineComponent({
+  __name: "UserSelectView",
+  setup(__props) {
+    const userStore = useUserStore();
+    const router = useRouter();
+    const selectedUser = ref(null);
+    const newUsername = ref("");
+    const loginError = ref("");
+    const createNewUser = computed(() => newUsername.value.trim() !== "");
+    const newUserExistsError = computed(() => {
+      return userStore.users.find((user) => user.username === newUsername.value) ? `User "${newUsername.value}" already exists` : "";
+    });
+    const error = computed(() => newUserExistsError.value || loginError.value);
+    const login = /* @__PURE__ */ __name(async () => {
+      try {
+        const user = createNewUser.value ? await userStore.createUser(newUsername.value) : selectedUser.value;
+        if (!user) {
+          throw new Error("No user selected");
+        }
+        userStore.login(user);
+        router.push("/");
+      } catch (err) {
+        loginError.value = err.message ?? JSON.stringify(err);
+      }
+    }, "login");
+    onMounted(async () => {
+      if (!userStore.initialized) {
+        await userStore.initialize();
+      }
+    });
+    return (_ctx, _cache) => {
+      return openBlock(), createElementBlock("div", _hoisted_1, [
+        createBaseVNode("main", _hoisted_2, [
+          _hoisted_3,
+          createBaseVNode("form", _hoisted_4, [
+            createBaseVNode("div", _hoisted_5, [
+              createBaseVNode("label", _hoisted_6, toDisplayString(_ctx.$t("userSelect.newUser")) + ":", 1),
+              createVNode(unref(script), {
+                id: "new-user-input",
+                modelValue: newUsername.value,
+                "onUpdate:modelValue": _cache[0] || (_cache[0] = ($event) => newUsername.value = $event),
+                placeholder: _ctx.$t("userSelect.enterUsername")
+              }, null, 8, ["modelValue", "placeholder"])
+            ]),
+            createVNode(unref(script$1)),
+            createBaseVNode("div", _hoisted_7, [
+              createBaseVNode("label", _hoisted_8, toDisplayString(_ctx.$t("userSelect.existingUser")) + ":", 1),
+              createVNode(unref(script$2), {
+                modelValue: selectedUser.value,
+                "onUpdate:modelValue": _cache[1] || (_cache[1] = ($event) => selectedUser.value = $event),
+                class: "w-full",
+                inputId: "existing-user-select",
+                options: unref(userStore).users,
+                "option-label": "username",
+                placeholder: _ctx.$t("userSelect.selectUser"),
+                disabled: createNewUser.value
+              }, null, 8, ["modelValue", "options", "placeholder", "disabled"]),
+              error.value ? (openBlock(), createBlock(unref(script$3), {
+                key: 0,
+                severity: "error"
+              }, {
+                default: withCtx(() => [
+                  createTextVNode(toDisplayString(error.value), 1)
+                ]),
+                _: 1
+              })) : createCommentVNode("", true)
+            ]),
+            createBaseVNode("footer", _hoisted_9, [
+              createVNode(unref(script$4), {
+                label: _ctx.$t("userSelect.next"),
+                onClick: login
+              }, null, 8, ["label"])
+            ])
+          ])
+        ])
+      ]);
+    };
+  }
+});
+export {
+  _sfc_main as default
+};
+//# sourceMappingURL=UserSelectView-DMDUPUPX.js.map

web/assets/UserSelectView-DMDUPUPX.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"UserSelectView-DMDUPUPX.js","sources":["../../src/views/UserSelectView.vue"],"sourcesContent":["<template>\n  <div\n    id=\"comfy-user-selection\"\n    class=\"font-sans flex flex-col items-center h-screen m-0 text-neutral-300 bg-neutral-900 dark-theme pointer-events-auto\"\n  >\n    <main\n      class=\"mt-[5vh] 2xl:mt-[20vh] min-w-84 relative rounded-lg bg-[var(--comfy-menu-bg)] p-5 px-10 shadow-lg\"\n    >\n      <h1 class=\"my-2.5 mb-7 font-normal\">ComfyUI</h1>\n      <form class=\"flex w-full flex-col items-center\">\n        <div class=\"flex w-full flex-col gap-2\">\n          <label for=\"new-user-input\">{{ $t('userSelect.newUser') }}:</label>\n          <InputText\n            id=\"new-user-input\"\n            v-model=\"newUsername\"\n            :placeholder=\"$t('userSelect.enterUsername')\"\n          />\n        </div>\n        <Divider />\n        <div class=\"flex w-full flex-col gap-2\">\n          <label for=\"existing-user-select\"\n            >{{ $t('userSelect.existingUser') }}:</label\n          >\n          <Select\n            v-model=\"selectedUser\"\n            class=\"w-full\"\n            inputId=\"existing-user-select\"\n            :options=\"userStore.users\"\n            option-label=\"username\"\n            :placeholder=\"$t('userSelect.selectUser')\"\n            :disabled=\"createNewUser\"\n          />\n          <Message v-if=\"error\" severity=\"error\">{{ error }}</Message>\n        </div>\n        <footer class=\"mt-5\">\n          <Button :label=\"$t('userSelect.next')\" @click=\"login\" />\n        </footer>\n      </form>\n    </main>\n  </div>\n</template>\n\n<script setup lang=\"ts\">\nimport Button from 'primevue/button'\nimport Divider from 'primevue/divider'\nimport InputText from 'primevue/inputtext'\nimport Select from 'primevue/select'\nimport Message from 'primevue/message'\nimport { User, useUserStore } from '@/stores/userStore'\nimport { useRouter } from 'vue-router'\nimport { computed, onMounted, ref } from 'vue'\n\nconst userStore = useUserStore()\nconst router = useRouter()\n\nconst selectedUser = ref<User | null>(null)\nconst newUsername = ref('')\nconst loginError = ref('')\n\nconst createNewUser = computed(() => newUsername.value.trim() !== '')\nconst newUserExistsError = computed(() => {\n  return userStore.users.find((user) => user.username === newUsername.value)\n    ? `User \"${newUsername.value}\" already exists`\n    : ''\n})\nconst error = computed(() => newUserExistsError.value || loginError.value)\n\nconst login = async () => {\n  try {\n    const user = createNewUser.value\n      ? await userStore.createUser(newUsername.value)\n      : selectedUser.value\n\n    if (!user) {\n      throw new Error('No user selected')\n    }\n\n    userStore.login(user)\n    router.push('/')\n  } catch (err) {\n    loginError.value = err.message ?? JSON.stringify(err)\n  }\n}\n\nonMounted(async () => {\n  if (!userStore.initialized) {\n    await userStore.initialize()\n  }\n})\n</script>\n"],"names":[],"mappings":";;;;;;;;;;;;;;;;;;AAoDA,UAAM,YAAY;AAClB,UAAM,SAAS;AAET,UAAA,eAAe,IAAiB,IAAI;AACpC,UAAA,cAAc,IAAI,EAAE;AACpB,UAAA,aAAa,IAAI,EAAE;AAEzB,UAAM,gBAAgB,SAAS,MAAM,YAAY,MAAM,KAAA,MAAW,EAAE;AAC9D,UAAA,qBAAqB,SAAS,MAAM;AACxC,aAAO,UAAU,MAAM,KAAK,CAAC,SAAS,KAAK,aAAa,YAAY,KAAK,IACrE,SAAS,YAAY,KAAK,qBAC1B;AAAA,IAAA,CACL;AACD,UAAM,QAAQ,SAAS,MAAM,mBAAmB,SAAS,WAAW,KAAK;AAEzE,UAAM,QAAQ,mCAAY;AACpB,UAAA;AACI,cAAA,OAAO,cAAc,QACvB,MAAM,UAAU,WAAW,YAAY,KAAK,IAC5C,aAAa;AAEjB,YAAI,CAAC,MAAM;AACH,gBAAA,IAAI,MAAM,kBAAkB;AAAA,QACpC;AAEA,kBAAU,MAAM,IAAI;AACpB,eAAO,KAAK,GAAG;AAAA,eACR,KAAK;AACZ,mBAAW,QAAQ,IAAI,WAAW,KAAK,UAAU,GAAG;AAAA,MACtD;AAAA,IAAA,GAdY;AAiBd,cAAU,YAAY;AAChB,UAAA,CAAC,UAAU,aAAa;AAC1B,cAAM,UAAU;MAClB;AAAA,IAAA,CACD;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;"}

web/assets/WelcomeView-BD34JMsC.css

@@ -1,5 +1,5 @@
 
-.animated-gradient-text[data-v-12b8b11b] {
+.animated-gradient-text[data-v-c4d014c5] {
   font-weight: 700;
   font-size: clamp(2rem, 8vw, 4rem);
   background: linear-gradient(to right, #12c2e9, #c471ed, #f64f59, #12c2e9);
@@ -7,12 +7,12 @@
   background-clip: text;
   -webkit-background-clip: text;
   -webkit-text-fill-color: transparent;
-  animation: gradient-12b8b11b 8s linear infinite;
+  animation: gradient-c4d014c5 8s linear infinite;
 }
-.text-glow[data-v-12b8b11b] {
+.text-glow[data-v-c4d014c5] {
   filter: drop-shadow(0 0 8px rgba(255, 255, 255, 0.3));
 }
-@keyframes gradient-12b8b11b {
+@keyframes gradient-c4d014c5 {
 0% {
     background-position: 0% center;
 }
@@ -20,11 +20,11 @@
     background-position: 300% center;
 }
 }
-.fade-in-up[data-v-12b8b11b] {
-  animation: fadeInUp-12b8b11b 1.5s ease-out;
+.fade-in-up[data-v-c4d014c5] {
+  animation: fadeInUp-c4d014c5 1.5s ease-out;
   animation-fill-mode: both;
 }
-@keyframes fadeInUp-12b8b11b {
+@keyframes fadeInUp-c4d014c5 {
 0% {
     opacity: 0;
     transform: translateY(20px);

web/assets/WelcomeView-C4D1cggT.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"WelcomeView-C4D1cggT.js","sources":[],"sourcesContent":[],"names":[],"mappings":";;;;;;;;;;;;;;;;;;;;;;;;;;;;;"}

web/assets/WelcomeView-D6WEsVyp.js

@@ -1,13 +1,17 @@
 var __defProp = Object.defineProperty;
 var __name = (target, value) => __defProp(target, "name", { value, configurable: true });
-import { d as defineComponent, g as openBlock, h as createElementBlock, A as createBaseVNode, a6 as toDisplayString, i as createVNode, z as unref, D as script, P as pushScopeId, Q as popScopeId, _ as _export_sfc } from "./index-CoOvI8ZH.js";
-const _withScopeId = /* @__PURE__ */ __name((n) => (pushScopeId("data-v-12b8b11b"), n = n(), popScopeId(), n), "_withScopeId");
+import { a as defineComponent, bU as useRouter, f as openBlock, g as createElementBlock, A as createBaseVNode, a8 as toDisplayString, h as createVNode, z as unref, D as script, R as pushScopeId, U as popScopeId, _ as _export_sfc } from "./index-CSl7lfOs.js";
+const _withScopeId = /* @__PURE__ */ __name((n) => (pushScopeId("data-v-c4d014c5"), n = n(), popScopeId(), n), "_withScopeId");
 const _hoisted_1 = { class: "font-sans flex flex-col justify-center items-center h-screen m-0 text-neutral-300 bg-neutral-900 dark-theme pointer-events-auto" };
 const _hoisted_2 = { class: "flex flex-col items-center justify-center gap-8 p-8" };
 const _hoisted_3 = { class: "animated-gradient-text text-glow select-none" };
 const _sfc_main = /* @__PURE__ */ defineComponent({
   __name: "WelcomeView",
   setup(__props) {
+    const router = useRouter();
+    const navigateTo = /* @__PURE__ */ __name((path) => {
+      router.push(path);
+    }, "navigateTo");
     return (_ctx, _cache) => {
       return openBlock(), createElementBlock("div", _hoisted_1, [
         createBaseVNode("div", _hoisted_2, [
@@ -18,7 +22,7 @@ const _sfc_main = /* @__PURE__ */ defineComponent({
             iconPos: "right",
             size: "large",
             rounded: "",
-            onClick: _cache[0] || (_cache[0] = ($event) => _ctx.$router.push("/install")),
+            onClick: _cache[0] || (_cache[0] = ($event) => navigateTo("/install")),
             class: "p-4 text-lg fade-in-up"
           }, null, 8, ["label"])
         ])
@@ -26,8 +30,8 @@ const _sfc_main = /* @__PURE__ */ defineComponent({
     };
   }
 });
-const WelcomeView = /* @__PURE__ */ _export_sfc(_sfc_main, [["__scopeId", "data-v-12b8b11b"]]);
+const WelcomeView = /* @__PURE__ */ _export_sfc(_sfc_main, [["__scopeId", "data-v-c4d014c5"]]);
 export {
   WelcomeView as default
 };
-//# sourceMappingURL=WelcomeView-C4D1cggT.js.map
+//# sourceMappingURL=WelcomeView-D6WEsVyp.js.map

web/assets/WelcomeView-D6WEsVyp.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"WelcomeView-D6WEsVyp.js","sources":["../../src/views/WelcomeView.vue"],"sourcesContent":["<template>\n  <div\n    class=\"font-sans flex flex-col justify-center items-center h-screen m-0 text-neutral-300 bg-neutral-900 dark-theme pointer-events-auto\"\n  >\n    <div class=\"flex flex-col items-center justify-center gap-8 p-8\">\n      <!-- Header -->\n      <h1 class=\"animated-gradient-text text-glow select-none\">\n        {{ $t('welcome.title') }}\n      </h1>\n\n      <!-- Get Started Button -->\n      <Button\n        :label=\"$t('welcome.getStarted')\"\n        icon=\"pi pi-arrow-right\"\n        iconPos=\"right\"\n        size=\"large\"\n        rounded\n        @click=\"navigateTo('/install')\"\n        class=\"p-4 text-lg fade-in-up\"\n      />\n    </div>\n  </div>\n</template>\n\n<script setup lang=\"ts\">\nimport Button from 'primevue/button'\nimport { useRouter } from 'vue-router'\n\nconst router = useRouter()\nconst navigateTo = (path: string) => {\n  router.push(path)\n}\n</script>\n\n<style scoped>\n.animated-gradient-text {\n  @apply font-bold;\n  font-size: clamp(2rem, 8vw, 4rem);\n  background: linear-gradient(to right, #12c2e9, #c471ed, #f64f59, #12c2e9);\n  background-size: 300% auto;\n  background-clip: text;\n  -webkit-background-clip: text;\n  -webkit-text-fill-color: transparent;\n  animation: gradient 8s linear infinite;\n}\n\n.text-glow {\n  filter: drop-shadow(0 0 8px rgba(255, 255, 255, 0.3));\n}\n\n@keyframes gradient {\n  0% {\n    background-position: 0% center;\n  }\n\n  100% {\n    background-position: 300% center;\n  }\n}\n\n.fade-in-up {\n  animation: fadeInUp 1.5s ease-out;\n  animation-fill-mode: both;\n}\n\n@keyframes fadeInUp {\n  0% {\n    opacity: 0;\n    transform: translateY(20px);\n  }\n\n  100% {\n    opacity: 1;\n    transform: translateY(0);\n  }\n}\n</style>\n"],"names":[],"mappings":";;;;;;;;;;AA4BA,UAAM,SAAS;AACT,UAAA,aAAa,wBAAC,SAAiB;AACnC,aAAO,KAAK,IAAI;AAAA,IAAA,GADC;;;;;;;;;;;;;;;;;;;;"}

web/assets/images/Git-Logo-White.svg

@@ -0,0 +1 @@
+<svg xmlns="http://www.w3.org/2000/svg" width="292" height="92pt" viewBox="0 0 219 92"><defs><clipPath id="a"><path d="M159 .79h25V69h-25Zm0 0"/></clipPath><clipPath id="b"><path d="M183 9h35.371v60H183Zm0 0"/></clipPath><clipPath id="c"><path d="M0 .79h92V92H0Zm0 0"/></clipPath></defs><path style="stroke:none;fill-rule:nonzero;fill:#fff;fill-opacity:1" d="M130.871 31.836c-4.785 0-8.351 2.352-8.351 8.008 0 4.261 2.347 7.222 8.093 7.222 4.871 0 8.18-2.867 8.18-7.398 0-5.133-2.961-7.832-7.922-7.832Zm-9.57 39.95c-1.133 1.39-2.262 2.87-2.262 4.612 0 3.48 4.434 4.524 10.527 4.524 5.051 0 11.926-.352 11.926-5.043 0-2.793-3.308-2.965-7.488-3.227Zm25.761-39.688c1.563 2.004 3.22 4.789 3.22 8.793 0 9.656-7.571 15.316-18.536 15.316-2.789 0-5.312-.348-6.879-.785l-2.87 4.613 8.526.52c15.059.96 23.934 1.398 23.934 12.968 0 10.008-8.789 15.665-23.934 15.665-15.75 0-21.757-4.004-21.757-10.88 0-3.917 1.742-6 4.789-8.878-2.875-1.211-3.828-3.387-3.828-5.739 0-1.914.953-3.656 2.523-5.312 1.566-1.652 3.305-3.305 5.395-5.219-4.262-2.09-7.485-6.617-7.485-13.058 0-10.008 6.613-16.88 19.93-16.88 3.742 0 6.004.344 8.008.872h16.972v7.394l-8.007.61"/><g clip-path="url(#a)"><path style="stroke:none;fill-rule:nonzero;fill:#fff;fill-opacity:1" d="M170.379 16.281c-4.961 0-7.832-2.87-7.832-7.836 0-4.957 2.871-7.656 7.832-7.656 5.05 0 7.922 2.7 7.922 7.656 0 4.965-2.871 7.836-7.922 7.836Zm-11.227 52.305V61.71l4.438-.606c1.219-.175 1.394-.437 1.394-1.746V33.773c0-.953-.261-1.566-1.132-1.824l-4.7-1.656.957-7.047h18.016V59.36c0 1.399.086 1.57 1.395 1.746l4.437.606v6.875h-24.805"/></g><g clip-path="url(#b)"><path style="stroke:none;fill-rule:nonzero;fill:#fff;fill-opacity:1" d="M218.371 65.21c-3.742 1.825-9.223 3.481-14.187 3.481-10.356 0-14.27-4.175-14.27-14.015V31.879c0-.524 0-.871-.7-.871h-6.093v-7.746c7.664-.871 10.707-4.703 11.664-14.188h8.27v12.36c0 .609 0 .87.695.87h12.27v8.704h-12.965v20.797c0 5.136 1.218 7.136 5.918 7.136 2.437 0 4.96-.609 7.047-1.39l2.351 7.66"/></g><g clip-path="url(#c)"><path style="stroke:none;fill-rule:nonzero;fill:#fff;fill-opacity:1" d="M89.422 42.371 49.629 2.582a5.868 5.868 0 0 0-8.3 0l-8.263 8.262 10.48 10.484a6.965 6.965 0 0 1 7.173 1.668 6.98 6.98 0 0 1 1.656 7.215l10.102 10.105a6.963 6.963 0 0 1 7.214 1.657 6.976 6.976 0 0 1 0 9.875 6.98 6.98 0 0 1-9.879 0 6.987 6.987 0 0 1-1.519-7.594l-9.422-9.422v24.793a6.979 6.979 0 0 1 1.848 1.32 6.988 6.988 0 0 1 0 9.88c-2.73 2.726-7.153 2.726-9.875 0a6.98 6.98 0 0 1 0-9.88 6.893 6.893 0 0 1 2.285-1.523V34.398a6.893 6.893 0 0 1-2.285-1.523 6.988 6.988 0 0 1-1.508-7.637L29.004 14.902 1.719 42.187a5.868 5.868 0 0 0 0 8.301l39.793 39.793a5.868 5.868 0 0 0 8.3 0l39.61-39.605a5.873 5.873 0 0 0 0-8.305"/></g></svg>

web/assets/index-1vLlIVor.css

@@ -68,26 +68,26 @@
     background-color: rgb(234 179 8 / var(--tw-bg-opacity))
 }
 
-.search-box-input[data-v-f28148d1] {
+.search-box-input[data-v-e10998c1] {
   width: 100%;
   padding-left: 36px;
 }
-.search-box-input.with-filter[data-v-f28148d1] {
+.search-box-input.with-filter[data-v-e10998c1] {
   padding-right: 36px;
 }
-.p-button.p-inputicon[data-v-f28148d1] {
+.p-button.p-inputicon[data-v-e10998c1] {
   padding: 0;
   width: auto;
   border: none !important;
 }
 
-.form-input[data-v-4fbf09d8] .input-slider .p-inputnumber input,
-.form-input[data-v-4fbf09d8] .input-slider .slider-part {
+.form-input[data-v-e54b447b] .input-slider .p-inputnumber input,
+.form-input[data-v-e54b447b] .input-slider .slider-part {
 
     width: 5rem
 }
-.form-input[data-v-4fbf09d8] .p-inputtext,
-.form-input[data-v-4fbf09d8] .p-select {
+.form-input[data-v-e54b447b] .p-inputtext,
+.form-input[data-v-e54b447b] .p-select {
 
     width: 11rem
 }
@@ -333,10 +333,10 @@
   overflow-y: hidden;
 }
 
-[data-v-e8581ca7] .p-terminal .xterm {
+[data-v-36dec989] .p-terminal .xterm {
   overflow-x: auto;
 }
-[data-v-e8581ca7] .p-terminal .xterm-screen {
+[data-v-36dec989] .p-terminal .xterm-screen {
   background-color: black;
   overflow-y: hidden;
 }
@@ -345,7 +345,7 @@
   padding-top: 0px !important;
 }
 
-.settings-container[data-v-dbb35a0c] {
+.settings-container[data-v-d85d6e64] {
   display: flex;
   height: 70vh;
   width: 60vw;
@@ -353,21 +353,24 @@
   overflow: hidden;
 }
 @media (max-width: 768px) {
-.settings-container[data-v-dbb35a0c] {
+.settings-container[data-v-d85d6e64] {
     flex-direction: column;
     height: auto;
 }
-.settings-sidebar[data-v-dbb35a0c] {
+.settings-sidebar[data-v-d85d6e64] {
     width: 100%;
 }
+.settings-content[data-v-d85d6e64] {
+    height: 350px;
+}
 }
 
 /* Show a separator line above the Keybinding tab */
 /* This indicates the start of custom setting panels */
-.settings-sidebar[data-v-dbb35a0c] .p-listbox-option[aria-label='Keybinding'] {
+.settings-sidebar[data-v-d85d6e64] .p-listbox-option[aria-label='Keybinding'] {
   position: relative;
 }
-.settings-sidebar[data-v-dbb35a0c] .p-listbox-option[aria-label='Keybinding']::before {
+.settings-sidebar[data-v-d85d6e64] .p-listbox-option[aria-label='Keybinding']::before {
   position: absolute;
   top: 0px;
   left: 0px;
@@ -377,25 +380,25 @@
   border-top: 1px solid var(--p-divider-border-color);
 }
 
-.pi-cog[data-v-f3b37ea3] {
+.pi-cog[data-v-43089afc] {
   font-size: 1.25rem;
   margin-right: 0.5rem;
 }
-.version-tag[data-v-f3b37ea3] {
+.version-tag[data-v-43089afc] {
   margin-left: 0.5rem;
 }
 
-.comfy-error-report[data-v-db438f98] {
+.comfy-error-report[data-v-5c200f18] {
   display: flex;
   flex-direction: column;
   gap: 1rem;
 }
-.action-container[data-v-db438f98] {
+.action-container[data-v-5c200f18] {
   display: flex;
   gap: 1rem;
   justify-content: flex-end;
 }
-.wrapper-pre[data-v-db438f98] {
+.wrapper-pre[data-v-5c200f18] {
   white-space: pre-wrap;
   word-wrap: break-word;
 }
@@ -407,6 +410,10 @@
 [data-v-98830966] .p-card-subtitle {
   text-align: center;
 }
+
+.prompt-dialog-content[data-v-abaaab2c] {
+  white-space: pre-wrap;
+}
 .mdi.rotate270::before {
   transform: rotate(270deg);
 }
@@ -736,7 +743,7 @@
   word-break: break-all;
 }
 
-[data-v-b7c3d32e] .tree-explorer-node-label {
+[data-v-82fb18cb] .tree-explorer-node-label {
   width: 100%;
   display: flex;
   align-items: center;
@@ -749,10 +756,10 @@
  * By setting the position to relative on the parent and using an absolutely positioned pseudo-element,
  * we can create a visual indicator for the drop target without affecting the layout of other elements.
  */
-[data-v-b7c3d32e] .p-tree-node-content:has(.tree-folder) {
+[data-v-82fb18cb] .p-tree-node-content:has(.tree-folder) {
   position: relative;
 }
-[data-v-b7c3d32e] .p-tree-node-content:has(.tree-folder.can-drop)::after {
+[data-v-82fb18cb] .p-tree-node-content:has(.tree-folder.can-drop)::after {
   content: '';
   position: absolute;
   top: 0;
@@ -842,23 +849,23 @@
   vertical-align: top;
 }
 
-[data-v-827f7782] .pi-fake-spacer {
+[data-v-31a92a0f] .pi-fake-spacer {
   height: 1px;
   width: 16px;
 }
 
-.slot_row[data-v-4b126b34] {
+.slot_row[data-v-e86c3783] {
   padding: 2px;
 }
 
 /* Original N-Sidebar styles */
-._sb_dot[data-v-4b126b34] {
+._sb_dot[data-v-e86c3783] {
   width: 8px;
   height: 8px;
   border-radius: 50%;
   background-color: grey;
 }
-.node_header[data-v-4b126b34] {
+.node_header[data-v-e86c3783] {
   line-height: 1;
   padding: 8px 13px 7px;
   margin-bottom: 5px;
@@ -868,37 +875,37 @@
   display: flex;
   align-items: center;
 }
-.headdot[data-v-4b126b34] {
+.headdot[data-v-e86c3783] {
   width: 10px;
   height: 10px;
   float: inline-start;
   margin-right: 8px;
 }
-.IMAGE[data-v-4b126b34] {
+.IMAGE[data-v-e86c3783] {
   background-color: #64b5f6;
 }
-.VAE[data-v-4b126b34] {
+.VAE[data-v-e86c3783] {
   background-color: #ff6e6e;
 }
-.LATENT[data-v-4b126b34] {
+.LATENT[data-v-e86c3783] {
   background-color: #ff9cf9;
 }
-.MASK[data-v-4b126b34] {
+.MASK[data-v-e86c3783] {
   background-color: #81c784;
 }
-.CONDITIONING[data-v-4b126b34] {
+.CONDITIONING[data-v-e86c3783] {
   background-color: #ffa931;
 }
-.CLIP[data-v-4b126b34] {
+.CLIP[data-v-e86c3783] {
   background-color: #ffd500;
 }
-.MODEL[data-v-4b126b34] {
+.MODEL[data-v-e86c3783] {
   background-color: #b39ddb;
 }
-.CONTROL_NET[data-v-4b126b34] {
+.CONTROL_NET[data-v-e86c3783] {
   background-color: #a5d6a7;
 }
-._sb_node_preview[data-v-4b126b34] {
+._sb_node_preview[data-v-e86c3783] {
   background-color: var(--comfy-menu-bg);
   font-family: 'Open Sans', sans-serif;
   font-size: small;
@@ -915,7 +922,7 @@
   font-size: 12px;
   padding-bottom: 10px;
 }
-._sb_node_preview ._sb_description[data-v-4b126b34] {
+._sb_node_preview ._sb_description[data-v-e86c3783] {
   margin: 10px;
   padding: 6px;
   background: var(--border-color);
@@ -925,7 +932,7 @@
   font-size: 0.9rem;
   word-break: break-word;
 }
-._sb_table[data-v-4b126b34] {
+._sb_table[data-v-e86c3783] {
   display: grid;
 
   grid-column-gap: 10px;
@@ -933,7 +940,7 @@
   width: 100%;
   /* Imposta la larghezza della tabella al 100% del contenitore */
 }
-._sb_row[data-v-4b126b34] {
+._sb_row[data-v-e86c3783] {
   display: grid;
   grid-template-columns: 10px 1fr 1fr 1fr 10px;
   grid-column-gap: 10px;
@@ -941,10 +948,10 @@
   padding-left: 9px;
   padding-right: 9px;
 }
-._sb_row_string[data-v-4b126b34] {
+._sb_row_string[data-v-e86c3783] {
   grid-template-columns: 10px 1fr 1fr 10fr 1fr;
 }
-._sb_col[data-v-4b126b34] {
+._sb_col[data-v-e86c3783] {
   border: 0px solid #000;
   display: flex;
   align-items: flex-end;
@@ -953,10 +960,10 @@
   align-content: flex-start;
   justify-content: flex-end;
 }
-._sb_inherit[data-v-4b126b34] {
+._sb_inherit[data-v-e86c3783] {
   display: inherit;
 }
-._long_field[data-v-4b126b34] {
+._long_field[data-v-e86c3783] {
   background: var(--bg-color);
   border: 2px solid var(--border-color);
   margin: 5px 5px 0 5px;
@@ -964,45 +971,45 @@
   line-height: 1.7;
   text-wrap: nowrap;
 }
-._sb_arrow[data-v-4b126b34] {
+._sb_arrow[data-v-e86c3783] {
   color: var(--fg-color);
 }
-._sb_preview_badge[data-v-4b126b34] {
+._sb_preview_badge[data-v-e86c3783] {
   text-align: center;
   background: var(--comfy-input-bg);
   font-weight: bold;
   color: var(--error-text);
 }
 
-.node-lib-node-container[data-v-90dfee08] {
+.node-lib-node-container[data-v-20bd95eb] {
     height: 100%;
     width: 100%
 }
 
-.p-selectbutton .p-button[data-v-91077f2a] {
+.p-selectbutton .p-button[data-v-29268946] {
   padding: 0.5rem;
 }
-.p-selectbutton .p-button .pi[data-v-91077f2a] {
+.p-selectbutton .p-button .pi[data-v-29268946] {
   font-size: 1.5rem;
 }
-.field[data-v-91077f2a] {
+.field[data-v-29268946] {
   display: flex;
   flex-direction: column;
   gap: 0.5rem;
 }
-.color-picker-container[data-v-91077f2a] {
+.color-picker-container[data-v-29268946] {
   display: flex;
   align-items: center;
   gap: 0.5rem;
 }
 
-._content[data-v-e7b35fd9] {
+._content[data-v-2fc57c5b] {
 
     display: flex;
 
     flex-direction: column
 }
-._content[data-v-e7b35fd9] > :not([hidden]) ~ :not([hidden]) {
+._content[data-v-2fc57c5b] > :not([hidden]) ~ :not([hidden]) {
 
     --tw-space-y-reverse: 0;
 
@@ -1010,7 +1017,7 @@
 
     margin-bottom: calc(0.5rem * var(--tw-space-y-reverse))
 }
-._footer[data-v-e7b35fd9] {
+._footer[data-v-2fc57c5b] {
 
     display: flex;
 
@@ -1021,10 +1028,10 @@
     padding-top: 1rem
 }
 
-.comfy-image-wrap[data-v-9bc23daf] {
+.comfy-image-wrap[data-v-ffe66146] {
   display: contents;
 }
-.comfy-image-blur[data-v-9bc23daf] {
+.comfy-image-blur[data-v-ffe66146] {
   position: absolute;
   top: 0;
   left: 0;
@@ -1033,7 +1040,7 @@
   -o-object-fit: cover;
      object-fit: cover;
 }
-.comfy-image-main[data-v-9bc23daf] {
+.comfy-image-main[data-v-ffe66146] {
   width: 100%;
   height: 100%;
   -o-object-fit: cover;
@@ -1042,19 +1049,19 @@
      object-position: center;
   z-index: 1;
 }
-.contain .comfy-image-wrap[data-v-9bc23daf] {
+.contain .comfy-image-wrap[data-v-ffe66146] {
   position: relative;
   width: 100%;
   height: 100%;
 }
-.contain .comfy-image-main[data-v-9bc23daf] {
+.contain .comfy-image-main[data-v-ffe66146] {
   -o-object-fit: contain;
      object-fit: contain;
   -webkit-backdrop-filter: blur(10px);
           backdrop-filter: blur(10px);
   position: absolute;
 }
-.broken-image-placeholder[data-v-9bc23daf] {
+.broken-image-placeholder[data-v-ffe66146] {
   display: flex;
   flex-direction: column;
   align-items: center;
@@ -1063,7 +1070,7 @@
   height: 100%;
   margin: 2rem;
 }
-.broken-image-placeholder i[data-v-9bc23daf] {
+.broken-image-placeholder i[data-v-ffe66146] {
   font-size: 3rem;
   margin-bottom: 0.5rem;
 }
@@ -1164,17 +1171,17 @@ img.galleria-image {
   z-index: 1;
 }
 
-.scroll-container[data-v-93f5af09] {
+.scroll-container[data-v-375f3c50] {
   height: 100%;
   overflow-y: auto;
 }
-.scroll-container[data-v-93f5af09]::-webkit-scrollbar {
+.scroll-container[data-v-375f3c50]::-webkit-scrollbar {
   width: 1px;
 }
-.scroll-container[data-v-93f5af09]::-webkit-scrollbar-thumb {
+.scroll-container[data-v-375f3c50]::-webkit-scrollbar-thumb {
   background-color: transparent;
 }
-.queue-grid[data-v-93f5af09] {
+.queue-grid[data-v-375f3c50] {
   display: grid;
   grid-template-columns: repeat(auto-fill, minmax(200px, 1fr));
   padding: 0.5rem;
@@ -2016,12 +2023,22 @@ img.galleria-image {
   .z-\[1000\]{
     z-index: 1000;
   }
+  .col-start-1{
+    grid-column-start: 1;
+  }
+  .row-start-1{
+    grid-row-start: 1;
+  }
   .m-0{
     margin: 0px;
   }
   .m-2{
     margin: 0.5rem;
   }
+  .mx-0{
+    margin-left: 0px;
+    margin-right: 0px;
+  }
   .mx-1{
     margin-left: 0.25rem;
     margin-right: 0.25rem;
@@ -2034,6 +2051,10 @@ img.galleria-image {
     margin-left: 1.5rem;
     margin-right: 1.5rem;
   }
+  .mx-auto{
+    margin-left: auto;
+    margin-right: auto;
+  }
   .my-0{
     margin-top: 0px;
     margin-bottom: 0px;
@@ -2042,6 +2063,14 @@ img.galleria-image {
     margin-top: 0.25rem;
     margin-bottom: 0.25rem;
   }
+  .my-2{
+    margin-top: 0.5rem;
+    margin-bottom: 0.5rem;
+  }
+  .my-2\.5{
+    margin-top: 0.625rem;
+    margin-bottom: 0.625rem;
+  }
   .my-4{
     margin-top: 1rem;
     margin-bottom: 1rem;
@@ -2058,6 +2087,9 @@ img.galleria-image {
   .mb-6{
     margin-bottom: 1.5rem;
   }
+  .mb-7{
+    margin-bottom: 1.75rem;
+  }
   .ml-2{
     margin-left: 0.5rem;
   }
@@ -2082,9 +2114,15 @@ img.galleria-image {
   .mt-2{
     margin-top: 0.5rem;
   }
+  .mt-24{
+    margin-top: 6rem;
+  }
   .mt-4{
     margin-top: 1rem;
   }
+  .mt-5{
+    margin-top: 1.25rem;
+  }
   .mt-\[5vh\]{
     margin-top: 5vh;
   }
@@ -2173,6 +2211,12 @@ img.galleria-image {
   .w-full{
     width: 100%;
   }
+  .w-screen{
+    width: 100vw;
+  }
+  .min-w-84{
+    min-width: 22rem;
+  }
   .min-w-96{
     min-width: 26rem;
   }
@@ -2185,6 +2229,9 @@ img.galleria-image {
   .max-w-full{
     max-width: 100%;
   }
+  .max-w-screen-sm{
+    max-width: 640px;
+  }
   .flex-1{
     flex: 1 1 0%;
   }
@@ -2197,6 +2244,9 @@ img.galleria-image {
   .flex-grow{
     flex-grow: 1;
   }
+  .flex-grow-0{
+    flex-grow: 0;
+  }
   .grow{
     flex-grow: 1;
   }
@@ -2225,6 +2275,9 @@ img.galleria-image {
   .resize{
     resize: both;
   }
+  .list-inside{
+    list-style-position: inside;
+  }
   .list-disc{
     list-style-type: disc;
   }
@@ -2234,6 +2287,9 @@ img.galleria-image {
   .flex-row{
     flex-direction: row;
   }
+  .flex-row-reverse{
+    flex-direction: row-reverse;
+  }
   .flex-col{
     flex-direction: column;
   }
@@ -2243,6 +2299,12 @@ img.galleria-image {
   .flex-nowrap{
     flex-wrap: nowrap;
   }
+  .place-content-center{
+    place-content: center;
+  }
+  .place-items-center{
+    place-items: center;
+  }
   .content-around{
     align-content: space-around;
   }
@@ -2289,6 +2351,11 @@ img.galleria-image {
     margin-top: calc(0.5rem * calc(1 - var(--tw-space-y-reverse)));
     margin-bottom: calc(0.5rem * var(--tw-space-y-reverse));
   }
+  .space-y-4 > :not([hidden]) ~ :not([hidden]){
+    --tw-space-y-reverse: 0;
+    margin-top: calc(1rem * calc(1 - var(--tw-space-y-reverse)));
+    margin-bottom: calc(1rem * var(--tw-space-y-reverse));
+  }
   .justify-self-end{
     justify-self: end;
   }
@@ -2343,6 +2410,9 @@ img.galleria-image {
   .border-none{
     border-style: none;
   }
+  .bg-\[var\(--comfy-menu-bg\)\]{
+    background-color: var(--comfy-menu-bg);
+  }
   .bg-\[var\(--p-tree-background\)\]{
     background-color: var(--p-tree-background);
   }
@@ -2358,6 +2428,10 @@ img.galleria-image {
     --tw-bg-opacity: 1;
     background-color: rgb(150 206 76 / var(--tw-bg-opacity));
   }
+  .bg-neutral-300{
+    --tw-bg-opacity: 1;
+    background-color: rgb(212 212 212 / var(--tw-bg-opacity));
+  }
   .bg-neutral-800{
     --tw-bg-opacity: 1;
     background-color: rgb(38 38 38 / var(--tw-bg-opacity));
@@ -2380,6 +2454,18 @@ img.galleria-image {
   .bg-opacity-50{
     --tw-bg-opacity: 0.5;
   }
+  .bg-\[url\(\'\/assets\/images\/Git-Logo-White\.svg\'\)\]{
+    background-image: url('../assets/images/Git-Logo-White.svg');
+  }
+  .bg-right-top{
+    background-position: right top;
+  }
+  .bg-no-repeat{
+    background-repeat: no-repeat;
+  }
+  .bg-origin-padding{
+    background-origin: padding-box;
+  }
   .object-cover{
     -o-object-fit: cover;
        object-fit: cover;
@@ -2399,6 +2485,9 @@ img.galleria-image {
   .p-4{
     padding: 1rem;
   }
+  .p-5{
+    padding: 1.25rem;
+  }
   .p-8{
     padding: 2rem;
   }
@@ -2406,6 +2495,10 @@ img.galleria-image {
     padding-left: 0px;
     padding-right: 0px;
   }
+  .px-10{
+    padding-left: 2.5rem;
+    padding-right: 2.5rem;
+  }
   .px-2{
     padding-left: 0.5rem;
     padding-right: 0.5rem;
@@ -2425,6 +2518,9 @@ img.galleria-image {
   .pb-0{
     padding-bottom: 0px;
   }
+  .pl-4{
+    padding-left: 1rem;
+  }
   .pl-6{
     padding-left: 1.5rem;
   }
@@ -2455,6 +2551,9 @@ img.galleria-image {
   .text-2xl{
     font-size: 1.5rem;
   }
+  .text-4xl{
+    font-size: 2.25rem;
+  }
   .text-lg{
     font-size: 1.125rem;
   }
@@ -2476,6 +2575,9 @@ img.galleria-image {
   .font-medium{
     font-weight: 500;
   }
+  .font-normal{
+    font-weight: 400;
+  }
   .font-semibold{
     font-weight: 600;
   }
@@ -2493,6 +2595,10 @@ img.galleria-image {
     --tw-text-opacity: 1;
     color: rgb(203 213 224 / var(--tw-text-opacity));
   }
+  .text-green-500{
+    --tw-text-opacity: 1;
+    color: rgb(150 206 76 / var(--tw-text-opacity));
+  }
   .text-highlight{
     color: var(--p-primary-color);
   }
@@ -2515,12 +2621,29 @@ img.galleria-image {
     --tw-text-opacity: 1;
     color: rgb(163 163 163 / var(--tw-text-opacity));
   }
+  .text-neutral-800{
+    --tw-text-opacity: 1;
+    color: rgb(38 38 38 / var(--tw-text-opacity));
+  }
+  .text-neutral-900{
+    --tw-text-opacity: 1;
+    color: rgb(23 23 23 / var(--tw-text-opacity));
+  }
+  .text-red-500{
+    --tw-text-opacity: 1;
+    color: rgb(239 68 68 / var(--tw-text-opacity));
+  }
   .no-underline{
     text-decoration-line: none;
   }
   .opacity-0{
     opacity: 0;
   }
+  .shadow-lg{
+    --tw-shadow: 0 10px 15px -3px rgb(0 0 0 / 0.1), 0 4px 6px -4px rgb(0 0 0 / 0.1);
+    --tw-shadow-colored: 0 10px 15px -3px var(--tw-shadow-color), 0 4px 6px -4px var(--tw-shadow-color);
+    box-shadow: var(--tw-ring-offset-shadow, 0 0 #0000), var(--tw-ring-shadow, 0 0 #0000), var(--tw-shadow);
+  }
   .outline{
     outline-style: solid;
   }
@@ -2580,6 +2703,7 @@ img.galleria-image {
   --fg-color: #000;
   --bg-color: #fff;
   --comfy-menu-bg: #353535;
+  --comfy-menu-secondary-bg: #292929;
   --comfy-input-bg: #222;
   --input-text: #ddd;
   --descrip-text: #999;
@@ -2750,8 +2874,8 @@ body {
 }
 
 .comfy-modal select,
-.comfy-modal input[type=button],
-.comfy-modal input[type=checkbox] {
+.comfy-modal input[type='button'],
+.comfy-modal input[type='checkbox'] {
   margin: 3px 3px 3px 4px;
 }
 
@@ -2865,8 +2989,8 @@ span.drag-handle {
   padding: 3px 4px;
   cursor: move;
   vertical-align: middle;
-  margin-top: -.4em;
-  margin-left: -.2em;
+  margin-top: -0.4em;
+  margin-left: -0.2em;
   font-size: 12px;
   font-family: sans-serif;
   letter-spacing: 2px;
@@ -2933,11 +3057,11 @@ button.comfy-queue-btn {
   z-index: 99;
 }
 
-.comfy-modal.comfy-settings input[type="range"] {
+.comfy-modal.comfy-settings input[type='range'] {
   vertical-align: middle;
 }
 
-.comfy-modal.comfy-settings input[type="range"] + input[type="number"] {
+.comfy-modal.comfy-settings input[type='range'] + input[type='number'] {
   width: 3.5em;
 }
 
@@ -2968,7 +3092,7 @@ button.comfy-queue-btn {
   .comfy-menu span.drag-handle {
     display: none;
   }
-  
+
   .comfy-menu-queue-size {
     flex: unset;
   }
@@ -3002,7 +3126,9 @@ button.comfy-queue-btn {
   padding-right: 8px;
 }
 
-.graphdialog input, .graphdialog textarea, .graphdialog select {
+.graphdialog input,
+.graphdialog textarea,
+.graphdialog select {
   background-color: var(--comfy-input-bg);
   border: 2px solid;
   border-color: var(--border-color);
@@ -3064,7 +3190,8 @@ dialog::backdrop {
   text-align: right;
 }
 
-#comfy-settings-dialog tbody button, #comfy-settings-dialog table > button {
+#comfy-settings-dialog tbody button,
+#comfy-settings-dialog table > button {
   background-color: var(--bg-color);
   border: 1px var(--border-color) solid;
   border-radius: 0;
@@ -3145,7 +3272,7 @@ dialog::backdrop {
 }
 
 .litemenu-entry.has_submenu::after {
-  content: ">";
+  content: '>';
   position: absolute;
   top: 0;
   right: 2px;
@@ -3159,7 +3286,8 @@ dialog::backdrop {
   will-change: transform;
 }
 
-.litegraph.litecontextmenu .litemenu-entry:hover:not(.disabled):not(.separator) {
+.litegraph.litecontextmenu
+  .litemenu-entry:hover:not(.disabled):not(.separator) {
   background-color: var(--comfy-menu-bg) !important;
   filter: brightness(155%);
   will-change: transform;
@@ -3288,6 +3416,17 @@ audio.comfy-audio.empty-audio-widget {
   opacity: 1;
 }
 
+@media (min-width: 768px){
+
+  .md\:flex{
+    display: flex;
+  }
+
+  .md\:hidden{
+    display: none;
+  }
+}
+
 @media (min-width: 1536px){
 
   .\32xl\:mx-4{

web/assets/index-BppSBmxJ.js

@@ -1,106 +0,0 @@
-const IPC_CHANNELS = {
-  LOADING_PROGRESS: "loading-progress",
-  IS_PACKAGED: "is-packaged",
-  RENDERER_READY: "renderer-ready",
-  RESTART_APP: "restart-app",
-  REINSTALL: "reinstall",
-  LOG_MESSAGE: "log-message",
-  OPEN_DIALOG: "open-dialog",
-  DOWNLOAD_PROGRESS: "download-progress",
-  START_DOWNLOAD: "start-download",
-  PAUSE_DOWNLOAD: "pause-download",
-  RESUME_DOWNLOAD: "resume-download",
-  CANCEL_DOWNLOAD: "cancel-download",
-  DELETE_MODEL: "delete-model",
-  GET_ALL_DOWNLOADS: "get-all-downloads",
-  GET_ELECTRON_VERSION: "get-electron-version",
-  SEND_ERROR_TO_SENTRY: "send-error-to-sentry",
-  GET_BASE_PATH: "get-base-path",
-  GET_MODEL_CONFIG_PATH: "get-model-config-path",
-  OPEN_PATH: "open-path",
-  OPEN_LOGS_PATH: "open-logs-path",
-  OPEN_DEV_TOOLS: "open-dev-tools",
-  TERMINAL_WRITE: "execute-terminal-command",
-  TERMINAL_RESIZE: "resize-terminal",
-  TERMINAL_RESTORE: "restore-terminal",
-  TERMINAL_ON_OUTPUT: "terminal-output",
-  IS_FIRST_TIME_SETUP: "is-first-time-setup",
-  GET_SYSTEM_PATHS: "get-system-paths",
-  VALIDATE_INSTALL_PATH: "validate-install-path",
-  VALIDATE_COMFYUI_SOURCE: "validate-comfyui-source",
-  SHOW_DIRECTORY_PICKER: "show-directory-picker",
-  INSTALL_COMFYUI: "install-comfyui"
-};
-var ProgressStatus = /* @__PURE__ */ ((ProgressStatus2) => {
-  ProgressStatus2["INITIAL_STATE"] = "initial-state";
-  ProgressStatus2["PYTHON_SETUP"] = "python-setup";
-  ProgressStatus2["STARTING_SERVER"] = "starting-server";
-  ProgressStatus2["READY"] = "ready";
-  ProgressStatus2["ERROR"] = "error";
-  return ProgressStatus2;
-})(ProgressStatus || {});
-const ProgressMessages = {
-  [
-    "initial-state"
-    /* INITIAL_STATE */
-  ]: "Loading...",
-  [
-    "python-setup"
-    /* PYTHON_SETUP */
-  ]: "Setting up Python Environment...",
-  [
-    "starting-server"
-    /* STARTING_SERVER */
-  ]: "Starting ComfyUI server...",
-  [
-    "ready"
-    /* READY */
-  ]: "Finishing...",
-  [
-    "error"
-    /* ERROR */
-  ]: "Was not able to start ComfyUI. Please check the logs for more details. You can open it from the Help menu. Please report issues to: https://forum.comfy.org"
-};
-const ELECTRON_BRIDGE_API = "electronAPI";
-const SENTRY_URL_ENDPOINT = "https://942cadba58d247c9cab96f45221aa813@o4507954455314432.ingest.us.sentry.io/4508007940685824";
-const MigrationItems = [
-  {
-    id: "user_files",
-    label: "User Files",
-    description: "Settings and user-created workflows"
-  },
-  {
-    id: "models",
-    label: "Models",
-    description: "Reference model files from existing ComfyUI installations. (No copy)"
-  }
-  // TODO: Decide whether we want to auto-migrate custom nodes, and install their dependencies.
-  // huchenlei: This is a very essential thing for migration experience.
-  // {
-  //   id: 'custom_nodes',
-  //   label: 'Custom Nodes',
-  //   description: 'Reference custom node files from existing ComfyUI installations. (No copy)',
-  // },
-];
-const DEFAULT_SERVER_ARGS = {
-  /** The host to use for the ComfyUI server. */
-  host: "127.0.0.1",
-  /** The port to use for the ComfyUI server. */
-  port: 8e3,
-  // Extra arguments to pass to the ComfyUI server.
-  extraServerArgs: {}
-};
-var DownloadStatus = /* @__PURE__ */ ((DownloadStatus2) => {
-  DownloadStatus2["PENDING"] = "pending";
-  DownloadStatus2["IN_PROGRESS"] = "in_progress";
-  DownloadStatus2["COMPLETED"] = "completed";
-  DownloadStatus2["PAUSED"] = "paused";
-  DownloadStatus2["ERROR"] = "error";
-  DownloadStatus2["CANCELLED"] = "cancelled";
-  return DownloadStatus2;
-})(DownloadStatus || {});
-export {
-  MigrationItems as M,
-  ProgressStatus as P
-};
-//# sourceMappingURL=index-BppSBmxJ.js.map