Recently, I’ve experimented with FLux LoRA training by using a new configuration and training method using multiple defined datasets and I’ve had a few successes in creating some interesting LoRAs that have a hybrid style. This one turned out particularly great and was trained with datasets from <a target="_blank" rel="ugc" href="https://civarchive.com/models/921689/mystic-sumi">Mystic Sumi</a> and <a target="_blank" rel="ugc" href="https://civarchive.com/models/993138/ink-wash-fusion">Ink Wash Fusion</a>.The LoRA will trigger three different styles. Using "dinkwfsn style" and "dismrt style" both trigger similar styles, but different compositions that are a mix of the triggered styles from Mystic Sumi and Ink Wash Fusion. Starting a prompt with "an ink wash painting" will create a more traditional sumi-style image.

Mystic Fusion

Mystic-Fusion_v10.safetensors

v1.0

Recently, I’ve experimented with FLux LoRA training by using a new configuration and training method using multiple defined datasets and I’ve had a few successes in creating some interesting LoRAs that have a hybrid style. This one turned out particularly great and was trained with datasets from <a href="https://civitai.com/models/921689/mystic-sumi" rel="ugc noopener noreferrer nofollow" target="_blank">Mystic Sumi</a> and <a href="https://civitai.com/models/993138/ink-wash-fusion" rel="ugc noopener noreferrer nofollow" target="_blank">Ink Wash Fusion</a>.The LoRA will trigger three different styles. Using "dinkwfsn style" and "dismrt style" both trigger similar styles, but different compositions that are a mix of the triggered styles from Mystic Sumi and Ink Wash Fusion. Starting a prompt with "an ink wash painting" will create a more traditional sumi-style image.

Files with SHA256: 7d01f15a0f27be8ecc5016abb6526022b42cfba9b3fef00fed0384d9eb5bebd8