Evie Lee Mikomin [flux] v1版本v1.0_512x (ID: 792600)

First attempt at a LoRA on CivitAI. Trained locally using sd_scripts @ 1024x1024 resolution. 20 image dataset trained for 70 epochs. Works best for me using the dpmpp_2s_ancestral beta sampler in ComfyUI @ 0.8 strength with 2.5 - 2.8 Flux guidance.

https://www.instagram.com/evieleemikomin/

https://www.patreon.com/mikomin

https://x.com/MikominCosplay

描述:

Exact same settings as V1.0 trained on same dataset just @ 512x512 instead of 1024x1024. Somethings are better in this version but generally I prefer v1.0.

训练词语: mik0cos