ilkerzgi Releases Over 1,500 Krea 2 Style LoRAs: A Massive Open-Source Style Library
ilkerzgi has released over 1,500 Krea 2 Style LoRAs on Hugging Face, spanning 7 categories from illustration to photographic. Each LoRA comes ready to use in ComfyUI.
Community member ilkerzgi has released over 1,500 Krea 2 Style LoRAs on Hugging Face, making it one of the largest open-source style collections available for the model. Each LoRA comes with both original safetensors and ComfyUI-compatible versions, ready to use out of the box.
The full collection is available at ilkerzgi/fal-Krea-2-Style-LoRAs.
What's Included
The collection spans 7 categories with 1,503 individual style LoRAs:
| Category | Count |
|---|---|
| 🎨 Illustration | 619 |
| 🖌️ Painterly | 289 |
| 📷 Photographic | 173 |
| 🖥️ Graphic | 126 |
| 🎬 Cinematic | 101 |
| 🏗️ 3D & Render | 97 |
| ✏️ Drawing | 98 |
Each LoRA has its own Hugging Face repository with full documentation, including usage examples, sample images, and the recommended trigger word for activation.
Background
The author describes this project as deeply personal. These LoRAs were trained on styles collected over several years — saved favorites, personal generations, and curated inspirations. Some styles were trained with as few as 5 images, others with around 10. After exploring Krea 2's LoRA training capabilities and seeing positive community feedback, ilkerzgi trained the full archive and released everything open-source in one batch.
"While exploring Krea 2, I trained a few LoRAs just to expand the styles I already had; once I saw they were getting genuinely good feedback, I trained my whole archive the same way, and the next day, once they were all ready, I released them open-source. So it really was a completely spontaneous thing."
How to Use in ComfyUI
All LoRAs are natively supported in ComfyUI. Download the ComfyUI-compatible safetensors from any LoRA's repository or from the collection index and load them through the standard LoRA loader node:
- Pick a style from the collection index
- Download the comfy safetensors file from the style's repo
- Place it in your
ComfyUI/models/loras/directory - Load the LoRA using the standard LoRA loader node
- Add the style's trigger word at the end of your prompt
- Recommended LoRA scale: 1.0 to 1.25
Make sure you have updated ComfyUI to the latest version first (update guide).
Comparison Across Models
The author ran an interesting comparison using a Bold Stripe Pattern Render LoRA (5 training images, 100 steps) across 8 different base models to show how Krea 2 handles low-shot style LoRAs compared to alternatives like FLUX.2 dev, Klein 9B, and others. The results demonstrate that Krea 2 achieves strong style consistency even with minimal training data.
License
These models are released under the Krea 2 Community License.