298977887/ComfyUI
1
0
Fork 0
mirror of https://github.com/comfyanonymous/ComfyUI.git synced 2025-07-08 00:07:09 +08:00
Code Issues Packages Projects Releases Wiki Activity

Add optional noise_seed to make augmentation deterministic

Browse Source
This commit is contained in:
Harel Cain 2024-04-17 12:46:33 +03:00
parent 45ec1cbe96
commit 43cf3f6ec2
1 changed files with 8 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
comfy_extras/nodes_video_model.py
View File

@ -34,7 +34,10 @@ class SVD_img2vid_Conditioning:
"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})
}} },
"optional": { "noise_seed": ("INT", {"default": 0, "min": 0, "max": 2**32-1}),
}
}
RETURN_TYPES = ("CONDITIONING", "CONDITIONING", "LATENT") RETURN_TYPES = ("CONDITIONING", "CONDITIONING", "LATENT")
RETURN_NAMES = ("positive", "negative", "latent") RETURN_NAMES = ("positive", "negative", "latent")
@ -42,12 +45,15 @@ 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, fps, augmentation_level): def encode(self, clip_vision, init_image, vae, width, height, video_frames, motion_bucket_id,
fps, augmentation_level, noise_seed=None):
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:
torch.manual_seed(noise_seed)
encode_pixels += torch.randn_like(pixels) * 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}]]