wangbo/ComfyUI
1
0
Fork 0
mirror of https://github.com/comfyanonymous/ComfyUI.git synced 2025-12-20 03:23:00 +08:00
Code Issues Actions 11 Packages Projects Releases Wiki Activity

Merge branch 'master' into dr-support-pip-cm

Browse Source
This commit is contained in:
Dr.Lt.Data 2025-09-19 12:00:10 +09:00
commit fa51f0c60a
4 changed files with 5 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
comfy/ldm/wan/model.py
View File

@ -1551,6 +1551,9 @@ class HumoWanModel(WanModel):
context_img_len = None context_img_len = None
if audio_embed is not None: if audio_embed is not None:
if reference_latent is not None:
zero_audio_pad = torch.zeros(audio_embed.shape[0], reference_latent.shape[-3], *audio_embed.shape[2:], device=audio_embed.device, dtype=audio_embed.dtype)
audio_embed = torch.cat([audio_embed, zero_audio_pad], dim=1)
audio = self.audio_proj(audio_embed).permute(0, 3, 1, 2).flatten(2).transpose(1, 2) audio = self.audio_proj(audio_embed).permute(0, 3, 1, 2).flatten(2).transpose(1, 2)
else: else:
audio = None audio = None

2
comfy/model_management.py
View File

@ -348,7 +348,7 @@ try:
# if any((a in arch) for a in ["gfx1201"]): # if any((a in arch) for a in ["gfx1201"]):
# ENABLE_PYTORCH_ATTENTION = True # ENABLE_PYTORCH_ATTENTION = True
if torch_version_numeric >= (2, 7) and rocm_version >= (6, 4): if torch_version_numeric >= (2, 7) and rocm_version >= (6, 4):
if any((a in arch) for a in ["gfx1201", "gfx942", "gfx950"]): # TODO: more arches if any((a in arch) for a in ["gfx1200", "gfx1201", "gfx942", "gfx950"]): # TODO: more arches
SUPPORT_FP8_OPS = True SUPPORT_FP8_OPS = True
except: except:

1
comfy_extras/nodes_post_processing.py
View File

@ -233,6 +233,7 @@ class Sharpen:
kernel_size = sharpen_radius * 2 + 1 kernel_size = sharpen_radius * 2 + 1
kernel = gaussian_kernel(kernel_size, sigma, device=image.device) * -(alpha*10) kernel = gaussian_kernel(kernel_size, sigma, device=image.device) * -(alpha*10)
kernel = kernel.to(dtype=image.dtype)
center = kernel_size // 2 center = kernel_size // 2
kernel[center, center] = kernel[center, center] - kernel.sum() + 1.0 kernel[center, center] = kernel[center, center] - kernel.sum() + 1.0
kernel = kernel.repeat(channels, 1, 1).unsqueeze(1) kernel = kernel.repeat(channels, 1, 1).unsqueeze(1)

4
comfy_extras/nodes_wan.py
View File

@ -1095,10 +1095,6 @@ class WanHuMoImageToVideo(io.ComfyNode):
audio_emb = torch.stack([feat0, feat1, feat2, feat3, feat4], dim=2)[0] # [T, 5, 1280] audio_emb = torch.stack([feat0, feat1, feat2, feat3, feat4], dim=2)[0] # [T, 5, 1280]
audio_emb, _ = get_audio_emb_window(audio_emb, length, frame0_idx=0) audio_emb, _ = get_audio_emb_window(audio_emb, length, frame0_idx=0)
# pad for ref latent
zero_audio_pad = torch.zeros(ref_latent.shape[2], *audio_emb.shape[1:], device=audio_emb.device, dtype=audio_emb.dtype)
audio_emb = torch.cat([audio_emb, zero_audio_pad], dim=0)
audio_emb = audio_emb.unsqueeze(0) audio_emb = audio_emb.unsqueeze(0)
audio_emb_neg = torch.zeros_like(audio_emb) audio_emb_neg = torch.zeros_like(audio_emb)
positive = node_helpers.conditioning_set_values(positive, {"audio_embed": audio_emb}) positive = node_helpers.conditioning_set_values(positive, {"audio_embed": audio_emb})