Saganaki22

HiDream O1 Image nodes + LoRA training for ComfyUI

48
11
89% credibility
Found May 17, 2026 at 49 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

HiDream_O1-ComfyUI is a free extension for ComfyUI that brings HiDream O1's powerful AI image generation directly into your creative workflow. You can create images from text descriptions alone, or guide the generation using up to 12 reference images for style and composition control. The extension supports multiple model variants optimized for different needsโ€”Full models for photorealistic detail, Dev models for faster, cleaner illustration work. You can also train your own custom style modifiers (LoRAs) using your own image collections, making it easy to teach the AI your preferred aesthetic. Everything runs locally on your own computer, giving you full control over your creative process.

How It Works

1
๐Ÿ” You discover HiDream O1

You hear about HiDream O1's impressive image generation and find it available as a ComfyUI extension.

2
๐Ÿ“ฆ You install the extension

You install HiDream O1 nodes through ComfyUI Manager or by cloning the repository into your custom nodes folder.

3
๐Ÿ’พ You download a model

You choose a model variant (Full for photorealism or Dev for faster, cleaner results) and download it to your ComfyUI models folder.

4
๐ŸŽ›๏ธ You build your workflow

You connect the model loader, add your text prompt, optionally attach reference images for style guidance, and chain in any custom style modifiers.

5
โœจ You generate your image

You click generate and watch as HiDream O1 creates your image step by step, with progress shown in real-time.

๐ŸŽ‰ You get your result

Your generated image appears, ready to use or refine further with additional edits.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 49 to 48 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is HiDream_O1-ComfyUI?

This brings HiDream's O1 image model directly into ComfyUI's node-based workflow. Instead of relying on their web demo, you run generation locally, feeding text prompts or optional reference images through a visual graph. It handles model loading across multiple precision levels (BF16, FP16, FP8) with automatic attention backend selection, plus a smoothing node that reduces visible seams on tiled outputs. For those experimenting with fine-tuning, it also bundles LoRA training nodes that follow the AI Toolkit recipe.

Why is it gaining traction?

The VRAM flexibility is the headline. FP8 runs on roughly 10-11GB of VRAM while the full model needs 18-20GB, covering most mid-range consumer GPUs. Reference image workflows are built-in natively, not awkwardly patched together. The smoothing node directly addresses a known upstream issue that produces grid artifacts in certain scenarios. Having both generation and training in one cohesive ComfyUI extension is uncommon.

Who should use this?

ComfyUI users who want HiDream O1 locally without abandoning their existing workflow. Artists exploring reference-driven image editing and style transfer. Developers training custom LoRAs for subject or style fine-tuning who prefer working inside ComfyUI rather than juggling separate training scripts.

Verdict

The credibility score of roughly 0.9% reflects the reality: 48 stars, minimal community documentation, and active upstream churn (PyTorch 2.9.x warnings, May 2026 breaking changes). This is bleeding-edge and experimental. The documentation is thorough, but test coverage and broader user validation are limited. Only jump in if you are comfortable troubleshooting edge cases and keeping pace with upstream changes.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.