HiDream-E1

HiDream-E1

HiDream-E1 is an image editing model built on HiDream-I1.

For more features and to experience the full capabilities of our product, please visit https://vivago.ai/.

Project Updates

• July 16, 2025: We've open-sourced the updated image editing model HiDream-E1.1. It supports dynamic resolution and is better in image quality and editing accuracy compared to HiDream-E1-Full.
• May 28, 2025: We've released our technical report HiDream-I1: A High-Efficient Image Generative Foundation Model with Sparse Diffusion Transformer. Please use the Bibtex below to cite the paper.
• April 28, 2025: We've open-sourced the instruction-based image editing model HiDream-E1.

Models

We offer the full version of HiDream-E1. For more information about the models, please refer to the link under Usage.

[!NOTE] The code and model are under development and will be updated frequently.

Quick Start

Please make sure you have installed Flash Attention and latest Diffusers. We recommend CUDA versions 12.4 for the manual installation.

pip install -r requirements.txt
pip install -U flash-attn --no-build-isolation
pip install -U git+https://github.com/huggingface/diffusers.git

For HiDream-E1.1, you can run the following script to generate images:

python ./inference_e1_1.py

For HiDream-E1-Full, you can run the following script to generate images:

python ./inference.py

[!NOTE] We add a refine_strength parameter to the pipeline to control the balance between editing and refinement stages. During the first (1 - refine_strength) portion of denoising steps, the model performs the main editing operation. The remaining refine_strength portion of steps uses HiDream-I1-Full for img2img refinement to enhance the final result. Set refine_strength to 0.0 to disable refinement.

[!NOTE] The inference script will try to automatically download meta-llama/Llama-3.1-8B-Instruct model files. You need to agree to the license of the Llama model on your HuggingFace account and login using huggingface-cli login in order to use the automatic downloader.

[!NOTE] For HiDream-E1.1, the model accepts direct instructions like convert the image into a Ghibli style and prompt refinement is no longer needed.

For HiDream-E1-Full, the model accepts instructions in the following format:

复制

Editing Instruction: {instruction}. Target Image Description: {description}

Example:

复制

Editing Instruction: Convert the image into a Ghibli style. Target Image Description: A person in a light pink t-shirt with short dark hair, depicted in a Ghibli style against a plain background.

To refine your instructions, use the provided script:

bash复制

python ./instruction_refinement.py --src_image ./test.jpeg --src_instruction "convert the image into a Ghibli style"

The instruction refinement script requires a VLM API key - you can either run vllm locally or use OpenAI's API.

Gradio Demo

We also provide a Gradio demo for interactive image editing. For HiDream-E1.1, you can run the demo with:

python gradio_demo_1_1.py 

For HiDream-E1-Full, you can run the demo with:

python gradio_demo.py 

Evaluation Metrics

Evaluation results on EmuEdit and ReasonEdit Benchmarks. Higher is better.

License

The code in this repository and the HiDream-E1 models are licensed under MIT License.

Citation

@article{hidreami1technicalreport,
  title={HiDream-I1: A High-Efficient Image Generative Foundation Model with Sparse Diffusion Transformer},
  author={Cai, Qi and Chen, Jingwen and Chen, Yang and Li, Yehao and Long, Fuchen and Pan, Yingwei and Qiu, Zhaofan and Zhang, Yiheng and Gao, Fengbin and Xu, Peihan and others},
  journal={arXiv preprint arXiv:2505.22705},
  year={2025}
}