Havuz

🔥 On-Policy Self-Distillation in Diffusion Models

The open .char format for consistent characters. Face, body, wardrobe & voice in one file, consistent across every model.

Refine and improve the quality of small faces in MiniMax H3 video. Per-frame face tracking, crop, refine with H3, stitch back.

Self-hosted, one-tab workbench for the whole training lifecycle: build Character/Concept/Style datasets (generate, scrape, or triage a big d

Official code for Echo-Memory: a controlled study of memory in action-conditioned video world models (Context, Compression, Spatial, State-S

DMAD: Distribution Matching as Adversarial Distillation for Fast Visual Generation

Local inference for diffusion language models (LLaDA, Dream), written in Mojo

Official implementation of "AReA: Adaptive Receptive-Field Attention for Efficient High-Resolution Diffusion Transformers"

NEWTON: Agentic Planning for Physically Grounded Video Generation

Split FLUX.2 and LTX 2.3 across two GPUs (LAN or same-machine) — NVENC compresses activations live on the wire. Icarus (ComfyUI node) + Daed

Official Implementation of LongLive-RAG: A general retrieval-augmented framework for long video generation.

Connect Claude to image generation with Agent Skills. Three levels: a zero-cost code-based design engine, a Three.js 3D renderer, and a real

[arXiv 2026] This is the official PyTorch implementation of "MeanFlowNFT: Bringing Forward-Process RL to Average-Velocity Generators".

🌹 Rose: Range-Of-Slice Equilibration PyTorch optimizer. Stateless optimization through range-normalized gradient updates.

[Preprint] TBSM: Three-Body Scattering for Generative Modeling.

CLI-native local AI image and video generation for people, scripts, and agents — CUDA on Linux, Metal on macOS, plus desktop, web, iPhone, R

AI Image Generator is a powerful and user-friendly desktop application designed to help creators produce stunning, high-quality artwork, pho

Distributed image and video generation in ComfyUI for two NVIDIA DGX Sparks.

Official code for InfoVis: one 9B model for slide image generation and editing, with the VisSlideBench benchmark and 15M+ training records.

[ACL 2026] Breaking Block Boundaries: Anchor-based History-stable Decoding for Diffusion Large Language Models

[ECCV 2026] WarpI2I: Image Warping for Image-to-Image Translation.

Building an Image-to-Video (I2V) Model from Scratch

A local AI film studio for one Windows PC with an 8 GB NVIDIA card. Write the story, cast the people and render the shots with LTX-2.5, Mini

Generative Models for Real-World Sensor Time Series

This is a comprehensive tool for merging, analyzing, and training Anima models (cosmos) in a hierarchical (layer-by-layer) manner.Multilingu

(ICML 2026) IDLM: Inverse-distilled Diffusion Language Models

Native macOS app for local AI image generation and editing on Apple silicon: Z-Image Turbo, Qwen-Image-2.1, Krea 2 Turbo, Anima Turbo and MO

[arXiv 2026] Code for "Elastic Token Compression for Pixel-Space Diffusion Transformers"

Official repository for "Persistence Forcing: Exploiting Feature Specialization in Pixel-Space Diffusion"

[ICML 2026] Alleviating Prompt Forgetting in Multimodal Diffusion Transformers

[NeurIPS 2026] Official implementation of Contrastive Distribution Matching for Amortized Sequential Monte Carlo in Discrete Diffusion

Full-pipeline ComfyUI node pack tailored for Krea 2 Turbo: NAG negative guidance, Variance diversity, custom schedulers, samplers & aestheti

Guía gratuita de IA generativa avanzada en español (311 págs.): Transformer, post-entrenamiento, inferencia, GPU, difusión, agentes, segurid

AnyFlow - flow-matching video generation: training, inference and VBench evaluation for t2v and i2v models.

ComfyUI top-bar button that reads a workflow's Markdown notes, downloads the models it needs into the right folder, and relinks loader nodes

ComfyUI deployment kit for Qwen-Image-2.1 Uncensored (GGUF): ready-to-use t2i & image-edit workflows, ComfyUI-GGUF architecture patches, and