This repository provides an unofficial ComfyUI implementation of APISR, a fast and effective super-resolution model specifically designed for anime content. It offers 2x and 4x upscaling capabilities, catering to both image and video processing, with modes optimized for high and low VRAM environments.

How It Works

The implementation integrates APISR models as custom nodes within the ComfyUI workflow. It supports two official APISR models (2x and 4x) which require manual download and placement. The core functionality includes batch processing for high-VRAM scenarios and iterative processing for low-VRAM environments, enabling upscaling of both static images and video frames.

Quick Start & Requirements

• Install: Recommended via ComfyUI Manager, or manually: cd custom_nodes && git clone https://github.com/ZHO-ZHO-ZHO/ComfyUI-APISR && cd ComfyUI-APISR && pip install -r requirements.txt.
• Prerequisites: ComfyUI, Python. Manual model download required for /ComfyUI/models/apisr. CPU generation is supported.
• Resources: VRAM requirements vary based on batch vs. iterative processing and video length.

Highlighted Details

• Supports both image and video upscaling.
• Offers "Batch" (high VRAM) and "Iterative" (low VRAM) processing modes for videos.
• Includes a APISR ModelLoader node.
• V1.1 adds CPU generation support.

Maintenance & Community

• Active development with recent updates (March 2024).
• Community support via QQ group (839821928).
• Author contact: zhozho3965@gmail.com.

Licensing & Compatibility

• License not explicitly stated in the README.
• Compatibility with ComfyUI workflows is the primary focus.

Limitations & Caveats

Processing long videos in batch mode may require significant VRAM. The implementation is unofficial, meaning it relies on the author's interpretation of the original APISR project.