nalltama/raiv

RAIV is a Windows image viewer written in Python that automatically enhances images using AI upscaling as you browse. When you open an image or folder, the viewer sends the images through one of two AI processing engines, Real-CUGAN or Real-ESRGAN, which increase image resolution and reduce visual noise. The result is displayed in place of the original, so you see a sharpened version without manually running a separate tool. You can also use a comparison slider to see the original and processed versions side by side. The viewer supports opening individual image files, entire folders, and common archive formats including ZIP, RAR, and 7z, which is useful for reading comic archives. It includes a spread view mode for displaying two pages side by side, as manga is commonly read. Prefetching is built in, meaning the next images are processed in the background while you read the current one, keeping page transitions fast even with large folders. Two AI engines are bundled. Real-CUGAN is suited for anime and illustration images and supports adjustable upscale ratios and noise reduction levels. Real-ESRGAN includes three models: a lightweight anime option, a general photo model, and a higher-quality anime model, all at a fixed 4x scale. Processed images can optionally be saved to subfolders so they do not need to be reprocessed on the next visit. The viewer also supports tone curve adjustments using GIMP curve files for fine-tuning how images look on screen, customizable keyboard and mouse button bindings, and both Japanese and English interfaces. Setup requires Python and a brief install step to download dependencies. The project is MIT-licensed and was built using AI-assisted coding with Codex.