giorgioskij / tripod

Fully-convolutional image sharpening for real world pictures

Tripod

This is Tripod, an AI Image Sharpener. It tries its best to remove or reduce motion blur from pictures.

Feel free to test it, use it, improve it.

Status

Right now, exetremely minimal usecases are covered:

• Tripod is only meant to receive as input square images with sizes that are powers of 2 (so 256x256, 512x512, 1024x1024). Crop your images accordingly before using it, or be prepared to read error logs.
• Tripod was trained on images that are artificially blurred. The guy is trying his best, don't expect him to perfectly sharpen a real-world blurry image.

Install

Clone the repo

git clone https://github.com/giorgioskij/tripod
cd tripod

Install the dependencies

conda env create -f environment.yml
conda activate tripod

Use

The best way to use Tripod is to interact with the code directly. However, if you just want to test the model, you can run

python scripts/demo.py <path to your image>

The sharpened image should be saved in the same directory as the input one.

Examples

Here follow a few examples of comparing blurred input, sharpened output, and sharp original.