theluyuan/DiffSynth-Studio
1
0
Fork 0
mirror of https://github.com/modelscope/DiffSynth-Studio.git synced 2026-03-24 10:18:12 +00:00
Code Issues Packages Projects Releases Wiki Activity

Merge pull request #859 from krahets/main

Browse Source 
Fix batch decoding for Wan-Video-VAE
This commit is contained in:
Zhongjie Duan
2025-08-29 12:45:49 +08:00
committed by GitHub
parent 5b6d112c15 419d47c195
commit 826bab5962
1 changed files with 12 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

18
diffsynth/models/wan_video_vae.py
View File

@@ -1216,7 +1216,6 @@ class WanVideoVAE(nn.Module):
def encode(self, videos, device, tiled=False, tile_size=(34, 34), tile_stride=(18, 16)): def encode(self, videos, device, tiled=False, tile_size=(34, 34), tile_stride=(18, 16)):
videos = [video.to("cpu") for video in videos] videos = [video.to("cpu") for video in videos]
hidden_states = [] hidden_states = []
for video in videos: for video in videos:
@@ -1234,11 +1233,18 @@ class WanVideoVAE(nn.Module):
def decode(self, hidden_states, device, tiled=False, tile_size=(34, 34), tile_stride=(18, 16)): def decode(self, hidden_states, device, tiled=False, tile_size=(34, 34), tile_stride=(18, 16)):
if tiled: hidden_states = [hidden_state.to("cpu") for hidden_state in hidden_states]
video = self.tiled_decode(hidden_states, device, tile_size, tile_stride) videos = []
else: for hidden_state in hidden_states:
video = self.single_decode(hidden_states, device) hidden_state = hidden_state.unsqueeze(0)
return video if tiled:
video = self.tiled_decode(hidden_state, device, tile_size, tile_stride)
else:
video = self.single_decode(hidden_state, device)
video = video.squeeze(0)
videos.append(video)
videos = torch.stack(videos)
return videos
@staticmethod @staticmethod