AI tools for Diffusion

NitroFusion2 is a text-to-image checkpoint designed to render a result in one or two diffusion steps. It builds on Z-Image-Turbo and targets sharp visual detail, including text

P-H3-Image is Pruna AI's open-weight image generator and editor derived from MiniMax-H3. It supports text-to-image work and edits conditioned on several reference images, with a

Clover-35B-A3B-Preview is DiffuSpace's open-weight diffusion language model. It targets text generation and agentic workloads with a 35-billion-parameter mixture-of-experts back

Open-H3-Regenerate-2K refines low-resolution draft videos into sharper high-resolution outputs. The open pipeline preserves the draft's scene and motion while regenerating fine

SceneBloom creates realistic indoor environments for training embodied agents. It translates room descriptions into varied furniture layouts and simulation-ready scenes, coverin

IDOL converts a single object image into a three-dimensional Gaussian splat. It accepts photographs, rendered objects, and generated images, reconstructing a viewable asset that

Streaming Code World Model is a causal video-rendering implementation that turns a running proxy world into successive generated chunks. It uses a sliding six-chunk attention sp

StreamRender-H3 turns a code-controlled game world into a continuously generated visual experience. A code game engine maintains state and rules while a two-step streaming diffu

AlphaProtein Novo is a generative modeling pipeline for de novo enzyme design around structural motifs. A diffusion model jointly generates protein structures and sequences cond

Kandinsky 6.0 is a family of diffusion models for generating synchronized video and audio. Lite and Pro variants generate short audio-visual clips from text or images, with a se

PDMD, or Projected Distribution Matching Distillation, is a method and implementation for faster diffusion generators. A projection removes the component of the training directi

ALoDLM-8B is an adaptively looped diffusion language model for parallel text and code generation. Shared transformer layers repeatedly refine unresolved latent states, with conf

Supra2-IMG is a small text-to-image model trained from scratch for lightweight image-generation experiments. Its approximately 104M-parameter diffusion transformer uses a frozen

Block3D generates 3D shape geometry from text using block-wise diffusion. It combines the efficiency of processing only part of a shape representation at a time with the ability

Intern Lumina U2 unifies language and visual tasks in a diffusion language-model architecture. It supports text question answering, image interpretation and editing, image creat

DiffusionOPSD is an on-policy self-distillation method for improving diffusion image models using reward guidance. It converts a final-image quality signal into explicit trainin

Atlas is World Labs omni world model for creating, reconstructing, and simulating spatial environments. It combines text, images, video, and 3D context to generate views and wor

UniMate generates articulated motion from a rigged 3D asset and a text prompt across diverse skeletal structures. It aims to animate humans, animals, and articulated objects wit

LLaDA-Image is a six-billion-parameter model family for image generation and instruction-guided editing. A single checkpoint can create an image from text or alter a reference w

Marigold V2 converts a single image into detailed depth and other dense visual predictions. Its model family covers ordinary depth, geometry behind transparent surfaces, surface