Trained on 20 images for 1700 steps, no keywords necessary. Best results are between 0.8-1.0 strength using Flux1 Dev, guidance around 4 and steps around 25-30.

Plug and play, trigger words or keywords do not have any effect