298977887/ComfyUI
1
0
Fork 0
mirror of https://github.com/comfyanonymous/ComfyUI.git synced 2025-01-11 10:25:16 +00:00
Code Issues Packages Projects Releases Wiki Activity

Move to using generator for the noise, making seed required

Browse Source
This commit is contained in:
Harel Cain 2024-04-18 10:28:20 +03:00
parent 43cf3f6ec2
commit 2d3c1d689f
1 changed files with 6 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
comfy_extras/nodes_video_model.py
View File

@ -33,9 +33,8 @@ class SVD_img2vid_Conditioning:
"video_frames": ("INT", {"default": 14, "min": 1, "max": 4096}), "video_frames": ("INT", {"default": 14, "min": 1, "max": 4096}),
"motion_bucket_id": ("INT", {"default": 127, "min": 1, "max": 1023}), "motion_bucket_id": ("INT", {"default": 127, "min": 1, "max": 1023}),
"fps": ("INT", {"default": 6, "min": 1, "max": 1024}), "fps": ("INT", {"default": 6, "min": 1, "max": 1024}),
"augmentation_level": ("FLOAT", {"default": 0.0, "min": 0.0, "max": 10.0, "step": 0.01}) "augmentation_level": ("FLOAT", {"default": 0.0, "min": 0.0, "max": 10.0, "step": 0.01}),
}, "noise_seed": ("INT", {"default": 0, "min": 0, "max": 2**32-1}),
"optional": { "noise_seed": ("INT", {"default": 0, "min": 0, "max": 2**32-1}),
} }
} }
RETURN_TYPES = ("CONDITIONING", "CONDITIONING", "LATENT") RETURN_TYPES = ("CONDITIONING", "CONDITIONING", "LATENT")
@ -46,15 +45,14 @@ class SVD_img2vid_Conditioning:
CATEGORY = "conditioning/video_models" CATEGORY = "conditioning/video_models"
def encode(self, clip_vision, init_image, vae, width, height, video_frames, motion_bucket_id, def encode(self, clip_vision, init_image, vae, width, height, video_frames, motion_bucket_id,
fps, augmentation_level, noise_seed=None): fps, augmentation_level, noise_seed):
output = clip_vision.encode_image(init_image) output = clip_vision.encode_image(init_image)
pooled = output.image_embeds.unsqueeze(0) pooled = output.image_embeds.unsqueeze(0)
pixels = comfy.utils.common_upscale(init_image.movedim(-1,1), width, height, "bilinear", "center").movedim(1,-1) pixels = comfy.utils.common_upscale(init_image.movedim(-1,1), width, height, "bilinear", "center").movedim(1,-1)
encode_pixels = pixels[:,:,:,:3] encode_pixels = pixels[:,:,:,:3]
if augmentation_level > 0: if augmentation_level > 0:
if noise_seed is not None: generator = torch.manual_seed(noise_seed)
torch.manual_seed(noise_seed) encode_pixels += torch.randn(pixels.shape, generator=generator) * augmentation_level
encode_pixels += torch.randn_like(pixels) * augmentation_level
t = vae.encode(encode_pixels) t = vae.encode(encode_pixels)
positive = [[pooled, {"motion_bucket_id": motion_bucket_id, "fps": fps, "augmentation_level": augmentation_level, "concat_latent_image": t}]] positive = [[pooled, {"motion_bucket_id": motion_bucket_id, "fps": fps, "augmentation_level": augmentation_level, "concat_latent_image": t}]]
negative = [[torch.zeros_like(pooled), {"motion_bucket_id": motion_bucket_id, "fps": fps, "augmentation_level": augmentation_level, "concat_latent_image": torch.zeros_like(t)}]] negative = [[torch.zeros_like(pooled), {"motion_bucket_id": motion_bucket_id, "fps": fps, "augmentation_level": augmentation_level, "concat_latent_image": torch.zeros_like(t)}]]