chaiNNer

The software provides a visual node editor that lets users assemble image‑processing pipelines by dragging and linking nodes. Each node represents an operation such as filtering, color adjustment, transformation, blending, or effect, and the connections define the order in which data flows. Users can construct simple or highly complex workflows without writing code, and the interface validates node types in real time to catch configuration errors before execution.

It supports batch processing of entire folders of images or video frames, and a command‑line interface is available for integration into larger automation scripts. The application includes GPU acceleration for NVIDIA (CUDA/TensorRT), AMD (ROCm/NCNN), Apple Silicon (MPS), and Intel (NCNN) hardware, with a CPU fallback when GPUs are unavailable. AI upscaling is handled through community‑trained models in formats such as PyTorch, ONNX, and TensorRT, allowing the use of architectures like ESRGAN and OmniSR.

The program is self‑contained, bundling an isolated Python runtime and managing dependencies automatically, so no external Python installation is required. It runs on Windows, macOS, and Linux, distributed as an installer or portable archive, making it suitable for users who need a customizable, cross‑platform solution for repetitive or advanced image‑processing tasks.