Create app.py

app.py

@@ -0,0 +1,189 @@
+import gradio as gr
+from diffusers import StableDiffusionPipeline
+import torch
+from PIL import Image, ImageDraw, ImageFont
+import io, zipfile
+import mediapipe as mp
+import numpy as np
+
+# ----------------------
+# Load AI model
+# ----------------------
+pipe = StableDiffusionPipeline.from_pretrained(
+    "stabilityai/stable-diffusion-2",
+    torch_dtype=torch.float16
+).to("cuda")
+
+mp_face = mp.solutions.face_detection.FaceDetection(model_selection=1, min_detection_confidence=0.5)
+
+# ----------------------
+# Helper functions
+# ----------------------
+def detect_characters(img):
+    """Detect faces in the image and return approximate (x,y) positions."""
+    img_rgb = img.convert("RGB")
+    width, height = img.size
+    results = mp_face.process(np.array(img_rgb))
+    positions = []
+    if results.detections:
+        for det in results.detections:
+            bbox = det.location_data.relative_bounding_box
+            x = int(bbox.xmin * width + bbox.width * width / 2)
+            y = int(bbox.ymin * height)
+            positions.append((x, y))
+    if not positions:
+        positions = [(50,50)]
+    return positions
+
+def add_speech_bubbles(img, bubbles, character_positions=None):
+    """Draw multiple bubbles with tails pointing to character positions."""
+    img = img.convert("RGBA")
+    draw = ImageDraw.Draw(img)
+    try:
+        font = ImageFont.truetype("ComicNeue-Bold.ttf", 24)
+        name_font = ImageFont.truetype("ComicNeue-Bold.ttf", 18)
+    except:
+        font = ImageFont.load_default()
+        name_font = ImageFont.load_default()
+
+    if not character_positions:
+        character_positions = [(50,50)]
+
+    for i, bubble in enumerate(bubbles):
+        text = bubble.get("text","")
+        char_name = bubble.get("character","")
+        x0, y0 = 20, 20 + i*120
+
+        # Wrap text
+        max_width = img.width - x0 - 20
+        lines = []
+        words = text.split()
+        current_line = ""
+        for word in words:
+            test_line = f"{current_line} {word}".strip()
+            w, h = draw.textsize(test_line, font=font)
+            if w <= max_width:
+                current_line = test_line
+            else:
+                lines.append(current_line)
+                current_line = word
+        if current_line:
+            lines.append(current_line)
+
+        line_height = font.getsize("A")[1] + 4
+        bubble_height = line_height * len(lines) + 20
+        bubble_width = max([draw.textsize(line, font=font)[0] for line in lines]) + 20
+
+        x1, y1 = x0 + bubble_width, y0 + bubble_height
+        draw.rounded_rectangle([x0, y0, x1, y1], radius=15, fill="white", outline="black", width=3)
+
+        # Tail points to closest character
+        char_x, char_y = character_positions[i % len(character_positions)]
+        tail_base_x = x0 + bubble_width // 2
+        tail_base_y = y1
+        draw.polygon([(tail_base_x-10, tail_base_y), (tail_base_x+10, tail_base_y), (char_x, char_y)], fill="white", outline="black")
+
+        # Draw text
+        y_text = y0 + 10
+        for line in lines:
+            draw.text((x0+10, y_text), line, font=font, fill="black")
+            y_text += line_height
+
+        if char_name:
+            draw.text((x0+10, y0-20), char_name, font=name_font, fill="black")
+
+    return img
+
+def generate_comic(title, story, pages, panels, art_style, theme, dialogues, explicit):
+    """Generate AI comic panels with bubbles."""
+    comic_panels = []
+    panels = int(panels)
+    panel_dialogues = dialogues.split("\n") if dialogues else [""] * panels
+
+    for i in range(panels):
+        prompt = f"{story}, {art_style} comic style, {theme} theme, colorful, detailed, panel {i+1}"
+        if explicit:
+            prompt += ", include blood, gore, offensive language, or dramatic action"
+
+        img = pipe(prompt, guidance_scale=7.5).images[0]
+        character_positions = detect_characters(img)
+
+        # Prepare bubbles
+        bubbles = []
+        if i < len(panel_dialogues) and panel_dialogues[i].strip():
+            bubble_texts = panel_dialogues[i].split("||")
+            for b in bubble_texts:
+                if ":" in b:
+                    char, text = b.split(":", 1)
+                else:
+                    char, text = "", b
+                bubbles.append({"text": text.strip(), "character": char.strip()})
+        if bubbles:
+            img = add_speech_bubbles(img, bubbles, character_positions)
+        comic_panels.append(img)
+
+    return comic_panels
+
+def create_zip(panels):
+    zip_buffer = io.BytesIO()
+    with zipfile.ZipFile(zip_buffer, "a", zipfile.ZIP_DEFLATED) as zf:
+        for i, img in enumerate(panels):
+            img_bytes = io.BytesIO()
+            img.save(img_bytes, format="PNG")
+            zf.writestr(f"panel_{i+1}.png", img_bytes.getvalue())
+    zip_buffer.seek(0)
+    return zip_buffer
+
+def generate_and_zip(title, story, pages, panels, art_style, theme, dialogues, explicit):
+    panels_list = generate_comic(title, story, pages, panels, art_style, theme, dialogues, explicit)
+    zip_file = create_zip(panels_list)
+    return panels_list, zip_file
+
+# ----------------------
+# Gradio UI
+# ----------------------
+custom_css = """
+<style>
+@import url('https://fonts.googleapis.com/css2?family=Bangers&family=Comic+Neue:wght@400;700&display=swap');
+body { background: linear-gradient(135deg, #7e22ce, #3b82f6); background-size: 400% 400%; animation: gradientBG 15s ease infinite; }
+@keyframes gradientBG { 0%{background-position:0% 50%} 50%{background-position:100% 50%} 100%{background-position:0% 50%} }
+h1, .comic-title { font-family: 'Bangers', cursive; }
+p, label, .comic-text { font-family: 'Comic Neue', cursive; }
+.comic-panel { border: 4px solid black; box-shadow: 8px 8px 0px rgba(0,0,0,0.3); border-radius: 8px; transition: transform 0.3s ease, box-shadow 0.3s ease; }
+.comic-panel:hover { transform: scale(1.05); box-shadow: 12px 12px 0px rgba(0,0,0,0.4); }
+</style>
+"""
+
+with gr.Blocks(css=custom_css) as demo:
+    gr.HTML("<h1 style='text-align:center; color:white; font-size:3em;'>AI Comic Crafter</h1>")
+    gr.HTML("<p style='text-align:center; color:white; font-size:1.2em;'>Professional multi-bubble comics with AI!</p>")
+
+    with gr.Row():
+        with gr.Column(scale=1):
+            title_input = gr.Textbox(label="Comic Book Title", placeholder="Enter your comic book title...")
+            story_input = gr.Textbox(label="Story Setup", placeholder="Describe characters, plot, etc.", lines=3)
+            pages_input = gr.Dropdown(["1","3","5","8","10"], label="Number of Pages", value="3")
+            panels_input = gr.Dropdown(["2","4","6"], label="Panels Per Page", value="4")
+            art_style_input = gr.Dropdown(["Comic","Anime","Manga","Noir","Realistic"], label="Art Style", value="Comic")
+            theme_input = gr.Dropdown(
+                ["Sci-Fi","Fantasy","Superhero","Horror","Noir","Cyberpunk","Mystery","Adventure"],
+                label="Theme", value="Sci-Fi"
+            )
+            dialogues_input = gr.Textbox(
+                label="Custom Dialogues (one line per panel, bubbles separated by '||', format: Character:Text)",
+                lines=6
+            )
+            explicit_input = gr.Checkbox(label="Allow explicit content (blood/gore/offensive language)")
+            generate_btn = gr.Button("Generate Comic Book")
+            download_btn = gr.File(label="Download Comic as ZIP")
+        
+        with gr.Column(scale=1):
+            output_gallery = gr.Gallery(label="Your Comic Preview").style(grid=[2], height="auto")
+
+    generate_btn.click(
+        generate_and_zip,
+        inputs=[title_input, story_input, pages_input, panels_input, art_style_input, theme_input, dialogues_input, explicit_input],
+        outputs=[output_gallery, download_btn]
+    )
+
+demo.launch()