https://github.com/stacylkv/preset-from-pair
Create a LUT model from a pair of images.
https://github.com/stacylkv/preset-from-pair
cv2 lut matplotlib nympy opencv pickle sklearn tqdm
Last synced: 3 months ago
JSON representation
Create a LUT model from a pair of images.
- Host: GitHub
- URL: https://github.com/stacylkv/preset-from-pair
- Owner: stacylkv
- License: mit
- Created: 2025-06-25T23:47:37.000Z (about 1 year ago)
- Default Branch: main
- Last Pushed: 2025-06-30T00:23:28.000Z (about 1 year ago)
- Last Synced: 2025-06-30T01:35:19.669Z (about 1 year ago)
- Topics: cv2, lut, matplotlib, nympy, opencv, pickle, sklearn, tqdm
- Language: Python
- Homepage:
- Size: 18.5 MB
- Stars: 1
- Watchers: 0
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE.txt
Awesome Lists containing this project
README
# 🎨 Preset-From-Pair — Generate a LUT from Before/After Images
This script allows you to **create a LUT model (color transformation)** from a pair of images — a raw image (`before`) and its edited version (`after`, e.g. from Tezza or VSCO) — and then **apply that color style to an entire folder of images**.
---
## 📂 What the Script Does
1. Loads a pair of images: `original.jpg` (before editing) and `target.jpg` (after editing).
2. Builds a LUT model based on the color differences between them.
3. Applies the LUT to every image in the `input_images/` folder.
4. Saves the results in the `output_images/` folder.
5. (Optional) Displays a side-by-side **Before/After preview**.
---
## 🚀 How to Use
1. Install dependencies:
```bash
pip install -r requirements.txt
```
2. Prepare the following files/folders:
- `original.jpg` — original image (unprocessed)
- `target.jpg` — edited version with your desired style
- `input_images/` — folder with images you want to style
3. Run the script:
```bash
python main.py
```
4. The styled images will appear in the `output_images/` folder.
---
## 🧪 Example Use Case
- `original.jpg`: raw photo from your camera
- `target.jpg`: styled photo from Tezza or another app
- The script "learns" the aesthetic from `target` and applies it to others
---
## 🧠 Features
- Saves the LUT model to `lut_model.pkl`
- Reuse the LUT by loading it later with `load_lut(path)`
- Displays Before/After comparison with matplotlib
- Fast and customizable
---
## 📦 Dependencies
Use the following `requirements.txt`:
```
numpy>=1.24
opencv-python>=4.8
matplotlib>=3.7
tqdm>=4.65
scikit-learn>=1.3
```
---
## 🧊 Potential Improvements
- Export to `.cube` LUT format
- Support for video input
- GUI for non-programmers
---
## 📸 Author
Built for creative workflows and quick aesthetic color transfer between images.
Useful for automating styles inspired by apps like Instagram, Tezza, VSCO, and more.
> **Note:** This project is in the development stage.
> In addition to `lut-filter.py`, there are two other Python scripts in this repository that can modify your images without LUT generation:
> - [`Simple Color Shift`](src/simple-color-shift.py)
> - [`Brightness Contrast Adjust`](src/brightness-contrast-adjust.py)
> Both scripts applie a Tezza-like style to images by adjusting exposure, brightness, fade, and warmth, but in different way. It processes all images in the specified input directory and saves the adjusted images to the output directory.
### 🧪 Usage
1. Place your input images in the `input_images` folder.
2. Run the script.
3. The processed images will be saved in the `output_images` folder.