enhanced_motion_examples.py

#!/usr/bin/env python3
"""
Enhanced Motion Video Generation Examples

This script demonstrates the enhanced motion capabilities of the Runway processor
with different motion intensity levels: subtle, moderate, dynamic, and cinematic.
"""

import json
import requests
import time
from typing import Dict, Any

def test_motion_intensity_levels():
    """Test video generation with different motion intensity levels."""

    # Base URL for the API
    base_url = "http://localhost:8000"

    # Test scenarios with different motion intensities
    test_scenarios = [
        {
            "name": "Ocean Scene - Subtle Motion",
            "image_prompt": "A peaceful ocean view with gentle waves at sunset",
            "motion_intensity": "subtle",
            "expected_motion": "Gentle ripples, soft movement, slow pan"
        },
        {
            "name": "Ocean Scene - Moderate Motion",
            "image_prompt": "A peaceful ocean view with gentle waves at sunset",
            "motion_intensity": "moderate",
            "expected_motion": "Flowing water, natural movement, smooth tracking"
        },
        {
            "name": "Ocean Scene - Dynamic Motion",
            "image_prompt": "A peaceful ocean view with gentle waves at sunset",
            "motion_intensity": "dynamic",
            "expected_motion": "Energetic waves, rhythmic motion, dynamic sweep"
        },
        {
            "name": "Ocean Scene - Cinematic Motion",
            "image_prompt": "A peaceful ocean view with gentle waves at sunset",
            "motion_intensity": "cinematic",
            "expected_motion": "Dramatic waves, sweeping motion, cinematic movement"
        },
        {
            "name": "Forest Scene - Cinematic Motion",
            "image_prompt": "A magical forest with tall trees and dappled sunlight",
            "motion_intensity": "cinematic",
            "expected_motion": "Majestic trees swaying, leaves cascading like rain, cinematic movement through forest canopy"
        },
        {
            "name": "City Scene - Dynamic Motion",
            "image_prompt": "A modern city skyline with glass buildings reflecting sunlight",
            "motion_intensity": "dynamic",
            "expected_motion": "Dynamic camera sweeping through scene, shadows moving, glass reflections shifting"
        }
    ]

    print("🎬 Enhanced Motion Video Generation Test")
    print("=" * 60)
    print()

    results = []

    for i, scenario in enumerate(test_scenarios, 1):
        print(f"📋 Test {i}: {scenario['name']}")
        print(f"   Image Prompt: {scenario['image_prompt']}")
        print(f"   Motion Intensity: {scenario['motion_intensity']}")
        print(f"   Expected Motion: {scenario['expected_motion']}")
        print()

        # Test with preview mode first to see the generated prompts
        print("   🔍 Preview Mode - Checking generated prompts...")
        preview_result = test_prompt_to_video_preview(
            image_prompt=scenario["image_prompt"],
            motion_intensity=scenario["motion_intensity"]
        )

        if preview_result:
            print("   ✅ Preview generated successfully")
            # Extract the video prompt from preview
            prompts = preview_result.get("prompts", {})
            video_gen = prompts.get("2_video_generation", {})
            final_prompt = video_gen.get("sanitized_prompt", "")
            print(f"   📝 Generated Video Prompt: {final_prompt}")

            # Add preview result to our results
            results.append({
                "scenario": scenario["name"],
                "motion_intensity": scenario["motion_intensity"],
                "preview_result": preview_result,
                "generated_prompt": final_prompt
            })
        else:
            print("   ❌ Preview failed")

        print()
        print("-" * 60)
        print()

    # Save results to file
    timestamp = int(time.time())
    results_file = f"enhanced_motion_test_results_{timestamp}.json"

    with open(results_file, 'w') as f:
        json.dump(results, f, indent=2, ensure_ascii=False)

    print(f"📄 Results saved to: {results_file}")

    # Print summary
    print("\n📊 Test Summary:")
    print("=" * 60)

    for result in results:
        print(f"• {result['scenario']} ({result['motion_intensity']})")
        prompt = result.get('generated_prompt', '')
        if prompt:
            # Extract motion keywords from the prompt
            motion_keywords = extract_motion_keywords(prompt)
            print(f"  Motion Keywords: {', '.join(motion_keywords)}")
        print()

    return results

def test_prompt_to_video_preview(image_prompt: str, motion_intensity: str = "dynamic") -> Dict[Any, Any]:
    """Test the preview mode for prompt-to-video generation."""
    try:
        payload = {
            "image_prompt": image_prompt,
            "motion_intensity": motion_intensity,
            "duration": 5,
            "ratio": "1280:720",
            "preview_only": True
        }

        response = requests.post("http://localhost:8000/generate-video-from-prompt", json=payload)

        if response.status_code == 200:
            result_data = response.json()

            # Extract the preview data from the result
            if hasattr(result_data, 'get'):
                # This is likely the VideoGenerationResult structure
                # We need to parse the actual preview from the response
                print("   ⚠️ Note: Getting result in API response format, may need to extract preview data")
                return result_data
            else:
                # Direct preview data
                return result_data
        else:
            print(f"   ❌ API request failed: {response.status_code}")
            print(f"   Error: {response.text}")
            return None

    except Exception as e:
        print(f"   ❌ Request failed: {e}")
        return None

def extract_motion_keywords(prompt: str) -> list:
    """Extract motion-related keywords from a video prompt."""
    motion_keywords = [
        "swaying", "dancing", "flowing", "moving", "shifting", "drifting",
        "cascading", "sweeping", "racing", "rippling", "billowing", "streaming",
        "camera", "cinematic", "dynamic", "gentle", "dramatic", "epic",
        "waves", "wind", "breeze", "motion", "movement", "animation"
    ]

    found_keywords = []
    prompt_lower = prompt.lower()

    for keyword in motion_keywords:
        if keyword in prompt_lower:
            found_keywords.append(keyword)

    return found_keywords

def test_api_health():
    """Check if the API server is running and healthy."""
    try:
        response = requests.get("http://localhost:8000/health")
        if response.status_code == 200:
            print("✅ API server is healthy")
            return True
        else:
            print(f"⚠️ API server returned status: {response.status_code}")
            return False
    except requests.exceptions.ConnectionError:
        print("❌ Cannot connect to API server at http://localhost:8000")
        print("   Please start the server with: python run_api.py")
        return False
    except Exception as e:
        print(f"❌ API health check failed: {e}")
        return False

def demonstrate_motion_comparison():
    """Demonstrate the differences between motion intensity levels."""
    print("\n🎭 Motion Intensity Comparison")
    print("=" * 60)

    base_prompt = "A serene mountain landscape with clouds and trees"

    intensities = ["subtle", "moderate", "dynamic", "cinematic"]

    print(f"Base Scene: {base_prompt}")
    print()

    for intensity in intensities:
        print(f"🎬 {intensity.upper()} Motion:")

        # This would call our enhanced motion generation function directly
        # For demonstration, we'll show what the motion would look like
        motion_descriptions = {
            "subtle": "Clouds drifting across sky, grass swaying gently, atmospheric mist moving softly, slow pan",
            "moderate": "Clouds flowing across sky, grass swaying gently, atmospheric mist moving softly, smooth tracking",
            "dynamic": "Clouds racing across sky, grass rippling like waves, mist flowing through valleys, light intensity shifting, dynamic sweep",
            "cinematic": "Dramatic clouds racing across vast sky, grass rippling like ocean waves, mist cascading through valleys, cinematic movement revealing landscape grandeur"
        }

        print(f"   Expected Motion: {motion_descriptions[intensity]}")
        print()

if __name__ == "__main__":
    print("🚀 Starting Enhanced Motion Video Generation Tests")
    print("=" * 60)

    # Check API health
    if not test_api_health():
        exit(1)

    print()

    # Demonstrate motion comparison
    demonstrate_motion_comparison()

    # Run the main test
    print()
    results = test_motion_intensity_levels()

    print("\n🎉 Enhanced Motion Testing Complete!")
    print("\n💡 Key Features Added:")
    print("• Motion intensity levels: subtle, moderate, dynamic, cinematic")
    print("• Enhanced scene detection for better motion generation")
    print("• Intensity-specific motion enhancers and camera movements")
    print("• API support for motion_intensity parameter")
    print("• Comprehensive motion keyword system")

    print("\n🛠️ Usage Examples:")
    print("• API: Include 'motion_intensity': 'cinematic' in your request")
    print("• CLI: Add --motion-intensity cinematic to your command")
    print("• Direct: motion_intensity='subtle' in function calls")