HiDream-O1-Image_clear

• Guide (External site): English | Japanese
• Civitai

Samples and Prompts

change to anime illustration

change to black and white manga artwork

A full-length portrait of a man in a fantasy world, holding a long sword and wearing plate armor, beaming with excitement as he looks forward to his upcoming adventures. The background features a medieval village with a wheat field in the distance.

A finetuned HiDream-O1-Image model, specifically optimized for generating clearer and sharper outputs.

Comparison with HiDream-O1-Image (Original)

HiDream-O1-Image_clear_v1 | HiDream-O1-Image (Original)

Model Format and Quality

Generally, model size and image quality are proportional.

• FP32 > FP16 ≒ BF16 > INT8_ConvRot_HQ > MXFP8 > FP8_scaled > NVFP4 > FP8

If you prioritize image quality, please use the FP16 / BF16 format.

vs_FP32	MAE_Similarity	SSIM_Similarity	Size
FP32	100%	100%	35.2 GB
FP16	92.7 %	87.6 %	17.6 GB
BF16	92.6 %	87.5 %	17.6 GB
MXFP8	92.0 %	83.8 %	10.3 GB
FP8_scaled	91.6 %	84.8 %	10.1 GB
FP8	85.8 %	59.1 %	8.8 GB
NVFP4	87.2 %	71.9 %	6.8 GB
INT8_ConvRot_HQ	92.6 %	88.6 %	10.6 GB

If you need speed or short on VRAM, choose INT8_ConvRot_HQ.

MXFP8, FP8_scaled, and NVFP4 are currently not recommended, as they are inferior to INT8_ConvRot_HQ in both quality and speed.

For a comparison of image quality and generation time by format, please see this page.

ComfyUI workflow

The ComfyUI workflows are embedded in these images.

Text to Image Anime (simple)

Text to Image Photoreal (reduce block noise)

Image to Image

Image Edit

To use these workflows, please install ComfyUI-uit-hidream-o1 custom node.

Recommended Settings

• Sampler: euler
• Scheduler: normal
• Steps: 12 (8-24)
• CFG scale: 4 (3-5)
• noise_scale: 6.5 (6.2-7.5)

---

Tips

• HiDream-O1-Image and UiT architecture: English | Japanese
• FP / BF Format Image Quality and Generation Time: English | Japanese
• Image Generative AI model License: English | Japanese

---

License

• MIT

---

Update History

2026.706

• Added INT8_ConvRot_HQ model

2026.6.12

• Added Model Format and Quality

2026.5.28

• Re-uploaded FP32, FP16, BF16, changed from Comfy-Org format to the original HiDream-O1-Image format. It can now be used with tools other than ComfyUI as well.

2026.5.27

• Uploaded HiDream-O1-Image_clear_v1_FP8_scaled.safetensors

2026.5.20

• Repository opened.