This LoRA is not mine, The author is lodestones, I'm just reprinting it for general purposes.
Trained with Danbooru art.
https://huggingface.co/lodestones/Kroma
This LoRA was produced from a K2 (Krea 2) fine-tune checkpoint and converted into the ComfyUI format: LoRA adapters use the native diffusion_model.* key naming, and the fully-fine-tuned RMSNorm / modulation tensors are stored as .diff deltas against the base model. Both are applied by ComfyUI's standard LoraLoaderI will delete it if the original author requests it.
Description
silveroxides/Kroma-LoRA-Extract
FAQ
Comments (6)
On 0.2, struggling to get anything close to your preview image. On comfy and with Neoforge. can you share a full workflow, or catbox link?
Workflow added
@Hikarias wow, major difference with the Wan 2.1 vae! that was the missing link, so much better results vs the recommended qwen_image_vae on HF. Do you always use the Wan Vae for all styles and use cases on Krea?
This is a masterpiece! Completely eliminates the AI-generated glitter effect compared to version 01.
I noticed that when using a value of 0.2, the characters appear younger. Is this 0.2 parameter specific to anime only?
I'm very curious about the tools or scripts used to extract this LORA. I saw quite a few of these non-fixed ranks, with names like "fro xxxx".