Metadata-Version: 2.2
Name: pyimagecuda
Version: 0.0.9
Summary: GPU-accelerated image processing library for Python
Author: Beltrán Offerrall
Requires-Python: >=3.10
Requires-Dist: pyvips[binary]
Description-Content-Type: text/markdown

# PyImageCUDA 0.0.9

[![PyPI version](https://img.shields.io/pypi/v/pyimagecuda.svg)](https://pypi.org/project/pyimagecuda/)
[![Build Status](https://github.com/offerrall/pyimagecuda/actions/workflows/build.yml/badge.svg)](https://github.com/offerrall/pyimagecuda/actions)
![Python](https://img.shields.io/badge/python-3.10%20|%203.11%20|%203.12%20|%203.13-blue)
![Platform](https://img.shields.io/badge/platform-Windows%20%7C%20Linux-brightgreen)
![Tests](https://img.shields.io/badge/tests-85%20passed-brightgreen)
[![NVIDIA](https://img.shields.io/badge/NVIDIA-CUDA-76B900?style=flat&logo=nvidia&logoColor=white)](https://developer.nvidia.com/cuda-zone)

**GPU-accelerated image compositing for Python.**

> PyImageCUDA is built for image composition, not computer vision. It provides GPU tools to create, modify, and blend images, rather than analyze or recognize objects within them.

## Quick Example

<img src="https://offerrall.github.io/pyimagecuda/images/quick.png" alt="Demo" width="400">

```python
from pyimagecuda import Image, Fill, Effect, Blend, Transform, save

with Image(1024, 1024) as bg:
    Fill.color(bg, (0, 1, 0.8, 1))
    with Image(512, 512) as card:
        Fill.gradient(card, (1, 0, 0, 1), (0, 0, 1, 1), 'radial')
        Effect.rounded_corners(card, 50)

        with Effect.stroke(card, 10, (1, 1, 1, 1)) as stroked:
            with Effect.drop_shadow(stroked, blur=50, color=(0, 0, 0, 1)) as shadowed:
                with Transform.rotate(shadowed, 45) as rotated:
                    Blend.normal(bg, rotated, anchor='center')

    save(bg, 'output.png')
```

## Key Features

* ✅ **Zero Dependencies:** No CUDA Toolkit, Visual Studio, or complex compilers needed. Is Plug & Play on both Windows and Linux.
* ✅ **Ultra-lightweight:** library weighs **~1 MB**.
* ✅ **Studio Quality:** 32-bit floating-point precision (float32) to prevent color banding.
* ✅ **Advanced Memory Control:** Reuse GPU buffers across operations and resize without reallocation—critical for video processing and batch workflows.
* ✅ **OpenGL Integration:** Direct GPU-to-GPU display for real-time preview widgets.
* ✅ **API Simplicity:** Intuitive, Pythonic API designed for ease of use.

## Use Cases

* **Generative Art:** Create thousands of unique variations in seconds.
* **Motion Graphics:** Process video frames or generate effects in real-time.
* **Image Compositing:** Complex multi-layer designs with GPU-accelerated effects.
* **Node Editors & Real-time Tools:** Build responsive image editors with instant preview.
* **Game Development:** Procedural UI assets, icons, and sprite generation.
* **Marketing Automation:** Mass-produce personalized graphics from templates.
* **Data Augmentation:** High-speed batch transformations for ML datasets.

## Installation
```bash
pip install pyimagecuda
```

**Note:** `pyvips` is the only mandatory dependency (installed automatically). It is used strictly for robust file I/O (JPG, PNG, WEBP...) and high-quality Text rendering on the CPU.

## Documentation

**⚠️ Alpha Release:** Many more features are planned and under development. If you have specific needs or bug reports, please open an issue on GitHub.

### Core Concepts
* [Getting Started Guide](https://offerrall.github.io/pyimagecuda/)
* [Image & Memory](https://offerrall.github.io/pyimagecuda/image/) (Buffer management)
* [IO](https://offerrall.github.io/pyimagecuda/io/) (Loading and Saving)
* [OpenGL Integration](https://offerrall.github.io/pyimagecuda/opengl/) (Real-time preview, zero-copy display)

### Operations
* [Fill](https://offerrall.github.io/pyimagecuda/fill/) (Solid colors, Gradients, Checkerboard, Grid, Stripes, Dots, Circle, Ngon, Noise, Perlin)
* [Text](https://offerrall.github.io/pyimagecuda/text/) (Rich typography, system fonts, HTML-like markup, letter spacing...)
* [Blend](https://offerrall.github.io/pyimagecuda/blend/) (Normal, Multiply, Screen, Add, Overlay, Soft Light, Hard Light, Mask)
* [Resize](https://offerrall.github.io/pyimagecuda/resize/) (Nearest, Bilinear, Bicubic, Lanczos)
* [Adjust](https://offerrall.github.io/pyimagecuda/adjust/) (Brightness, Contrast, Saturation, Gamma, Opacity)
* [Transform](https://offerrall.github.io/pyimagecuda/transform/) (Flip, Rotate, Crop)
* [Filter](https://offerrall.github.io/pyimagecuda/filter/) (Gaussian Blur, Sharpen, Sepia, Invert, Threshold, Solarize, Sobel, Emboss)
* [Effect](https://offerrall.github.io/pyimagecuda/effect/) (Drop Shadow, Rounded Corners, Stroke, Vignette)

## Performance

PyImageCUDA shows significant speedups for GPU-friendly operations like blending, filtering, and transformations. Performance varies by operation complexity and workflow:

- Complex operations (blur, blend, rotate) see **10-260x improvements**
- Simple operations (flip, crop) see **3-20x improvements**  
- Real-world pipelines with file I/O typically see **1.5-2.5x speedups**

Results depend on your hardware, batch size, and whether you reuse GPU buffers.

**[→ View Detailed Benchmarks](https://offerrall.github.io/pyimagecuda/benchmarks/)**

## Requirements

* **OS:** 
    - Windows 10 or 11 (64-bit).
    - Linux: Any modern distribution (Ubuntu, Fedora, Debian, Arch, WSL2, etc.).
* **GPU:** NVIDIA GPU (Maxwell architecture / GTX 900 series or newer).
* **Drivers:** Standard NVIDIA Drivers installed.

**NOT REQUIRED:** Visual Studio, CUDA Toolkit, or Conda.

## Linux Compatibility & Troubleshooting

PyImageCUDA is currently tested primarily on **Ubuntu LTS** releases with up-to-date NVIDIA drivers.

If you encounter the following error on Linux:

```text
RuntimeError: Kernel launch failed: the provided PTX was compiled with an unsupported toolchain.
```

Solution: This indicates your installed NVIDIA drivers are too old to execute the kernels included in the library. Please update your NVIDIA drivers to the latest version available for your distribution (Proprietary drivers recommended).

We are actively investigating ways to broaden compatibility for older drivers and legacy Linux distributions in future releases.

## Tests
```bash
pytest tests/tests.py
```

## Contributing
Contributions welcome! Open issues or submit PRs

## License
MIT License. See [LICENSE](LICENSE) for details.