Update to script-to-shots app with FLUX model

app.py

@@ -1,154 +1,321 @@
 import gradio as gr
-import numpy as np
-import random
-
-# import spaces #[uncomment to use ZeroGPU]
-from diffusers import DiffusionPipeline
-import torch
-
-device = "cuda" if torch.cuda.is_available() else "cpu"
-model_repo_id = "stabilityai/sdxl-turbo"  # Replace to the model you would like to use
-
-if torch.cuda.is_available():
-    torch_dtype = torch.float16
-else:
-    torch_dtype = torch.float32
-
-pipe = DiffusionPipeline.from_pretrained(model_repo_id, torch_dtype=torch_dtype)
-pipe = pipe.to(device)
-
-MAX_SEED = np.iinfo(np.int32).max
-MAX_IMAGE_SIZE = 1024
-
-
-# @spaces.GPU #[uncomment to use ZeroGPU]
-def infer(
-    prompt,
-    negative_prompt,
-    seed,
-    randomize_seed,
-    width,
-    height,
-    guidance_scale,
-    num_inference_steps,
-    progress=gr.Progress(track_tqdm=True),
-):
-    if randomize_seed:
-        seed = random.randint(0, MAX_SEED)
-
-    generator = torch.Generator().manual_seed(seed)
-
-    image = pipe(
-        prompt=prompt,
-        negative_prompt=negative_prompt,
-        guidance_scale=guidance_scale,
-        num_inference_steps=num_inference_steps,
-        width=width,
-        height=height,
-        generator=generator,
-    ).images[0]
-
-    return image, seed
-
-
-examples = [
-    "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k",
-    "An astronaut riding a green horse",
-    "A delicious ceviche cheesecake slice",
-]
-
-css = """
-#col-container {
-    margin: 0 auto;
-    max-width: 640px;
-}
+import requests
+import json
+import re
+from typing import List, Dict, Any
+import os
+
+# Hugging Face configuration
+HF_TOKEN = os.getenv("HUGGING_FACE_API_TOKEN", "")
+HF_API_URL = "https://api-inference.huggingface.co/models/black-forest-labs/FLUX.1-dev"
+
+def parse_script(script_text: str) -> Dict[str, Any]:
+    """Parse script text and extract scenes and characters"""
+    lines = script_text.strip().split('\n')
+    scenes = []
+    characters = set()
+    current_scene = None
+    
+    for line in lines:
+        line = line.strip()
+        if not line:
+            continue
+            
+        # Scene headers (INT./EXT.)
+        if line.upper().startswith(('INT.', 'EXT.', 'SCENE')):
+            if current_scene:
+                scenes.append(current_scene)
+            current_scene = {
+                'location': line,
+                'dialogue': [],
+                'action': []
+            }
+        
+        # Character dialogue (ALL CAPS followed by dialogue)
+        elif line.isupper() and len(line.split()) <= 3 and current_scene:
+            characters.add(line)
+            current_scene['dialogue'].append({'character': line, 'lines': []})
+        
+        # Dialogue lines
+        elif current_scene and current_scene['dialogue'] and not line.isupper():
+            current_scene['dialogue'][-1]['lines'].append(line)
+        
+        # Action lines
+        elif current_scene and not line.isupper():
+            current_scene['action'].append(line)
+    
+    if current_scene:
+        scenes.append(current_scene)
+    
+    return {
+        'scenes': scenes,
+        'characters': list(characters),
+        'total_scenes': len(scenes)
+    }
+
+def generate_shot_list(script_data: Dict[str, Any]) -> List[Dict[str, Any]]:
+    """Generate shot list from parsed script"""
+    shots = []
+    shot_id = 1
+    
+    for scene_idx, scene in enumerate(script_data['scenes']):
+        # Establishing shot
+        shots.append({
+            'id': shot_id,
+            'type': 'Establishing Shot',
+            'description': f"Wide shot of {scene['location']}",
+            'scene': scene_idx + 1,
+            'location': scene['location']
+        })
+        shot_id += 1
+        
+        # Character shots
+        dialogue_chars = set()
+        for dialogue in scene['dialogue']:
+            char = dialogue['character']
+            if char not in dialogue_chars:
+                shots.append({
+                    'id': shot_id,
+                    'type': 'Medium Shot',
+                    'description': f"Medium shot of {char}",
+                    'scene': scene_idx + 1,
+                    'character': char,
+                    'location': scene['location']
+                })
+                dialogue_chars.add(char)
+                shot_id += 1
+        
+        # Action shots
+        for action in scene['action']:
+            if len(action) > 20:  # Only significant action lines
+                shots.append({
+                    'id': shot_id,
+                    'type': 'Action Shot',
+                    'description': action[:100] + "..." if len(action) > 100 else action,
+                    'scene': scene_idx + 1,
+                    'location': scene['location']
+                })
+                shot_id += 1
+    
+    return shots
+
+def generate_image(prompt: str) -> str:
+    """Generate image using Hugging Face API"""
+    if not HF_TOKEN:
+        return "https://via.placeholder.com/512x512?text=No+API+Key"
+    
+    headers = {
+        "Authorization": f"Bearer {HF_TOKEN}",
+        "Content-Type": "application/json"
+    }
+    
+    payload = {
+        "inputs": f"{prompt}, cinematic, professional, high quality"
+    }
+    
+    try:
+        response = requests.post(HF_API_URL, headers=headers, json=payload, timeout=30)
+        if response.status_code == 200:
+            # Save image and return path
+            import base64
+            image_data = response.content
+            image_b64 = base64.b64encode(image_data).decode()
+            return f"data:image/png;base64,{image_b64}"
+        else:
+            return f"https://via.placeholder.com/512x512?text=API+Error+{response.status_code}"
+    except Exception as e:
+        return f"https://via.placeholder.com/512x512?text=Error"
+
+def process_script(script_text: str, generate_images: bool = True):
+    """Main function to process script and generate shot list"""
+    if not script_text.strip():
+        return "Please enter a script.", "", ""
+    
+    # Parse script
+    script_data = parse_script(script_text)
+    
+    # Generate shot list
+    shots = generate_shot_list(script_data)
+    
+    # Create summary
+    summary = f"""
+## Script Analysis Summary
+- **Total Scenes:** {script_data['total_scenes']}
+- **Characters:** {', '.join(script_data['characters'])}
+- **Generated Shots:** {len(shots)}
 """
+    
+    # Create shot list display
+    shot_list_html = "<div style='max-height: 400px; overflow-y: auto;'>"
+    
+    for shot in shots:
+        # Generate image if requested
+        image_html = ""
+        if generate_images:
+            image_url = generate_image(shot['description'])
+            image_html = f'<img src="{image_url}" style="width: 200px; height: 150px; object-fit: cover; border-radius: 8px;" />'
+        
+        shot_list_html += f"""
+        <div style='border: 1px solid #ddd; margin: 10px 0; padding: 15px; border-radius: 8px; background: #f9f9f9;'>
+            <div style='display: flex; gap: 15px; align-items: flex-start;'>
+                <div style='flex: 1;'>
+                    <h3 style='margin: 0 0 10px 0; color: #333;'>Shot {shot['id']}: {shot['type']}</h3>
+                    <p style='margin: 5px 0;'><strong>Scene:</strong> {shot['scene']}</p>
+                    <p style='margin: 5px 0;'><strong>Description:</strong> {shot['description']}</p>
+                    {f"<p style='margin: 5px 0;'><strong>Character:</strong> {shot.get('character', 'N/A')}</p>" if shot.get('character') else ""}
+                    <p style='margin: 5px 0;'><strong>Location:</strong> {shot.get('location', 'N/A')}</p>
+                </div>
+                <div style='flex-shrink: 0;'>
+                    {image_html}
+                </div>
+            </div>
+        </div>
+        """
+    
+    shot_list_html += "</div>"
+    
+    return summary, shot_list_html, f"Generated {len(shots)} shots successfully!"
 
-with gr.Blocks(css=css) as demo:
-    with gr.Column(elem_id="col-container"):
-        gr.Markdown(" # Text-to-Image Gradio Template")
-
-        with gr.Row():
-            prompt = gr.Text(
-                label="Prompt",
-                show_label=False,
-                max_lines=1,
-                placeholder="Enter your prompt",
-                container=False,
-            )
+# Sample script for demo
+SAMPLE_SCRIPT = """INT. COFFEE SHOP - DAY
 
-            run_button = gr.Button("Run", scale=0, variant="primary")
+A bustling coffee shop filled with the morning crowd. Steam rises from espresso machines.
 
-        result = gr.Image(label="Result", show_label=False)
+SARAH sits at a corner table, typing furiously on her laptop. She glances at her watch nervously.
 
-        with gr.Accordion("Advanced Settings", open=False):
-            negative_prompt = gr.Text(
-                label="Negative prompt",
-                max_lines=1,
-                placeholder="Enter a negative prompt",
-                visible=False,
-            )
+SARAH
+(muttering to herself)
+Come on, come on... where is he?
 
-            seed = gr.Slider(
-                label="Seed",
-                minimum=0,
-                maximum=MAX_SEED,
-                step=1,
-                value=0,
-            )
+The door chimes as MIKE enters, scanning the room. He spots Sarah and approaches.
 
-            randomize_seed = gr.Checkbox(label="Randomize seed", value=True)
+MIKE
+Sorry I'm late! Traffic was insane.
 
-            with gr.Row():
-                width = gr.Slider(
-                    label="Width",
-                    minimum=256,
-                    maximum=MAX_IMAGE_SIZE,
-                    step=32,
-                    value=1024,  # Replace with defaults that work for your model
-                )
+SARAH
+(relieved)
+Thank god you're here. I've been going crazy waiting.
 
-                height = gr.Slider(
-                    label="Height",
-                    minimum=256,
-                    maximum=MAX_IMAGE_SIZE,
-                    step=32,
-                    value=1024,  # Replace with defaults that work for your model
-                )
+Mike sits down across from her.
+
+MIKE
+So, what's this big emergency about?
+
+Sarah closes her laptop and leans in conspiratorially.
+
+SARAH
+I found something. Something that could change everything.
+
+EXT. CITY STREET - DAY
 
+Sarah and Mike walk quickly down a busy sidewalk, weaving through pedestrians.
+
+MIKE
+Are you sure about this?
+
+SARAH
+I've never been more sure of anything in my life.
+
+They stop at a red light, looking around nervously."""
+
+# Create Gradio interface
+with gr.Blocks(title="Script to Shots - AI Storyboard Generator") as demo:
+    gr.Markdown("""
+    # 🎬 Script to Shots - AI Storyboard Generator
+    
+    Transform your scripts into visual shot lists with AI-generated reference images!
+    
+    **How it works:**
+    1. Paste your script in the text area below
+    2. Choose whether to generate AI images
+    3. Get an automated shot list with visual references
+    """)
+    
+    with gr.Row():
+        with gr.Column(scale=1):
+            script_input = gr.Textbox(
+                label="Script Text",
+                placeholder="Paste your script here...",
+                lines=15,
+                value=SAMPLE_SCRIPT
+            )
+            
             with gr.Row():
-                guidance_scale = gr.Slider(
-                    label="Guidance scale",
-                    minimum=0.0,
-                    maximum=10.0,
-                    step=0.1,
-                    value=0.0,  # Replace with defaults that work for your model
+                generate_images_checkbox = gr.Checkbox(
+                    label="Generate AI Images", 
+                    value=True,
+                    info="Generate visual references (requires API key)"
                 )
+                process_btn = gr.Button("Generate Shot List", variant="primary")
+        
+        with gr.Column(scale=1):
+            summary_output = gr.Markdown(label="Analysis Summary")
+            status_output = gr.Textbox(label="Status", interactive=False)
+    
+    with gr.Row():
+        shot_list_output = gr.HTML(label="Generated Shot List")
+    
+    # Example scripts
+    gr.Markdown("### 📝 Example Scripts")
+    with gr.Row():
+        example_btn1 = gr.Button("Coffee Shop Scene")
+        example_btn2 = gr.Button("Action Sequence")
+        example_btn3 = gr.Button("Clear Script")
+    
+    # Event handlers
+    process_btn.click(
+        fn=process_script,
+        inputs=[script_input, generate_images_checkbox],
+        outputs=[summary_output, shot_list_output, status_output]
+    )
+    
+    example_btn1.click(
+        fn=lambda: SAMPLE_SCRIPT,
+        outputs=script_input
+    )
+    
+    example_btn2.click(
+        fn=lambda: """EXT. ROOFTOP - NIGHT
 
-                num_inference_steps = gr.Slider(
-                    label="Number of inference steps",
-                    minimum=1,
-                    maximum=50,
-                    step=1,
-                    value=2,  # Replace with defaults that work for your model
-                )
+Rain pours down on the city skyline. Lightning illuminates the darkness.
 
-        gr.Examples(examples=examples, inputs=[prompt])
-    gr.on(
-        triggers=[run_button.click, prompt.submit],
-        fn=infer,
-        inputs=[
-            prompt,
-            negative_prompt,
-            seed,
-            randomize_seed,
-            width,
-            height,
-            guidance_scale,
-            num_inference_steps,
-        ],
-        outputs=[result, seed],
+ALEX crouches behind an air conditioning unit, breathing heavily.
+
+ALEX
+(into radio)
+I'm in position. Do you see them?
+
+VOICE (V.O.)
+(filtered)
+Two guards on the east side. Move now!
+
+Alex sprints across the rooftop, water splashing with each step.
+
+Suddenly, a spotlight sweeps across the roof. Alex dives behind a chimney just in time.
+
+GUARD
+(shouting)
+There! On the roof!
+
+Gunshots ring out. Alex pulls out a grappling hook and fires it toward the next building.""",
+        outputs=script_input
+    )
+    
+    example_btn3.click(
+        fn=lambda: "",
+        outputs=script_input
     )
+    
+    gr.Markdown("""
+    ### 🔧 Setup Instructions
+    
+    To enable AI image generation, you need a Hugging Face API token:
+    1. Get a free token at [huggingface.co/settings/tokens](https://huggingface.co/settings/tokens)
+    2. Set it as an environment variable: `HUGGING_FACE_API_TOKEN`
+    3. Restart the application
+    
+    **Note:** Without an API token, placeholder images will be shown instead.
+    """)
 
 if __name__ == "__main__":
-    demo.launch()
+    demo.launch()