Update merged_app2.py

merged_app2.py

@@ -116,7 +116,7 @@ model_path3 = './checkpoints/sam2_hiera_large.pt'
 model_path4 = './checkpoints/config.json'
 model_path5 = './checkpoints/preprocessor_config.json'
 model_path6 = './configs/sam2_hiera_l.yaml'
-model_path7 = './mvadapter_i2mv_sdxl.safetensors'
+#model_path7 = './mvadapter_i2mv_sdxl.safetensors'
 
 # Base URL for the repository
 BASE_URL = 'https://huggingface.co/Ashoka74/Placement/resolve/main/'
@@ -129,7 +129,7 @@ model_urls = {
     model_path4: 'config.json',
     model_path5: 'preprocessor_config.json',
     model_path6: 'sam2_hiera_l.yaml',
-    model_path7: 'mvadapter_i2mv_sdxl.safetensors'
+#    model_path7: 'mvadapter_i2mv_sdxl.safetensors'
 }
 
 # Ensure directories exist
@@ -153,13 +153,13 @@ ensure_directories()
 download_models()
 
 
-hf_hub_download(repo_id="black-forest-labs/FLUX.1-Redux-dev", filename="flux1-redux-dev.safetensors", local_dir="models/style_models")
-hf_hub_download(repo_id="black-forest-labs/FLUX.1-Depth-dev", filename="flux1-depth-dev.safetensors", local_dir="models/diffusion_models")
-hf_hub_download(repo_id="Comfy-Org/sigclip_vision_384", filename="sigclip_vision_patch14_384.safetensors", local_dir="models/clip_vision")
-hf_hub_download(repo_id="Kijai/DepthAnythingV2-safetensors", filename="depth_anything_v2_vitl_fp32.safetensors", local_dir="models/depthanything")
-hf_hub_download(repo_id="black-forest-labs/FLUX.1-dev", filename="ae.safetensors", local_dir="models/vae/FLUX1")
-hf_hub_download(repo_id="comfyanonymous/flux_text_encoders", filename="clip_l.safetensors", local_dir="models/text_encoders")
-t5_path = hf_hub_download(repo_id="comfyanonymous/flux_text_encoders", filename="t5xxl_fp16.safetensors", local_dir="models/text_encoders/t5")
+# hf_hub_download(repo_id="black-forest-labs/FLUX.1-Redux-dev", filename="flux1-redux-dev.safetensors", local_dir="models/style_models")
+# hf_hub_download(repo_id="black-forest-labs/FLUX.1-Depth-dev", filename="flux1-depth-dev.safetensors", local_dir="models/diffusion_models")
+# hf_hub_download(repo_id="Comfy-Org/sigclip_vision_384", filename="sigclip_vision_patch14_384.safetensors", local_dir="models/clip_vision")
+# hf_hub_download(repo_id="Kijai/DepthAnythingV2-safetensors", filename="depth_anything_v2_vitl_fp32.safetensors", local_dir="models/depthanything")
+# hf_hub_download(repo_id="black-forest-labs/FLUX.1-dev", filename="ae.safetensors", local_dir="models/vae/FLUX1")
+# hf_hub_download(repo_id="comfyanonymous/flux_text_encoders", filename="clip_l.safetensors", local_dir="models/text_encoders")
+# t5_path = hf_hub_download(repo_id="comfyanonymous/flux_text_encoders", filename="t5xxl_fp16.safetensors", local_dir="models/text_encoders/t5")
 
 
 sd15_name = 'stablediffusionapi/realistic-vision-v51'
@@ -759,7 +759,9 @@ def process(input_fg, prompt, image_width, image_height, num_samples, seed, step
     input_height, input_width = input_fg.shape[:2]
 
     bg_source = BGSource(bg_source)
-
+    
+    if bg_source == BGSource.NONE:
+        pass
     if bg_source == BGSource.UPLOAD:
         pass
     elif bg_source == BGSource.UPLOAD_FLIP:
@@ -1373,8 +1375,8 @@ def add_extra_model_paths() -> None:
         print("Could not find the extra_model_paths config file.")
 
 # Initialize paths
-add_comfyui_directory_to_sys_path()
-add_extra_model_paths()
+# add_comfyui_directory_to_sys_path()
+# add_extra_model_paths()
 
 def import_custom_nodes() -> None:
     import asyncio
@@ -1388,91 +1390,93 @@ def import_custom_nodes() -> None:
     init_extra_nodes()
 
 # Import all necessary nodes
-from nodes import (
-    StyleModelLoader,
-    VAEEncode,
-    NODE_CLASS_MAPPINGS,
-    LoadImage,
-    CLIPVisionLoader,
-    SaveImage,
-    VAELoader,
-    CLIPVisionEncode,
-    DualCLIPLoader,
-    EmptyLatentImage,
-    VAEDecode,
-    UNETLoader,
-    CLIPTextEncode,
-)
+# from nodes import (
+#     StyleModelLoader,
+#     VAEEncode,
+#     NODE_CLASS_MAPPINGS,
+#     LoadImage,
+#     CLIPVisionLoader,
+#     SaveImage,
+#     VAELoader,
+#     CLIPVisionEncode,
+#     DualCLIPLoader,
+#     EmptyLatentImage,
+#     VAEDecode,
+#     UNETLoader,
+#     CLIPTextEncode,
+# )
 
 # Initialize all constant nodes and models in global context
-import_custom_nodes()
+# import_custom_nodes()
 
 # Global variables for preloaded models and constants
 #with torch.inference_mode():
     # Initialize constants
-intconstant = NODE_CLASS_MAPPINGS["INTConstant"]()
-CONST_1024 = intconstant.get_value(value=1024)
-
-# Load CLIP
-dualcliploader = DualCLIPLoader()
-CLIP_MODEL = dualcliploader.load_clip(
-    clip_name1="t5/t5xxl_fp16.safetensors",
-    clip_name2="clip_l.safetensors",
-    type="flux",
-)
-
-# Load VAE
-vaeloader = VAELoader()
-VAE_MODEL = vaeloader.load_vae(vae_name="FLUX1/ae.safetensors")
+# intconstant = NODE_CLASS_MAPPINGS["INTConstant"]()
+# CONST_1024 = intconstant.get_value(value=1024)
+
+# # Load CLIP
+# dualcliploader = DualCLIPLoader()
+# CLIP_MODEL = dualcliploader.load_clip(
+#     clip_name1="t5/t5xxl_fp16.safetensors",
+#     clip_name2="clip_l.safetensors",
+#     type="flux",
+# )
+
+# # Load VAE
+# vaeloader = VAELoader()
+# VAE_MODEL = vaeloader.load_vae(vae_name="FLUX1/ae.safetensors")
+
+# # Load UNET
+# unetloader = UNETLoader()
+# UNET_MODEL = unetloader.load_unet(
+#     unet_name="flux1-depth-dev.safetensors", weight_dtype="default"
+# )
+
+# # Load CLIP Vision
+# clipvisionloader = CLIPVisionLoader()
+# CLIP_VISION_MODEL = clipvisionloader.load_clip(
+#     clip_name="sigclip_vision_patch14_384.safetensors"
+# )
+
+# # Load Style Model
+# stylemodelloader = StyleModelLoader()
+# STYLE_MODEL = stylemodelloader.load_style_model(
+#     style_model_name="flux1-redux-dev.safetensors"
+# )
+
+# # Initialize samplers
+# ksamplerselect = NODE_CLASS_MAPPINGS["KSamplerSelect"]()
+# SAMPLER = ksamplerselect.get_sampler(sampler_name="euler")
+
+# # Initialize depth model
+# cr_clip_input_switch = NODE_CLASS_MAPPINGS["CR Clip Input Switch"]()
+# downloadandloaddepthanythingv2model = NODE_CLASS_MAPPINGS["DownloadAndLoadDepthAnythingV2Model"]()
+# DEPTH_MODEL = downloadandloaddepthanythingv2model.loadmodel(
+#     model="depth_anything_v2_vitl_fp32.safetensors"
+# )
+# cliptextencode = CLIPTextEncode()
+# loadimage = LoadImage()
+# vaeencode = VAEEncode()
+# fluxguidance = NODE_CLASS_MAPPINGS["FluxGuidance"]()
+# instructpixtopixconditioning = NODE_CLASS_MAPPINGS["InstructPixToPixConditioning"]()
+# clipvisionencode = CLIPVisionEncode()
+# stylemodelapplyadvanced = NODE_CLASS_MAPPINGS["StyleModelApplyAdvanced"]()
+# emptylatentimage = EmptyLatentImage()
+# basicguider = NODE_CLASS_MAPPINGS["BasicGuider"]()
+# basicscheduler = NODE_CLASS_MAPPINGS["BasicScheduler"]()        
+# randomnoise = NODE_CLASS_MAPPINGS["RandomNoise"]()
+# samplercustomadvanced = NODE_CLASS_MAPPINGS["SamplerCustomAdvanced"]()
+# vaedecode = VAEDecode()
+# cr_text = NODE_CLASS_MAPPINGS["CR Text"]()
+# saveimage = SaveImage()
+# getimagesizeandcount = NODE_CLASS_MAPPINGS["GetImageSizeAndCount"]()
+# depthanything_v2 = NODE_CLASS_MAPPINGS["DepthAnything_V2"]()
+# imageresize = NODE_CLASS_MAPPINGS["ImageResize+"]()
 
-# Load UNET
-unetloader = UNETLoader()
-UNET_MODEL = unetloader.load_unet(
-    unet_name="flux1-depth-dev.safetensors", weight_dtype="default"
-)
-
-# Load CLIP Vision
-clipvisionloader = CLIPVisionLoader()
-CLIP_VISION_MODEL = clipvisionloader.load_clip(
-    clip_name="sigclip_vision_patch14_384.safetensors"
-)
-
-# Load Style Model
-stylemodelloader = StyleModelLoader()
-STYLE_MODEL = stylemodelloader.load_style_model(
-    style_model_name="flux1-redux-dev.safetensors"
-)
-
-# Initialize samplers
-ksamplerselect = NODE_CLASS_MAPPINGS["KSamplerSelect"]()
-SAMPLER = ksamplerselect.get_sampler(sampler_name="euler")
-
-# Initialize depth model
-cr_clip_input_switch = NODE_CLASS_MAPPINGS["CR Clip Input Switch"]()
-downloadandloaddepthanythingv2model = NODE_CLASS_MAPPINGS["DownloadAndLoadDepthAnythingV2Model"]()
-DEPTH_MODEL = downloadandloaddepthanythingv2model.loadmodel(
-    model="depth_anything_v2_vitl_fp32.safetensors"
-)
-cliptextencode = CLIPTextEncode()
-loadimage = LoadImage()
-vaeencode = VAEEncode()
-fluxguidance = NODE_CLASS_MAPPINGS["FluxGuidance"]()
-instructpixtopixconditioning = NODE_CLASS_MAPPINGS["InstructPixToPixConditioning"]()
-clipvisionencode = CLIPVisionEncode()
-stylemodelapplyadvanced = NODE_CLASS_MAPPINGS["StyleModelApplyAdvanced"]()
-emptylatentimage = EmptyLatentImage()
-basicguider = NODE_CLASS_MAPPINGS["BasicGuider"]()
-basicscheduler = NODE_CLASS_MAPPINGS["BasicScheduler"]()        
-randomnoise = NODE_CLASS_MAPPINGS["RandomNoise"]()
-samplercustomadvanced = NODE_CLASS_MAPPINGS["SamplerCustomAdvanced"]()
-vaedecode = VAEDecode()
-cr_text = NODE_CLASS_MAPPINGS["CR Text"]()
-saveimage = SaveImage()
-getimagesizeandcount = NODE_CLASS_MAPPINGS["GetImageSizeAndCount"]()
-depthanything_v2 = NODE_CLASS_MAPPINGS["DepthAnything_V2"]()
-imageresize = NODE_CLASS_MAPPINGS["ImageResize+"]()
 
 @spaces.GPU
+@torch.inference_mode
 def generate_image(prompt, structure_image, style_image, depth_strength=15, style_strength=0.5, progress=gr.Progress(track_tqdm=True)) -> str:
     """Main generation function that processes inputs and returns the path to the generated image."""
 
@@ -1672,10 +1676,10 @@ with gr.Blocks() as app:
                         with gr.Group():
                             gr.Markdown("Outpaint")
                             with gr.Row():
-                                with gr.Column(scale=2):
-                                    prompt_fill = gr.Textbox(label="Prompt (Optional)")
-                                with gr.Column(scale=1):
-                                    fill_button = gr.Button("Generate")
+                                # with gr.Column(scale=2):
+                                #     # prompt_fill = gr.Textbox(label="Prompt (Optional)")
+                                # with gr.Column(scale=1):
+                                fill_button = gr.Button("Generate")
                             target_ratio = gr.Radio(
                                 label="Image Ratio",
                                 choices=["9:16", "16:9", "1:1", "Custom"],
@@ -1850,9 +1854,9 @@ with gr.Blocks() as app:
         ).then(
             fn=inpaint,
             inputs=[dummy_image_for_outputs, width_slider, height_slider, overlap_percentage, num_inference_steps,
-                    resize_option, custom_resize_percentage, prompt_fill, alignment_dropdown,
+                    resize_option, custom_resize_percentage, prompt, alignment_dropdown,
                     overlap_left, overlap_right, overlap_top, overlap_bottom],
-            outputs=[fill_result]).then(clear_memory, inputs=[], outputs=[])
+            outputs=[fill_result])#.then(clear_memory, inputs=[], outputs=[])
         # ).then(
         #     fn=lambda: gr.update(visible=True),
         #     inputs=None,
@@ -1866,9 +1870,9 @@ with gr.Blocks() as app:
         ).then(
             fn=inpaint,
             inputs=[dummy_image_for_outputs, width_slider, height_slider, overlap_percentage, num_inference_steps,
-                    resize_option, custom_resize_percentage, prompt_fill, alignment_dropdown,
+                    resize_option, custom_resize_percentage, prompt, alignment_dropdown,
                     overlap_left, overlap_right, overlap_top, overlap_bottom],
-            outputs=[fill_result]).then(clear_memory, inputs=[], outputs=[])
+            outputs=[fill_result])#.then(clear_memory, inputs=[], outputs=[])
 
         def convert_to_pil(image):
             try:
@@ -1905,7 +1909,7 @@ with gr.Blocks() as app:
             outputs=[extracted_fg, x_slider, y_slider]
         ).then(clear_memory, inputs=[], outputs=[])
         
-    with gr.Tab("Style Transfer"):
+    with gr.Tab("Style Transfer", visible=False):
         gr.Markdown("## Apply the style of an image to another one")
         with gr.Row():
             with gr.Column():