OmniGen2: Exploring Advanced Multimodal Generation

1. Tutorial Introduction

OmniGen2 is an open source multimodal generation model released by the Beijing Academy of Artificial Intelligence (BAAI) on June 16, 2025. It aims to provide a unified solution for a variety of generation tasks, including text-to-image generation, image editing, and context generation. Unlike OmniGen v1, OmniGen2 designs two independent decoding paths for text and image modalities, using non-shared parameters and separate image segmenters. This design enables OmniGen2 to build on existing multimodal understanding models without having to re-adapt VAE inputs, thereby retaining the original text generation capabilities. Its core innovation lies in the dual-path architecture and self-reflection mechanism, which has become a new benchmark for current open source multimodal models. The relevant paper results are "OmniGen2: Exploration to Advanced Multimodal Generation".

The computing resources of this tutorial use a single RTX A6000 card, and the English prompts are currently more effective.

2. Effect display

Some examples of effects with OmniGen2:

3. Operation steps

1. Start the container

2. Usage steps

If "Bad Gateway" is displayed, it means the model is initializing. Since the model is large, please wait about 2-3 minutes and refresh the page.

The first example is image description, the second and third examples are viz images, and the remaining examples are image editing.

Specific parameters:

• Height: height.
• Width: width.
• Text Guidance Scale: Text guidance scale.
• Image Guidance Scale: Image guidance scale.
• CFG Range Start: Range start.
• CFG Range End: Range end.
• Scheduler: Scheduler.
• Inference Steps: Inference steps.
• Number of images per prompt: The number of images per prompt.
• Seed: seed.
• max_input_image_side_length: Maximum input image side length.
• max_pixels: Maximum pixels.

result

4. Discussion

🖌️ If you see a high-quality project, please leave a message in the background to recommend it! In addition, we have also established a tutorial exchange group. Welcome friends to scan the QR code and remark [SD Tutorial] to join the group to discuss various technical issues and share application effects↓

Citation Information

The citation information for this project is as follows:

@article{wu2025omnigen2,
  title={OmniGen2: Exploration to Advanced Multimodal Generation},
  author={Chenyuan Wu and Pengfei Zheng and Ruiran Yan and Shitao Xiao and Xin Luo and Yueze Wang and Wanli Li and Xiyan Jiang and Yexin Liu and Junjie Zhou and Ze Liu and Ziyi Xia and Chaofan Li and Haoge Deng and Jiahao Wang and Kun Luo and Bo Zhang and Defu Lian and Xinlong Wang and Zhongyuan Wang and Tiejun Huang and Zheng Liu},
  journal={arXiv preprint arXiv:2506.18871},
  year={2025}
}