The most powerful and modular diffusion model GUI, api and backend with a graph/nodes interface. The fastest local inference engine in the world.
Remove unecessary clones in the wan2.2 VAE. (#9083)
comfyanonymousc60dc4177c16
diff --git a/comfy/ldm/wan/vae2_2.py b/comfy/ldm/wan/vae2_2.pyindex c2c150e10..b9c2d1a26 100644--- a/comfy/ldm/wan/vae2_2.py+++ b/comfy/ldm/wan/vae2_2.py@@ -136,7 +136,7 @@ class ResidualBlock(nn.Module):if in_dim != out_dim else nn.Identity())def forward(self, x, feat_cache=None, feat_idx=[0]):- h = self.shortcut(x)+ old_x = xfor layer in self.residual:if isinstance(layer, CausalConv3d) and feat_cache is not None:idx = feat_idx[0]@@ -156,7 +156,7 @@ class ResidualBlock(nn.Module):feat_idx[0] += 1else:x = layer(x)- return x + h+ return x + self.shortcut(old_x)def patchify(x, patch_size):@@ -327,7 +327,7 @@ class Down_ResidualBlock(nn.Module):self.downsamples = nn.Sequential(*downsamples)def forward(self, x, feat_cache=None, feat_idx=[0]):- x_copy = x.clone()+ x_copy = xfor module in self.downsamples:x = module(x, feat_cache, feat_idx)@@ -369,7 +369,7 @@ class Up_ResidualBlock(nn.Module):self.upsamples = nn.Sequential(*upsamples)def forward(self, x, feat_cache=None, feat_idx=[0], first_chunk=False):- x_main = x.clone()+ x_main = xfor module in self.upsamples:x_main = module(x_main, feat_cache, feat_idx)if self.avg_shortcut is not None: