[pull] master from comfyanonymous:master

comfy/ldm/flux/controlnet.py

@@ -121,6 +121,9 @@ def forward_orig(
         if img.ndim != 3 or txt.ndim != 3:
             raise ValueError("Input img and txt tensors must have 3 dimensions.")
 
+        if y is None:
+            y = torch.zeros((img.shape[0], self.params.vec_in_dim), device=img.device, dtype=img.dtype)
+
         # running on sequences img
         img = self.img_in(img)
 
@@ -174,7 +177,7 @@ def forward_orig(
             out["output"] = out_output[:self.main_model_single]
         return out
 
-    def forward(self, x, timesteps, context, y, guidance=None, hint=None, **kwargs):
+    def forward(self, x, timesteps, context, y=None, guidance=None, hint=None, **kwargs):
         patch_size = 2
         if self.latent_input:
             hint = comfy.ldm.common_dit.pad_to_patch_size(hint, (patch_size, patch_size))

comfy/ldm/flux/model.py

@@ -101,6 +101,10 @@ def forward_orig(
         transformer_options={},
         attn_mask: Tensor = None,
     ) -> Tensor:
+
+        if y is None:
+            y = torch.zeros((img.shape[0], self.params.vec_in_dim), device=img.device, dtype=img.dtype)
+
         patches_replace = transformer_options.get("patches_replace", {})
         if img.ndim != 3 or txt.ndim != 3:
             raise ValueError("Input img and txt tensors must have 3 dimensions.")
@@ -188,7 +192,7 @@ def block_wrap(args):
         img = self.final_layer(img, vec)  # (N, T, patch_size ** 2 * out_channels)
         return img
 
-    def forward(self, x, timestep, context, y, guidance=None, control=None, transformer_options={}, **kwargs):
+    def forward(self, x, timestep, context, y=None, guidance=None, control=None, transformer_options={}, **kwargs):
         bs, c, h, w = x.shape
         patch_size = self.patch_size
         x = comfy.ldm.common_dit.pad_to_patch_size(x, (patch_size, patch_size))