mirrors/ComfyUI
1
0
Fork 0
mirror of https://github.com/comfyanonymous/ComfyUI.git synced 2025-09-10 11:35:40 +00:00
Code Issues Packages Projects Releases Wiki Activity

Support new hunyuan video i2v model.

Browse Source 
Use the new "v2 (replace)" guidance type in HunyuanImageToVideo and set
image_interleave to 4 on the "Text Encode Hunyuan Video" node.
This commit is contained in:
comfyanonymous
2025-03-08 20:25:14 -05:00
parent 0952569493
commit 7395b0c0d1
4 changed files with 72 additions and 24 deletions
Download Patch File Download Diff File Expand all files Collapse all files

17
comfy_extras/nodes_hunyuan.py
View File

@@ -68,7 +68,6 @@ class TextEncodeHunyuanVideo_ImageToVideo:
tokens = clip.tokenize(prompt, llama_template=PROMPT_TEMPLATE_ENCODE_VIDEO_I2V, image_embeds=clip_vision_output.mm_projected, image_interleave=image_interleave)
return (clip.encode_from_tokens_scheduled(tokens), )
class HunyuanImageToVideo:
@classmethod
def INPUT_TYPES(s):
@@ -78,6 +77,7 @@ class HunyuanImageToVideo:
"height": ("INT", {"default": 480, "min": 16, "max": nodes.MAX_RESOLUTION, "step": 16}),
"length": ("INT", {"default": 53, "min": 1, "max": nodes.MAX_RESOLUTION, "step": 4}),
"batch_size": ("INT", {"default": 1, "min": 1, "max": 4096}),
"guidance_type": (["v1 (concat)", "v2 (replace)"], )
},
"optional": {"start_image": ("IMAGE", ),
}}
@@ -88,8 +88,10 @@ class HunyuanImageToVideo:
CATEGORY = "conditioning/video_models"
def encode(self, positive, vae, width, height, length, batch_size, start_image=None):
def encode(self, positive, vae, width, height, length, batch_size, guidance_type, start_image=None):
latent = torch.zeros([batch_size, 16, ((length - 1) // 4) + 1, height // 8, width // 8], device=comfy.model_management.intermediate_device())
out_latent = {}
if start_image is not None:
start_image = comfy.utils.common_upscale(start_image[:length, :, :, :3].movedim(-1, 1), width, height, "bilinear", "center").movedim(1, -1)
@@ -97,13 +99,20 @@ class HunyuanImageToVideo:
mask = torch.ones((1, 1, latent.shape[2], concat_latent_image.shape[-2], concat_latent_image.shape[-1]), device=start_image.device, dtype=start_image.dtype)
mask[:, :, :((start_image.shape[0] - 1) // 4) + 1] = 0.0
positive = node_helpers.conditioning_set_values(positive, {"concat_latent_image": concat_latent_image, "concat_mask": mask})
if guidance_type == "v1 (concat)":
cond = {"concat_latent_image": concat_latent_image, "concat_mask": mask}
else:
cond = {'guiding_frame_index': 0}
latent[:, :, :concat_latent_image.shape[2]] = concat_latent_image
out_latent["noise_mask"] = mask
positive = node_helpers.conditioning_set_values(positive, cond)
out_latent = {}
out_latent["samples"] = latent
return (positive, out_latent)
NODE_CLASS_MAPPINGS = {
"CLIPTextEncodeHunyuanDiT": CLIPTextEncodeHunyuanDiT,
"TextEncodeHunyuanVideo_ImageToVideo": TextEncodeHunyuanVideo_ImageToVideo,