Bring Your Memories Back to Life

Transform grainy 8mm home movies into stunning HD videos with our GPU-accelerated AI pipeline. Featuring optimized Real-ESRGAN, GFPGAN face restoration, and professional-grade enhancement. Process 10-second clips in just 5-6 minutes!

See the Transformation

📼 Original

✨ AI Restored

Watch side-by-side comparison: Grainy original vs AI-enhanced restoration

📷 How I Digitized the Films

I used my Kodak scanner to digitize these precious 8mm films frame by frame, preserving every moment before applying AI restoration.

My Kodak scanner setup for frame-by-frame digitization

98%

GPU Utilization

2x-4x

AI Upscaling

7-Frame

Temporal Denoising

6 Steps

Enhancement Pipeline

Optimized GPU-Accelerated Pipeline

🎯

Stabilization + Crop

Remove camera shake with 10% edge crop. Eliminates black borders and damaged film edges for professional results.

✨

Deflickering

Fix brightness flickering from old projectors. Temporal median filtering for consistent, smooth playback.

🧹

GPU Denoising

93x faster than CPU! GPU-accelerated temporal grain removal (0.14 sec/frame) while preserving details.

�

Face Restoration

GFPGAN v1.4 with weight 0.7 for natural enhancement. Runs BEFORE interpolation to prevent flickering.

�

Detail Enhancement

Adaptive unsharp masking (crisp preset). Edge-aware sharpening that ignores grain. Tested and optimized.

🎨

Color Correction

Natural "soft" preset with minimal processing. Preserves authentic film look without over-enhancement.

Get Started in Minutes

Clone the Repository

git clone https://github.com/avisalmon/MOV8mm.git

Download the complete pipeline to your computer.

Install Dependencies

pip install -r requirements.txt

Requires Python 3.12+, CUDA 12.1+, and 8GB+ GPU VRAM.

Run Setup

python setup.py

Automatically downloads AI models and configures your environment.

Restore Your Films!

python scripts/enhance_video.py input_videos/film.MP4 --cut 0 10 --setup

Process your first 10-second test clip in just 5-6 minutes!

System Requirements

🖥️

Operating System

Windows 10/11, Linux, or macOS with Python 3.12+

🎮

GPU Required

NVIDIA GPU with 8GB+ VRAM (tested on RTX 4060). CUDA 12.1+ required.

⏱️

Processing Time

~5-6 minutes for 10 seconds @ 20fps (200 frames). 30-second clip: ~15 min. 60-second clip: ~30 min (RTX 4060).