add compatibility with ipadapter

InstantID.py

@@ -51,7 +51,7 @@ def draw_kps(image_pil, kps, color_list=[(255,0,0), (0,255,0), (0,0,255), (255,2
     out_img_pil = PIL.Image.fromarray(out_img.astype(np.uint8))
     return out_img_pil
 
-class CrossAttentionPatch:
+class CrossAttentionPatchIID:
     # forward for patching
     def __init__(self, weight, instantid, number, cond, uncond, mask=None, sigma_start=0.0, sigma_end=1.0):
         self.weights = [weight]
@@ -191,14 +191,14 @@ def __init__(self, state_dict):
             self.to_kvs[k] = torch.nn.Linear(value.shape[1], value.shape[0], bias=False)
             self.to_kvs[k].weight.data = value
 
-def set_model_patch_replace(model, patch_kwargs, key):
+def _set_model_patch_replace(model, patch_kwargs, key):
     to = model.model_options["transformer_options"]
     if "patches_replace" not in to:
         to["patches_replace"] = {}
     if "attn2" not in to["patches_replace"]:
         to["patches_replace"]["attn2"] = {}
-    if key not in to["patches_replace"]["attn2"]:
-        patch = CrossAttentionPatch(**patch_kwargs)
+    if key not in to["patches_replace"]["attn2"] or not isinstance(to["patches_replace"]["attn2"][key], CrossAttentionPatchIID):
+        patch = CrossAttentionPatchIID(**patch_kwargs)
         to["patches_replace"]["attn2"][key] = patch
     else:
         to["patches_replace"]["attn2"][key].set_new_condition(**patch_kwargs)
@@ -388,25 +388,25 @@ def apply_instantid(self, instantid, insightface, image_features, model, positiv
 
         if not is_sdxl:
             for id in [1,2,4,5,7,8]: # id of input_blocks that have cross attention
-                set_model_patch_replace(work_model, patch_kwargs, ("input", id))
+                _set_model_patch_replace(work_model, patch_kwargs, ("input", id))
                 patch_kwargs["number"] += 1
             for id in [3,4,5,6,7,8,9,10,11]: # id of output_blocks that have cross attention
-                set_model_patch_replace(work_model, patch_kwargs, ("output", id))
+                _set_model_patch_replace(work_model, patch_kwargs, ("output", id))
                 patch_kwargs["number"] += 1
-            set_model_patch_replace(work_model, patch_kwargs, ("middle", 0))
+            _set_model_patch_replace(work_model, patch_kwargs, ("middle", 0))
         else:
             for id in [4,5,7,8]: # id of input_blocks that have cross attention
                 block_indices = range(2) if id in [4, 5] else range(10) # transformer_depth
                 for index in block_indices:
-                    set_model_patch_replace(work_model, patch_kwargs, ("input", id, index))
+                    _set_model_patch_replace(work_model, patch_kwargs, ("input", id, index))
                     patch_kwargs["number"] += 1
             for id in range(6): # id of output_blocks that have cross attention
                 block_indices = range(2) if id in [3, 4, 5] else range(10) # transformer_depth
                 for index in block_indices:
-                    set_model_patch_replace(work_model, patch_kwargs, ("output", id, index))
+                    _set_model_patch_replace(work_model, patch_kwargs, ("output", id, index))
                     patch_kwargs["number"] += 1
             for index in range(10):
-                set_model_patch_replace(work_model, patch_kwargs, ("middle", 0, index))
+                _set_model_patch_replace(work_model, patch_kwargs, ("middle", 0, index))
                 patch_kwargs["number"] += 1
 
         pos = []