Upscale images directly within GIMP (2.10 & 3.0) using the RealESRGAN AI upscaler powered by ncnn and Vulkan for fast, high-quality enhancement of photos, art, and anime.

Upscaler image and video (and gif) with simple user interface

Picsart Creative APIs SDK for Java. Includes helper methods and functions for Programmable Image APIs (e.g. Remove Background, Upscale, Enhance, Effects) and the GenAI APIs (e.g. Text2Image, Replace, Expand Image).

Designed for texture files with transparency to ensure the alpha channel remains intact during scaling.

Picsart Creative APIs SDK for Typescript. Includes helper methods and functions for Programmable Image APIs (e.g. Remove Background, Upscale, Enhance, Effects) and the GenAI APIs (e.g. Text2Image, Replace, Expand Image).

sharpify-gui is a powerful, open‑source desktop application built with PyQt6, leveraging Real‑ESRGAN AI models and FFmpeg to batch upscale your favorite anime images and videos with ease

TheAtlas is a robust and intuitive tool designed for seamless image and video processing on Windows. With its user-friendly interface, AtlasGUI simplifies running models for upscaling and interpolating tasks, including ESRGAN, RIFE, and CUGAN.

COG implementation of SEESR with Stable Diffusion Turbo

Zippy is a simple, fast, and smart macOS app - combining media conversion, downloading, extraction, and AI-powered image tools in one place. 🎉

Manage ingesting mangas, placing them in the right places and merging multiple sources together, while also enabling automatic upscaling of said mangas.

go-upxcale is a self-contained Go service that orchestrates external AI tools to upscale images and videos

Picsart Plugin for Figma. Includes Remove Background and Upscale.

Unofficial implementation of the SRGAN model. With this repository you will be able to implement the model from scratch with your own dataset.

PyTorch implementation of ESRGAN for 4× image super-resolution. Features two-phase training (PSNR + GAN), RRDBNet architecture with 23 RRDB blocks, and comprehensive monitoring guides. Achieves 25.5-26 dB PSNR on validation data. Trained on DIV2K, Flickr2K, and ISR datasets.