My first lora intended to be a proof of concept more than anything. A better version will likely come later. Intended to produce images of NCAA gymnasts on balance beam, uneven bars, and floor. You might want to keep the lora weight around 0.4 to 0.5

Gymnasts and Gymnastics

Gymnasts_and_Gymnastics_V3.safetensors

v3.0

v2.0

A much better version trained on many more images. Some things to take note of:<ul><li>Works best with guidance between 4 and 6</li><li>Use DPM++ SDE</li><li>Clip Skip 2</li><li>Fingers and eyes are still a bit iffy here and there, so adjust your negative prompts accordingly</li></ul>

Files with SHA256: d4a77bd844c634358c22b98668e4a4e8f3c07e46808f91930edba46dbf843bb2