Update app.py

app.py

@@ -2,172 +2,258 @@ import gradio as gr
 import spaces
 import torch
 from diffusers import AutoencoderKL, TCDScheduler
-# (Assume ControlNet manual load or from_pretrained is already working)
-from controlnet_union import ControlNetModel_Union
-from pipeline_fill_sd_xl import StableDiffusionXLFillPipeline
+from diffusers.models.model_loading_utils import load_state_dict
 from gradio_imageslider import ImageSlider
 from huggingface_hub import hf_hub_download
 
+from controlnet_union import ControlNetModel_Union
+from pipeline_fill_sd_xl import StableDiffusionXLFillPipeline
+
 from PIL import Image, ImageDraw
 import numpy as np
 
-# --- Load ControlNet and SDXL Fill Pipeline ---
-# (Either manual download or via from_pretrained)
-controlnet_model = ControlNetModel_Union.from_pretrained(
+config_file = hf_hub_download(
     "xinsir/controlnet-union-sdxl-1.0",
-    torch_dtype=torch.float16,
-    variant="fp16"
-).to("cuda")
+    filename="config_promax.json",
+)
+
+config = ControlNetModel_Union.load_config(config_file)
+controlnet_model = ControlNetModel_Union.from_config(config)
+model_file = hf_hub_download(
+    "xinsir/controlnet-union-sdxl-1.0",
+    filename="diffusion_pytorch_model_promax.safetensors",
+)
+
+sstate_dict = load_state_dict(model_file)
+model, _, _, _, _ = ControlNetModel_Union._load_pretrained_model(
+    controlnet_model, sstate_dict, model_file, "xinsir/controlnet-union-sdxl-1.0"
+)
+model.to(device="cuda", dtype=torch.float16)
+#----------------------
 
 vae = AutoencoderKL.from_pretrained(
-    "madebyollin/sdxl-vae-fp16-fix",
-    torch_dtype=torch.float16
+    "madebyollin/sdxl-vae-fp16-fix", torch_dtype=torch.float16
 ).to("cuda")
 
 pipe = StableDiffusionXLFillPipeline.from_pretrained(
     "SG161222/RealVisXL_V5.0_Lightning",
     torch_dtype=torch.float16,
     vae=vae,
-    controlnet=controlnet_model,
+    controlnet=model,
     variant="fp16",
 ).to("cuda")
+
 pipe.scheduler = TCDScheduler.from_config(pipe.scheduler.config)
 
-# --- Utility functions ---
 def can_expand(source_width, source_height, target_width, target_height, alignment):
+    """Checks if the image can be expanded based on the alignment."""
     if alignment in ("Left", "Right") and source_width >= target_width:
         return False
     if alignment in ("Top", "Bottom") and source_height >= target_height:
         return False
     return True
 
+def prepare_image_and_mask(image, width, height, overlap_percentage, resize_option, custom_resize_percentage, alignment, overlap_left, overlap_right, overlap_top, overlap_bottom):
+    target_size = (width, height)
+
+    # Calculate the scaling factor to fit the image within the target size
+    scale_factor = min(target_size[0] / image.width, target_size[1] / image.height)
+    new_width = int(image.width * scale_factor)
+    new_height = int(image.height * scale_factor)
+    
+    # Resize the source image to fit within target size
+    source = image.resize((new_width, new_height), Image.LANCZOS)
+
+    # Apply resize option using percentages
+    if resize_option == "Full":
+        resize_percentage = 100
+    elif resize_option == "50%":
+        resize_percentage = 50
+    elif resize_option == "33%":
+        resize_percentage = 33
+    elif resize_option == "25%":
+        resize_percentage = 25
+    else:  # Custom
+        resize_percentage = custom_resize_percentage
+
+    # Calculate new dimensions based on percentage
+    resize_factor = resize_percentage / 100
+    new_width = int(source.width * resize_factor)
+    new_height = int(source.height * resize_factor)
+
+    # Ensure minimum size of 64 pixels
+    new_width = max(new_width, 64)
+    new_height = max(new_height, 64)
+
+    # Resize the image
+    source = source.resize((new_width, new_height), Image.LANCZOS)
+
+    # Calculate the overlap in pixels based on the percentage
+    overlap_x = int(new_width * (overlap_percentage / 100))
+    overlap_y = int(new_height * (overlap_percentage / 100))
+
+    # Ensure minimum overlap of 1 pixel
+    overlap_x = max(overlap_x, 1)
+    overlap_y = max(overlap_y, 1)
+
+    # Calculate margins based on alignment
+    if alignment == "Middle":
+        margin_x = (target_size[0] - new_width) // 2
+        margin_y = (target_size[1] - new_height) // 2
+    elif alignment == "Left":
+        margin_x = 0
+        margin_y = (target_size[1] - new_height) // 2
+    elif alignment == "Right":
+        margin_x = target_size[0] - new_width
+        margin_y = (target_size[1] - new_height) // 2
+    elif alignment == "Top":
+        margin_x = (target_size[0] - new_width) // 2
+        margin_y = 0
+    elif alignment == "Bottom":
+        margin_x = (target_size[0] - new_width) // 2
+        margin_y = target_size[1] - new_height
+
+    # Adjust margins to eliminate gaps
+    margin_x = max(0, min(margin_x, target_size[0] - new_width))
+    margin_y = max(0, min(margin_y, target_size[1] - new_height))
+
+    # Create a new background image and paste the resized source image
+    background = Image.new('RGB', target_size, (255, 255, 255))
+    background.paste(source, (margin_x, margin_y))
+
+    # Create the mask
+    mask = Image.new('L', target_size, 255)
+    mask_draw = ImageDraw.Draw(mask)
+
+    # Calculate overlap areas
+    white_gaps_patch = 2
+
+    left_overlap = margin_x + overlap_x if overlap_left else margin_x + white_gaps_patch
+    right_overlap = margin_x + new_width - overlap_x if overlap_right else margin_x + new_width - white_gaps_patch
+    top_overlap = margin_y + overlap_y if overlap_top else margin_y + white_gaps_patch
+    bottom_overlap = margin_y + new_height - overlap_y if overlap_bottom else margin_y + new_height - white_gaps_patch
+    
+    if alignment == "Left":
+        left_overlap = margin_x + overlap_x if overlap_left else margin_x
+    elif alignment == "Right":
+        right_overlap = margin_x + new_width - overlap_x if overlap_right else margin_x + new_width
+    elif alignment == "Top":
+        top_overlap = margin_y + overlap_y if overlap_top else margin_y
+    elif alignment == "Bottom":
+        bottom_overlap = margin_y + new_height - overlap_y if overlap_bottom else margin_y + new_height
+
+
+    # Draw the mask
+    mask_draw.rectangle([
+        (left_overlap, top_overlap),
+        (right_overlap, bottom_overlap)
+    ], fill=0)
+
+    return background, mask
+
+def preview_image_and_mask(image, width, height, overlap_percentage, resize_option, custom_resize_percentage, alignment, overlap_left, overlap_right, overlap_top, overlap_bottom):
+    background, mask = prepare_image_and_mask(image, width, height, overlap_percentage, resize_option, custom_resize_percentage, alignment, overlap_left, overlap_right, overlap_top, overlap_bottom)
+    
+    # Create a preview image showing the mask
+    preview = background.copy().convert('RGBA')
+    
+    # Create a semi-transparent red overlay
+    red_overlay = Image.new('RGBA', background.size, (255, 0, 0, 64))  # Reduced alpha to 64 (25% opacity)
+    
+    # Convert black pixels in the mask to semi-transparent red
+    red_mask = Image.new('RGBA', background.size, (0, 0, 0, 0))
+    red_mask.paste(red_overlay, (0, 0), mask)
+    
+    # Overlay the red mask on the background
+    preview = Image.alpha_composite(preview, red_mask)
+    
+    return preview
 
-def prepare_image_and_mask(image, width, height, overlap_percentage,
-                           resize_option, custom_resize_percentage,
-                           alignment, overlap_left, overlap_right,
-                           overlap_top, overlap_bottom):
-    target = (width, height)
-    scale = min(target[0] / image.width, target[1] / image.height)
-    w, h = int(image.width * scale), int(image.height * scale)
-    src = image.resize((w, h), Image.LANCZOS)
-
-    # Resize percentage
-    if resize_option == "Full": pct = 100
-    elif resize_option == "50%": pct = 50
-    elif resize_option == "33%": pct = 33
-    elif resize_option == "25%": pct = 25
-    else: pct = custom_resize_percentage
-
-    rw, rh = max(int(src.width * pct / 100), 64), max(int(src.height * pct / 100), 64)
-    src = src.resize((rw, rh), Image.LANCZOS)
-
-    ox = max(int(rw * overlap_percentage / 100), 1)
-    oy = max(int(rh * overlap_percentage / 100), 1)
-
-    # Margins
-    if alignment == "Middle": mx, my = (width - rw)//2, (height - rh)//2
-    elif alignment == "Left": mx, my = 0, (height - rh)//2
-    elif alignment == "Right": mx, my = width - rw, (height - rh)//2
-    elif alignment == "Top": mx, my = (width - rw)//2, 0
-    else: mx, my = (width - rw)//2, height - rh
-
-    mx, my = max(0, min(mx, width - rw)), max(0, min(my, height - rh))
-
-    bg = Image.new("RGB", target, (255,255,255))
-    bg.paste(src, (mx, my))
-
-    mask = Image.new("L", target, 255)
-    d = ImageDraw.Draw(mask)
-
-    lx = mx + (ox if overlap_left else 2)
-    rx = mx + rw - (ox if overlap_right else 2)
-    ty = my + (oy if overlap_top else 2)
-    by = my + rh - (oy if overlap_bottom else 2)
-
-    # Edge adjustments
-    if alignment == "Left": lx = mx + (ox if overlap_left else 0)
-    if alignment == "Right": rx = mx + rw - (ox if overlap_right else 0)
-    if alignment == "Top": ty = my + (oy if overlap_top else 0)
-    if alignment == "Bottom": by = my + rh - (oy if overlap_bottom else 0)
-
-    d.rectangle([(lx, ty), (rx, by)], fill=0)
-    return bg, mask
-
-
-def preview_image_and_mask(*args):
-    bg, mask = prepare_image_and_mask(*args)
-    vis = bg.copy().convert("RGBA")
-    red = Image.new("RGBA", bg.size, (255,0,0,64))
-    overlay = Image.new("RGBA", bg.size, (0,0,0,0))
-    overlay.paste(red, (0,0), mask)
-    return Image.alpha_composite(vis, overlay)
-
-# --- Fixed infer: return list for slider ---
 @spaces.GPU(duration=24)
-def infer(image, width, height, overlap_percentage, num_inference_steps,
-          resize_option, custom_resize_percentage, prompt_input,
-          alignment, overlap_left, overlap_right, overlap_top, overlap_bottom):
-    background, mask = prepare_image_and_mask(
-        image, width, height, overlap_percentage,
-        resize_option, custom_resize_percentage,
-        alignment, overlap_left, overlap_right,
-        overlap_top, overlap_bottom
-    )
+def infer(image, width, height, overlap_percentage, num_inference_steps, resize_option, custom_resize_percentage, prompt_input, alignment, overlap_left, overlap_right, overlap_top, overlap_bottom):
+    background, mask = prepare_image_and_mask(image, width, height, overlap_percentage, resize_option, custom_resize_percentage, alignment, overlap_left, overlap_right, overlap_top, overlap_bottom)
+    
     if not can_expand(background.width, background.height, width, height, alignment):
         alignment = "Middle"
 
-    hole = background.copy()
-    hole.paste(0, (0,0), mask)
+    cnet_image = background.copy()
+    cnet_image.paste(0, (0, 0), mask)
 
     final_prompt = f"{prompt_input} , high quality, 4k"
-    embeds = pipe.encode_prompt(final_prompt, "cuda", True)
-
-    # Run pipeline and grab last frame
-    gen = pipe(
-        prompt_embeds=embeds[0],
-        negative_prompt_embeds=embeds[1],
-        pooled_prompt_embeds=embeds[2],
-        negative_pooled_prompt_embeds=embeds[3],
-        image=hole,
-        num_inference_steps=num_inference_steps
-    )
-    last = None
-    for img in gen:
-        last = img
 
-    out = last.convert("RGBA")
-    hole.paste(out, (0,0), mask)
+    (
+        prompt_embeds,
+        negative_prompt_embeds,
+        pooled_prompt_embeds,
+        negative_pooled_prompt_embeds,
+    ) = pipe.encode_prompt(final_prompt, "cuda", True)
+
+    for image in pipe(
+        prompt_embeds=prompt_embeds,
+        negative_prompt_embeds=negative_prompt_embeds,
+        pooled_prompt_embeds=pooled_prompt_embeds,
+        negative_pooled_prompt_embeds=negative_pooled_prompt_embeds,
+        image=cnet_image,
+        num_inference_steps=num_inference_steps
+    ):
+        yield cnet_image, image
 
-    # Return a list: [input_hole_image, final_output]
-    return [background, hole]
+    image = image.convert("RGBA")
+    cnet_image.paste(image, (0, 0), mask)
 
+    yield background, cnet_image
 
 def clear_result():
+    """Clears the result ImageSlider."""
     return gr.update(value=None)
 
-def preload_presets(ratio, w, h):
-    if ratio == "9:16": return 720, 1280, gr.update()
-    if ratio == "16:9": return 1280, 720, gr.update()
-    if ratio == "1:1": return 1024, 1024, gr.update()
-    return w, h, gr.update(open=True)
-
-def select_the_right_preset(w, h):
-    if (w,h) == (720,1280): return "9:16"
-    if (w,h) == (1280,720): return "16:9"
-    if (w,h) == (1024,1024): return "1:1"
-    return "Custom"
-
-def toggle_custom_resize_slider(opt):
-    return gr.update(visible=(opt=="Custom"))
-
-def update_history(img, history):
-    history = history or []
-    history.insert(0, img)
+def preload_presets(target_ratio, ui_width, ui_height):
+    """Updates the width and height sliders based on the selected aspect ratio."""
+    if target_ratio == "9:16":
+        changed_width = 720
+        changed_height = 1280
+        return changed_width, changed_height, gr.update()
+    elif target_ratio == "16:9":
+        changed_width = 1280
+        changed_height = 720
+        return changed_width, changed_height, gr.update()
+    elif target_ratio == "1:1":
+        changed_width = 1024
+        changed_height = 1024
+        return changed_width, changed_height, gr.update()
+    elif target_ratio == "Custom":
+        return ui_width, ui_height, gr.update(open=True)
+
+def select_the_right_preset(user_width, user_height):
+    if user_width == 720 and user_height == 1280:
+        return "9:16"
+    elif user_width == 1280 and user_height == 720:
+        return "16:9"
+    elif user_width == 1024 and user_height == 1024:
+        return "1:1"
+    else:
+        return "Custom"
+
+def toggle_custom_resize_slider(resize_option):
+    return gr.update(visible=(resize_option == "Custom"))
+
+def update_history(new_image, history):
+    """Updates the history gallery with the new image."""
+    if history is None:
+        history = []
+    history.insert(0, new_image)
     return history
 
-css = ".gradio-container { width: 1200px !important; }"
-title = "<h1 align='center'>Diffusers Image Outpaint Lightning</h1>"
+css = """
+.gradio-container {
+    width: 1200px !important;
+}
+h1 { text-align: center; }
+footer { visibility: hidden; }
+"""
 
+title = """<h1 align="center">Diffusers Image Outpaint Lightning</h1>
+"""
 with gr.Blocks(css=css) as demo:
     gr.HTML(title)
     with gr.Row():