Instructions to use HiDream-ai/HiDream-O1-Image with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use HiDream-ai/HiDream-O1-Image with Transformers:
# Load model directly from transformers import AutoProcessor, AutoModelForMultimodalLM processor = AutoProcessor.from_pretrained("HiDream-ai/HiDream-O1-Image") model = AutoModelForMultimodalLM.from_pretrained("HiDream-ai/HiDream-O1-Image", device_map="auto") - Notebooks
- Google Colab
- Kaggle
add a true img2img/edit mode
#6
by Sambler - opened
Could you add a true img2img/edit mode with strength control for HiDream O1?
The current image_1 input works more like a reference image. For interior rendering workflows, we need to preserve the exact source image layout, geometry, camera angle and materials, similar to img2img denoise strength in SD/FLUX.
Suggested controls:
- edit_strength / denoise_strength
- image_guidance_scale
- structure_preservation
- preserve_image_layout mode