High-Performance AI Image Upscaling

License: MIT Platforms: Linux, macOS

High-performance REST API for AI-based image upscaling of massive files.

// Features

Why choose MLCUpscale.

AI Upscaling

High-quality 2x, 3x, and 4x image upscaling using Real-ESRGAN models.

Massive Image Support

Specifically designed to handle huge images up to 10,000x10,000 pixels and beyond.

GPU Acceleration

Optimized for Vulkan-compatible GPUs (NVIDIA, AMD, Intel) with automatic CPU fallback.

Async REST API

Modern asynchronous API for job submission, real-time status tracking, and download.

Specialized Models

Includes optimized models for general photos, anime art, and video frames.

Docker Ready

Fully containerized for easy deployment with health checks, metrics, and rate limiting.

MLCUpscale is a high-performance REST API designed for one specific task: upscaling massive images using state-of-the-art AI models. While many tools struggle with images over a few megapixels, MLCUpscale is built to handle files as large as 10,000x10,000 pixels by utilizing advanced tiling strategies and GPU acceleration.

Beyond Consumer Hardware Limits

By offloading the compute-intensive upscaling process to a dedicated server, clients can process professional-grade imagery without local hardware constraints. MLCUpscale utilizes Vulkan-based hardware acceleration, making it compatible with a wide range of GPUs on both Linux and macOS (including Apple Silicon).

Asynchronous Workflow

Submit: Send a POST request with your image and desired scale factor (2x, 3x, or 4x).
Track: Poll the status endpoint to monitor progress in real-time.
Download: Once completed, retrieve the high-resolution result.

API Quickstart

Submit a new upscaling job:

curl -X POST http://api.mlcgo.eu/v1/upscale \
  -F "image=@photo.jpg" \
  -F "scale=4" \
  -F "model_name=realesrgan-x4plus"

Check the status:

curl http://api.mlcgo.eu/v1/status/YOUR_JOB_ID

Download the result:

curl -O http://api.mlcgo.eu/v1/download/YOUR_JOB_ID