## HimawariMix mix models

These models are merged from style LoRAs into the himawarimix model.

### V2:

Parameters used when training the LoRA models:
* **resolution:** 1536x1536
* **learning_rate:** 1e-04
* **network dim, alpha:** 96,48
* **caption dropout:** 0.08

**v2.1b**:

![alt text](./sample-images/v2-sample1.png)