chaiNNer is a free, open-source, node-based image processing editor built for AI upscaling — drag nodes onto a canvas, connect them, and run powerful super-resolution pipelines locally.
chaiNNer turns image processing into a visual graph. Each node does one thing — load an image, upscale it with a model, restore a face, fix colors, save the result — and you wire nodes together into a chain that can be as simple as load-upscale-save or as elaborate as a conditional batch pipeline over whole folders and videos. It was originally written for AI upscaling and remains the tool enthusiasts reach for when they want full control over which model processes their images.
Through its Spandrel library it supports ESRGAN, SPAN, OmniSR and dozens of other architectures, with community models from OpenModelDB, running on PyTorch, NCNN, ONNX or TensorRT across Nvidia, AMD, Intel and Apple hardware. Chains can also be executed from the command line.
An active Discord community and a steady release cadence keep chaiNNer current with new model architectures as they appear. Saved chains can also be shared between users, so beginners can start from community templates for common jobs such as anime upscaling or game texture work instead of building every pipeline from a blank canvas.
chaiNNer is completely free and open source. There are no subscriptions, credits or locked features; an optional Ko-fi page exists purely for voluntary support.
chaiNNer is for upscaling enthusiasts, game texture modders, archivists and technical users who want to choose their own models and build automated pipelines instead of trusting a website’s black box.