nihui/realsr-ncnn-vulkan

Health v2 · maintenance only

Adoption not part of the score

1213 stars · 112 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

A command-line tool implementing the RealSR real-world super-resolution model using the ncnn inference framework with Vulkan GPU acceleration. It upscales images 4x on Intel, AMD, and Nvidia GPUs without requiring CUDA or other deep learning runtimes.

Use cases

• upscale low-resolution photos 4x
• enhance JPEG-compressed images with super resolution
• batch upscale a folder of images on GPU
• run super resolution without CUDA or PyTorch
• upscale anime or game screenshots
• process images on AMD or Intel GPUs

When to choose

• you need portable GPU-accelerated image upscaling with no Python or CUDA dependencies
• you want a simple CLI for batch super-resolution on Windows/Linux/macOS
• your GPU is AMD or Intel where CUDA tools don't work

When to avoid

• you need real-time or video super-resolution
• you want to train or fine-tune the model yourself
• you need arbitrary upscale ratios other than 4x

Facets

cli-tool · maturity maintenance

image-processing machine-learning llm-inference image-processing computer-vision machine-learning windows cross-platform cli super-resolution ncnn vulkan upscaling image-enhancement gpu-accelerated linux macos gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="nihui/realsr-ncnn-vulkan")