Awesome Reference Image Generation
Awesome-Reference-Image-Generation
A curated list of image-conditioned image generation methods with release dates, parameters, teams, and links.
Contents
Overview
This repository collects major image-conditioned image generation methods from 2023 to 2026, including ControlNet, IP-Adapter, image editing, image composition, and reference-based generation approaches. Each entry includes the model name, release date, parameter count, releasing team, open-source link, and technical report.
2023
| Model Name | Release Date | Parameters | Team | Open Source Link | Technical Report |
|---|---|---|---|---|---|
| ControlNet | 2023-02-08 | SD UNet Encoder + SD (1.5 or XL) | Stanford University | GitHub | arXiv |
| T2I-Adapter | 2023-02-16 | CNN Adapter + Stable Diffusion (1.5 or XL) | ARC Lab, Tencent PCG | GitHub | arXiv |
| ELITE | 2023-02-16 | CLIP image encoder + MLP + Stable Diffusion 1.5 | Harbin Institute of Technology | GitHub | arXiv |
| InstructPix2Pix | 2023-03-14 | 0.9 B (Stable Diffusion 1.5) | University of California, Berkeley | GitHub | arXiv |
| FreeDoM | 2023-03-17 | 0.9 B (Stable Diffusion 1.5) | Peking University | GitHub | arXiv |
| MagicBrush | 2023-06-16 | 0.9 B (Stable Diffusion 1.5) | The Ohio State University | GitHub | arXiv |
| IP-Adapter | 2023-08-13 | CLIP Image Encoder + MLP + Stable Diffusion (1.5 or XL) | Tencent AI Lab | GitHub | arXiv |
| InstructDiffusion | 2023-09-26 | 0.9 B (Stable Diffusion 1.5) | Microsoft Research Asia | GitHub | arXiv |
| SmartEdit | 2023-12-11 | LLaVA-1.1 (7B or 13B) + QFormer + BIM module + LoRA + InstructDiffusion UNet | ARC Lab, Tencent PCG | GitHub | arXiv |
| AnyDoor | 2023-12-17 | SD2.1 + DINOv2 + AnyDoor Projection | The University of Hong Kong, Alibaba TongYi Vision Intelligence Lab | GitHub | arXiv |
2024
| Model Name | Release Date | Parameters | Team | Open Source Link | Technical Report |
|---|---|---|---|---|---|
| BrushNet | 2024-03-11 | Dual branch Stable Diffusion (1.5 or XL) | ARC Lab, Tencent PCG | GitHub | arXiv |
| MS-Diffusion | 2024-06-11 | Stable Diffusion XL | Alibaba Group | GitHub | arXiv |
| MimicBrush | 2024-06-12 | Stable Diffusion 1.5 inpainting + Stable Diffusion 1.5 | The University of Hong Kong, Alibaba TongYi Vision Intelligence Lab | GitHub | arXiv |
| UltraEdit | 2024-07-07 | Stable Diffusion 1.5 or Stable Diffusion XL or Stable Diffusion 3 | Peking University | GitHub | arXiv |
| FLUX.1-dev-Controlnet-Union (InstantX) | 2024-08-14 | FLUX.1-dev ControlNet | InstantX | HuggingFace | None |
| FLUX.1-dev-ControlNet-Union-Pro (XLabs-AI) | 2024-08-20 v2: 2025-04-16 |
FLUX.1-dev ControlNet | XLabs-AI | HuggingFace v2 |
None |
| flux-ip-adapter (XLabs-AI) | 2024-08-20 v2: 2024-11-04 |
clip-vit-large-patch14 + FLUX.1-dev | XLabs-AI | HuggingFace v2 |
None |
| In-Context LoRA (IC-LoRA) | 2024-11-07 | 12 B (LoRA on FLUX.1-dev) | Tongyi Lab, Alibaba | GitHub | arXiv |
| flux-ip-adapter (InstantX) | 2024-11-12 | siglip-so400m-patch14-384 + FLUX.1-dev | InstantX | HuggingFace | None |
| FLUX.1 Redux | 2024-11-21 | 0.9 B (0.88 B siglip + 0.02 B MLP) | Black Forest Labs | HuggingFace | Blog |
| FLUX.1-Fill | 2024-11-21 | 12 B | Black Forest Labs | HuggingFace | Blog |
| AnyEdit | 2024-12-23 | 0.9 B (Stable Diffusion 1.5) | Zhejiang University | GitHub | arXiv |
| OminiControl | 2024-12-26 | 12 B (LoRA on FLUX.1-dev) | xML Lab, National University of Singapore | GitHub | arXiv |
2025
| Model Name | Release Date | Parameters | Team | Open Source Link | Technical Report |
|---|---|---|---|---|---|
| ACE++ | 2025-01-06 | 12 B (LoRA on FLUX.1-Fill) | Tongyi Lab, Alibaba | GitHub | arXiv |
| SmartFreeEdit | 2025-02-17 | LISA-7B + grounding_dino + BrushNet | Institute of Artificial Intelligence, China Telecom (TeleAI) | GitHub | arXiv |
| EasyControl | 2025-03-18 | 12 B (LoRA on FLUX.1-dev) | The Chinese University of Hong Kong | GitHub | arXiv |
| InfiniteYou | 2025-03-20 | Face ID Encoder + Projection Network + InfuseNet + FLUX.1-dev | Intelligent Creation Lab, ByteDance | GitHub | arXiv |
| UNO | 2025-04-03 | 12 B (LoRA on FLUX.1-dev) | UXO Team, Intelligent Creation Lab, ByteDance | GitHub | arXiv |
| DreamFuse | 2025-04-11 | 12 B (LoRA on FLUX.1-dev) | Intelligent Creation Lab, ByteDance | GitHub | arXiv |
| Insert Anything | 2025-04-22 | FLUX.1 Redux + FLUX.1-Fill (LoRA on 12 B FLUX.1-Fill) | Zhejiang University | GitHub | arXiv |
| Step1X-Edit | 2025-04-25 | 12 B | StepFun | GitHub | arXiv |
| In-Context Edit | 2025-04-30 | 12 B (LoRA on FLUX.1-Fill) | Zhejiang University | GitHub | arXiv |
| OminiControl2 | 2025-05-12 | 12 B (LoRA on FLUX.1-dev) | The Chinese University of Hong Kong | GitHub | arXiv |
| BAGEL | 2025-05-20 | Total 14 B, activate 7 B | Seed Team, ByteDance | GitHub | arXiv |
| FLUX.1-Kontext | 2025-05-29 | 12 B | Black Forest Labs | HuggingFace | Blog arXiv |
| OmniGen2 | 2025-06-16 | 3 B autoregressive + 4 B diffusion | Beijing Academy of Artificial Intelligence | GitHub | arXiv |
| XVerse | 2025-06-26 | 12 B (using LoRA on FLUX.1-dev) | Intelligent Creation Team, ByteDance | GitHub | arXiv |
| IC-Custom | 2025-07-03 | FLUX.1 Redux + FLUX.1-Fill (LoRA on 12 B FLUX.1-Fill) | ARC Lab, Tencent PCG | GitHub | GitHub |
| Qwen-Image-Edit | 2025-08-18 v2509: 2025-09-22 v2511: 2025-11-22 |
20 B | Qwen Team, Alibaba | HuggingFace | arXiv |
| Qwen-Image-ControlNet-Union | 2025-08-27 | Qwen-Image ControlNet | InstantX Team | HuggingFace | None |
| Qwen-Image-ControlNet-Inpainting | 2025-08-27 | Qwen-Image ControlNet | InstantX Team | HuggingFace | None |
| USO | 2025-08-27 | 12 B (using LoRA on FLUX.1-dev) | UXO Team, Intelligent Creation Lab, ByteDance | GitHub | arXiv |
| UMO | 2025-09-12 | Using LoRA on FLUX.1-dev or OmniGen2 | UXO Team, Intelligent Creation Lab, ByteDance | GitHub | arXiv |
| DreamOmni2 | 2025-10-09 | 12 B (using LoRA on FLUX Kontext) | The Chinese University of Hong Kong, JIA Lab | GitHub | arXiv |
| ContextGen | 2025-10-15 | 12 B (using LoRA on FLUX Kontext) | Zhejiang University | GitHub | arXiv |
| UniWorld-V2 | 2025-10-19 | 12 B (Edit-R1-FLUX.1-Kontext-dev) 20 B (Edit-R1-Qwen-Image-Edit-2509) |
Peking University | GitHub | arXiv |
| FLUX 2 | 2025-11-25 | 32 B | Black Forest Labs | HuggingFace | Blog Research |
| Z-Image-Turbo-Fun-Controlnet-Union | 2025-12-03 v2.1: 2026-02-26 |
Z-Image-Turbo ControlNet | Alibaba-PAI | HuggingFace v2.1 |
None |
| LongCat Image | 2025-12-05 | 6 B | LongCat Team, Meituan | GitHub | arXiv |
| RePlan | 2025-12-19 | Qwen2.5-VL-7B + Qwen Image Edit or FLUX 2 Klein | Tencent AI Lab, The Chinese University of Hong Kong, JIA Lab | GitHub | arXiv |
2026
| Model Name | Release Date | Parameters | Team | Open Source Link | Technical Report |
|---|---|---|---|---|---|
| UniPic-3 | 2026-01-09 | 20 B (Qwen Image) | Skywork | GitHub | arXiv |
| GLM Image | 2026-01-14 | 9 B autoregressive + 7 B diffusion | Zhipu | GitHub | Blog |
| FLUX 2 Klein | 2026-01-15 | 4 B 9 B |
Black Forest Labs | HuggingFace | Blog |
| TeleStyle | v1: 2026-01-28 v2: 2026-06-10 |
20 B (using LoRA on Qwen-Image-Edit) | Institute of Artificial Intelligence, China Telecom (TeleAI) | GitHub v1 GitHub v2 |
arXiv v1 arXiv v2 |
| FireRed-Image-Edit | v1: 2026-02-14 v1.1: 2026-03-03 |
20 B | Super Intelligence Team, Xiaohongshu | GitHub | arXiv |
| HiFi-Inpaint | 2026-03-04 | 12 B (using LoRA on FLUX.1-dev) | University of Chinese Academy of Sciences | GitHub | arXiv |
| A²-Edit | 2026-03-20 | ByteDance SigLIP image encoder + FLUX.1-Fill | Shanghai Jiao Tong University | GitHub | arXiv |
| JoyAI Image | edit: 2026-04-02 edit-plus: 2026-06-23 |
16 B | Joy Future Academy, JD | GitHub | arXiv |
| Boogu-Image-0.1 | 2026-06-16 | 10 B | Celia Lab, Huawei | GitHub | arXiv |
License
This repository is for informational purposes only. Each model has its own license — please refer to the respective open-source links for license details.
Contributing
Contributions are welcome! If you’d like to add a new model or update an existing entry, please submit a pull request.
Awesome Reference Image Generation
https://huan-yin.github.io/2026/07/16/awesome-reference-image-generation/