Fix channel padding with new Comfy core API (#43)

* Fix channel padding with new Comfy core API

* nit

README.md

@@ -19,11 +19,8 @@ There are 2 models:
 
 After you download these models, please put them under `ComfyUI/models/unet` and load them with `UNETLoader` node.
 
-### [Important!] Required nodes
-You MUST install following nodes first for IC light to work properly.
-- [ComfyUI-layerdiffuse](https://github.com/huchenlei/ComfyUI-layerdiffuse): Although not used in the workflow, the patching of weight load in layerdiffuse is a dependency for IC-Light nodes to work properly.
-
 ### Recommended nodes
+
 - [ComfyUI-KJNodes](https://github.com/kijai/ComfyUI-KJNodes): Provides various mask nodes to create light map.
 - [ComfyUI-Easy-Use](https://github.com/yolain/ComfyUI-Easy-Use): A giant node pack of everything. The remove bg node used in workflow comes from this pack.
 - [ComfyUI_essentials](https://github.com/cubiq/ComfyUI_essentials): Many useful tooling nodes. Image resize node used in the workflow comes from this pack.
@@ -61,12 +58,7 @@ Light from left
 
 ## Common Issues
 IC-Light's unet is accepting extra inputs on top of the common noise input. FG model accepts extra 1 input (4 channels). BG model accepts 2 extra input (8 channels).
-The original unet's input is 4 channels as well. 
-
-If you see following issue, it means IC-Light's unet is not properly loaded, and you need to install [ComfyUI-layerdiffuse](https://github.com/huchenlei/ComfyUI-layerdiffuse) first.
-```
-RuntimeError: Given groups=1, weight of size [320, 4, 3, 3], expected input[2, 8, 64, 64] to have 4 channels, but got 8 channels instead
-```
+The original unet's input is 4 channels as well.
 
 If you see following error, it means you are using FG workflow but loaded the BG model.
 ```

ic_light_nodes.py

@@ -120,7 +120,14 @@ def wrapper_func(unet_apply: Callable, params: UnetParams):
         ic_model_state_dict = ic_model.model.diffusion_model.state_dict()
         work_model.add_patches(
             patches={
-                ("diffusion_model." + key): (value.to(dtype=dtype, device=device),)
+                ("diffusion_model." + key): (
+                    "diff",
+                    [
+                        value.to(dtype=dtype, device=device),
+                        # The extra flag to pad the weight if the diff's shape is larger than the weight
+                        {"pad_weight": key == "input_blocks.0.0.weight"},
+                    ],
+                )
                 for key, value in ic_model_state_dict.items()
             }
         )