Spaces:
Running
title: 4K Upscaler
emoji: 🐠
colorFrom: blue
colorTo: red
sdk: docker
pinned: false
license: apache-2.0
🎬 4K Upscaler AI-Powered Image & Video Upscaling to Ultra HD Quality Transform your images and videos to stunning 4K resolution using advanced neural upscaling techniques powered by PyTorch and GPU acceleration. 🚀 What It Does This application uses sophisticated neural network interpolation and enhancement algorithms to upscale images and videos to 4K resolution (4x the original size) while maintaining and improving visual quality. Unlike simple resizing, it employs:
Progressive Neural Upscaling: Multi-stage bicubic interpolation with anti-aliasing AI-Enhanced Sharpening: Custom convolution kernels for detail enhancement Adaptive Contrast Optimization: Automatic brightness and contrast adjustment GPU-Accelerated Processing: CUDA optimization for fast processing Intelligent Smoothing: Noise reduction while preserving important details
✨ Features 🎨 Smart Image Processing
Upscale images from any resolution to 4K (3840x2160 or equivalent 4x scaling) Advanced sharpening algorithms that enhance details without artifacts Automatic contrast and brightness optimization Support for PNG, JPG, JPEG, and GIF formats
🎬 Professional Video Enhancement
Frame-by-frame 4K upscaling for videos Maintains original frame rate and aspect ratio Batch processing for efficiency Support for MP4, AVI, MOV, and MKV formats
⚡ Performance & Efficiency
GPU Acceleration: Automatic CUDA detection and optimization CPU Fallback: Seamless fallback when GPU unavailable Memory Management: Smart memory allocation and cleanup Progressive Processing: Handles large files without memory overflow
🧹 Automatic Storage Management
Auto-Cleanup: Files automatically deleted after 1 hour Background Scheduler: Cleanup runs every 10 minutes Storage Statistics: Real-time monitoring of disk usage Manual Controls: Force cleanup and cache clearing
📊 Real-Time Monitoring
Live processing logs with timestamps GPU memory usage tracking Processing progress indicators System performance metrics
🛠️ Technical Specifications AI Processing Pipeline
Input Normalization: Convert to floating-point tensors Progressive Upscaling: 2x → 4x scaling stages Neural Enhancement: Custom sharpening kernels Quality Optimization: Adaptive contrast and smoothing Output Conversion: High-quality image/video encoding
System Requirements
GPU: NVIDIA GPU with CUDA support (recommended) Memory: 4GB+ RAM, 2GB+ GPU VRAM for large files Storage: Automatic cleanup manages disk space Formats: Images (PNG, JPG, GIF), Videos (MP4, AVI, MOV, MKV)
🎯 Use Cases
Content Creation: Enhance low-resolution footage for professional projects Archive Restoration: Upscale old photos and videos to modern standards Social Media: Create high-quality content from mobile captures Gaming: Enhance screenshots and recordings Research: Improve visual data quality for analysis Personal: Transform family photos and memories to 4K
🚀 Quick Start
Upload: Drag and drop your image or video file Process: AI automatically detects file type and starts 4K upscaling Monitor: Watch real-time progress in the processing logs Download: Get your enhanced 4K file when processing completes
The application automatically handles GPU detection, memory management, and file cleanup - no configuration needed! 📈 Performance
Images: Typical 1080p → 4K upscaling in 10-30 seconds (GPU) Videos: Processing speed depends on length and GPU capability Quality: Advanced algorithms preserve and enhance detail quality Efficiency: Optimized memory usage prevents system overload
🔧 API Endpoints
POST /api/upload - Upload and process files GET /api/processing-status - Check processing progress GET /api/download/ - Download processed files GET /api/system - System and GPU information POST /api/cleanup-now - Manual storage cleanup GET /api/storage-stats - Detailed storage statistics
🧹 Storage Management Files are automatically managed to prevent storage overflow:
Auto-deletion: Files removed after 1 hour Regular cleanup: Runs every 10 minutes Smart monitoring: Tracks storage usage and cleanup statistics Manual controls: Force cleanup when needed
👨💻 Created By Angel E. Pariente This project combines cutting-edge neural network techniques with practical engineering to make professional-quality 4K upscaling accessible to everyone. 📄 License Licensed under the Apache License 2.0 - see the license for details. 🌟 Technology Stack
Backend: Python, Flask, PyTorch Computer Vision: OpenCV, PIL GPU Computing: CUDA, cuDNN Container: Docker Scheduling: Background task automation