An interactive visualization tool for educational purposes that explains the geometric intuitions behind diffusion and flow-based generative models.

An open-weight 8B Diffusion Transformer text-to-image model by Baidu that excels at text rendering, complex instruction following, and structured visual compositions.

A collection of ComfyUI custom nodes optimized for the Z-Image Turbo model, offering enhanced style control, low-step consistency, and compositional variety.

Image MetaHub is a local‑first desktop app (Electron + React) that indexes AI‑generated images, video and audio, extracts generation metadata from tools like ComfyUI, Automatic1111, InvokeAI, Midjourney, etc., and provides fast faceted search, visual‑similarity lookup, lineage tracking and an embedded ComfyUI workspace. The core app is open‑source (MPL 2.0); a $39 offline Pro license unlocks deeper integrations, compare view, unlimited clustering and analytics.

CV‑CUDA is NVIDIA’s open‑source GPU‑accelerated computer‑vision library. It provides fast, scalable image/video operators (resize, filter, color conversion, etc.) that run entirely on the GPU and integrate with C/C++ and Python AI frameworks. Install via pre‑built wheels (`cvcuda-cu12` or `cvcuda-cu13`) for Linux, decode images directly to GPU with `nvimgcodec`, and chain CV‑CUDA operators before feeding data to deep‑learning models. The library targets modern NVIDIA GPUs (Turing and newer) and is used in production systems such as Microsoft Bing visual search and Tencent Cloud video processing.

A Blender add-on that integrates Stable Diffusion to generate seamless textures, concept art, and image assets directly within the 3D workspace.

A text prompt weighting and blending library for transformers-type text embedding systems, enabling precise control over prompt influence in image generation pipelines.

A GUI shell for multiple AI image upscaling tools like waifu2x, Real-CUGAN, and Real-ESRGAN, providing a drag-and-drop interface for image enhancement.

A JavaScript library for AI-powered image enhancement, allowing developers to increase resolution, denoise, and deblur images in the browser or Node.js.

A ComfyUI extension that prevents image artifacts and over-saturation by automatically rescaling CFG and provides tools to accelerate generation speed.

PathML is an open‑source Python library (also available as a Docker image) that streamlines the whole‑slide pathology workflow: loading diverse slide formats, applying scalable preprocessing (stain normalisation, tiling, GPU transforms), converting data to PyTorch tensors, building cell‑level graphs, and running deep‑learning or ONNX models. It aims to lower the barrier to AI‑driven computational pathology and is already used in many high‑impact biomedical publications.

A fast and efficient method for fine-tuning text-to-image diffusion models to learn new concepts from a few images with minimal storage overhead.

A collection of Google Colab notebooks for preparing datasets and training Stable Diffusion LoRA models, including support for SDXL.

SITS is an open‑source R package (with a Python wrapper) for building, regularising, and classifying satellite‑image time‑series data cubes. It connects to many public EO collections, offers a suite of ML/DL models (including GPU‑accelerated CNNs and attention encoders), and provides end‑to‑end tools for training, classification, smoothing, uncertainty estimation, and active learning.

A Docker-based deployment for the FaceFusion face manipulation platform, allowing users to easily run the software across various hardware accelerators like CUDA and ROCm.

A curated collection of research papers and datasets focused on anime and cartoon generation, covering topics from image synthesis to animation production.

PIX is a JAX‑based image‑processing library that provides GPU/TPU‑accelerated functions (e.g., flips, resize) which can be compiled with `jax.jit`, vectorised with `jax.vmap`, or run on multiple devices with `jax.pmap`. Install JAX first, then `pip install dm-pix`. Use it just like NumPy image code, but with full JAX performance and automatic differentiation.

A tool for generating multi-view optical illusions using pixel-based diffusion models, creating images that change identity when rotated, flipped, or rearranged.

A training-free iterative framework for long story visualization that maintains semantic consistency across up to 100 frames using a global reference cross-attention module.

MACE is a finetuning framework for diffusion models that enables the mass erasure of up to 100 unwanted concepts, such as celebrities or explicit content, while preserving unrelated model knowledge.

A two-stage training framework that uses dynamic reward optimization to enable image generation models to accurately combine multiple reference images of different types.

A set of ComfyUI custom nodes that load and stack LoRA weights for Nunchaku Qwen‑Image, Z‑Image‑Turbo, and Krea2 ControlNet models, with UI controls, experimental AWQ‑modulation support, and DiffSynth ControlNet integration.

HistomicsTK is an open‑source Python toolkit for analyzing whole‑slide pathology images. It offers stain normalization, nuclei segmentation, and feature extraction, usable as a standalone library or as a plugin for the Digital Slide Archive web platform.

ResShift is an efficient diffusion model for image super-resolution and restoration that reduces sampling steps to improve inference speed without sacrificing image quality.

A local-first AI image workspace that combines an infinite canvas with agentic planning for multi-step prompt-to-image generation.

An image composition toolbox that provides a unified set of tools for realistic object insertion, including harmonization, shadow generation, and placement assessment.

A set of ComfyUI custom nodes that save images with embedded generation metadata, ensuring compatibility with Civitai and Prompthero.

Eynollah is a Python library and command‑line tool for deep‑learning based document layout analysis, image binarization, enhancement, OCR, and reading‑order detection. It supports 10 layout classes, offers both CNN‑RNN and Transformer OCR models, outputs PAGE‑XML, and can be installed via PyPI or Docker. Pre‑trained ONNX models enable fast inference; full TensorFlow/PyTorch checkpoints are available for fine‑tuning. The project targets high‑quality processing of historical and heterogeneous scanned documents.

ObjectClear is an object removal model that jointly eliminates target objects and their associated effects, such as shadows, while preserving background consistency.

FlowEdit is an open‑source PyTorch implementation of an ICCV 2025 paper that enables fast, inversion‑free text‑guided image editing by reusing pre‑trained diffusion models (e.g., Stable Diffusion 3, Flux) together with a flow network. Users provide a source image, a source caption, and a target caption; the system computes a concise edit code and produces the edited image without costly latent‑space inversion. The repo includes ready‑to‑run scripts, YAML‑based configuration, and links to ComfyUI nodes for broader integration.

A PyTorch implementation of Mean Flows and Improved Mean Flows for high-quality, one-step image generation.

A versatile image inpainting model that supports text-guided object insertion, object removal, and outpainting using tailored task prompts.

Panel Cleaner is an open‑source Python tool that uses machine‑learning models to locate and mask text in comic‑book speech bubbles, then either removes the text (with optional LaMa in‑painting) or extracts it via OCR. It works from the command line or a Qt GUI, supports batch processing, CUDA acceleration, and configurable profiles. The project is GPL‑v3 licensed and available as pip packages, pre‑built binaries, Flatpak, AUR, and Docker images.

kraken is an open‑source, Python‑based OCR toolkit for historical and non‑Latin scripts. It provides trainable layout analysis, reading‑order detection, and neural‑network character recognition, supporting right‑to‑left, bidirectional and vertical scripts. The tool outputs standard XML formats (ALTO, PageXML, hOCR), offers word‑level bounding boxes, and ships with a public model repository. Install via pip/pipx on Linux or macOS, fetch a model with `kraken get …`, and run a one‑line command to binarise, segment, and OCR an image.

A prompt compilation kit for Claude Code that transforms vague image requests into professional, validated prompts for gpt-image-2.

An unofficial PyTorch implementation of the Noise2Noise paper, providing scripts to train and test a U‑Net denoiser on noisy‑only data using Gaussian, Poisson, text‑overlay, or Monte‑Carlo rendering noise.

A 3D-aware generative model that transforms 2D GANs into Multiplane Images to enable 3D view synthesis and mesh extraction.

An end-to-end framework for Matryoshka Diffusion Models that enables efficient training of high-resolution text-to-image synthesis models up to 1024x1024 pixels.

A collection of StyleGAN2-based face generators that create high-fidelity, unique virtual human faces for use in film, advertising, and gaming.

A distilled, faster version of the Qwen‑Image text‑to‑image model (4‑ or 8‑step diffusion) with fp32/bf16/fp8 checkpoints, LoRA adapters, and ready‑to‑use ComfyUI workflows. Provides 12–25× speed‑up with modest quality loss and integrates with Diffusers, ComfyUI, Nunchaku, and Cache‑dit.

A model merging extension for AUTOMATIC1111's Stable Diffusion WebUI that allows users to merge models in-memory and perform block-by-block weight adjustments without saving to disk.

A PyTorch implementation of autoregressive image generation that eliminates the need for vector quantization by using a diffusion-based approach.

A hierarchical generative network and library for creating and animating vector graphics (SVG) using deep learning and differentiable tensors.

A historical PyTorch implementation of the original StyleGAN architecture for generating high-resolution realistic images.

A set of ComfyUI nodes that enhance image details and reduce background blur by adjusting noise schedules (sigmas) during sampling, specifically optimized for Flux and SDXL models.

A PyTorch implementation of GANILLA, a generative adversarial network designed to translate real images into artistic illustrations.

FaceFusion Pinokio is a face manipulation platform that allows users to swap and modify faces in images and videos with a simplified installation process via Pinokio.

ColorFlow is a retrieval-augmented diffusion framework that colorizes black-and-white image sequences while maintaining consistent character and object identities across frames.