# Yuanshi9815/OminiControl

[ICCV 2025 Highlight] OminiControl: Minimal and Universal Control for Diffusion Transformer

Repository: https://github.com/Yuanshi9815/OminiControl
Canonical: https://ross.abutalabs.com/products/ominicontrol
Language: Python
License: Apache-2.0
License Family: permissive
Last push: 2026-07-02T06:12:46+00:00

## Health v2 (maintenance only)
Score: 62/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 90, release rhythm 35, longevity 46
- inputs: {"age_days": 654, "days_push": 62, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1927, forks 150 (observed 2026-08-28T04:05:55.645688+00:00)

## What it is
OminiControl is a universal control framework for Diffusion Transformer models like FLUX, supporting subject-driven and spatial control (edge-guided generation, inpainting) with only ~0.1% additional parameters. It includes training code, a companion efficient-conditioning method (OminiControl2), and pretrained models on HuggingFace.

## Use cases
- control FLUX image generation with a reference subject
- generate images guided by edge maps or sketches
- inpaint parts of an image with a diffusion transformer
- train a subject-driven LoRA adapter for FLUX
- speed up conditioned diffusion inference with KV-cache
- add spatial control to diffusion transformer models

## When to choose
- you need subject-driven or spatially controlled generation with FLUX or similar DiT models
- you want a lightweight control adapter that barely modifies the base model
- you want published, paper-backed methods (ICCV 2025) with training code

## When to avoid
- you use UNet-based diffusion models like SD 1.5 instead of diffusion transformers
- you need a production image-generation service rather than a research library
- you lack a GPU for training or inference

## Facets
- artifact type: library
- maturity: active
- function: image-processing, machine-learning, deep-learning
- domain: artificial-intelligence, image-processing, deep-learning
- platform: python
- tags: diffusion-transformer, flux, controllable-generation, subject-driven-generation, inpainting, lora, research, gpu

## Member repositories
- Yuanshi9815/OminiControl (main) score 62

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:55.645688+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T03:09:09.121035+00:00, confidence not recorded.
  - readme: https://github.com/Yuanshi9815/OminiControl (fetched 2026-08-28T04:05:55.645688+00:00, sha 0c4a3cf8418c)
- Data as of 2026-08-30T08:39:29.467469+00:00.