MCPcopy Create free account
hub / github.com/OpenImagingLab/FlashVSR / decode

Method decode

diffsynth/models/stepvideo_vae.py:1103–1110  ·  view source on GitHub ↗
(self, hidden_states, device, tiled=False, tile_size=(34, 34), tile_stride=(16, 16), smooth_scale=0.6)

Source from the content-addressed store, hash-verified

1101 return values / weight
1102
1103 def decode(self, hidden_states, device, tiled=False, tile_size=(34, 34), tile_stride=(16, 16), smooth_scale=0.6):
1104 hidden_states = hidden_states.to("cpu")
1105 if tiled:
1106 video = self.tiled_decode(hidden_states, device, tile_size, tile_stride)
1107 else:
1108 video = self.single_decode(hidden_states, device)
1109 video = self.mix(video, smooth_scale=smooth_scale)
1110 return video
1111
1112 @staticmethod
1113 def state_dict_converter():

Callers

nothing calls this directly

Calls 4

tiled_decodeMethod · 0.95
single_decodeMethod · 0.95
mixMethod · 0.95
toMethod · 0.45

Tested by

no test coverage detected