feat(anima): add image-to-image and inpainting support

modules/modeldata.py

@@ -72,7 +72,7 @@ def get_model_type(pipe):
         model_type = 'sana'
     elif "HiDream" in name:
         model_type = 'h1'
-    elif "AnimaTextToImage" in name:
+    elif name.startswith("Anima") and "AnimateDiff" not in name:
         model_type = 'anima'
     elif "Cosmos2TextToImage" in name:
         model_type = 'cosmos'

modules/processing_args.py

@@ -272,7 +272,7 @@ def set_pipeline_args(p, model, prompts:list, negative_prompts:list, prompts_2:l
             kwargs['output_type'] = 'np' # only set latent if model has vae
 
     # model specific
-    if 'Kandinsky' in model.__class__.__name__ or 'Cosmos2' in model.__class__.__name__ or 'Anima' in model.__class__.__name__ or 'OmniGen2' in model.__class__.__name__:
+    if 'Kandinsky' in model.__class__.__name__ or 'Cosmos2' in model.__class__.__name__ or 'OmniGen2' in model.__class__.__name__:
         kwargs['output_type'] = 'np' # only set latent if model has vae
     if 'StableCascade' in model.__class__.__name__:
         kwargs.pop("guidance_scale") # remove

modules/processing_vae.py

@@ -247,6 +247,8 @@ def vae_postprocess(tensor, model, output_type='np'):
                 if tensor.ndim == 6 and tensor.shape[1] == 1:
                     tensor = tensor.squeeze(0)
                 images = model.video_processor.postprocess_video(tensor, output_type='pil')
+                if isinstance(images, list) and len(images) > 0 and isinstance(images[0], list):
+                    images = [frame for batch in images for frame in batch]
             elif hasattr(model, 'image_processor'):
                 if tensor.ndim == 5 and tensor.shape[1] == 3: # Qwen Image
                     tensor = tensor[:, :, 0]