Lets sum it up, style training on Photography with a bigger database via Loras is a pain, chose the version you like the most. I won't spent more time on this :DIn general training styles work best on 10-20 images max, singlecaption with mixed no caption, or no caption at all seems also very good, but yeah small datasets everything else should be finetuned

epiCFluX-Photo LORA

FluXPhoto.safetensors

V4-LastEvo

v3-unique

v2-Mixed

V2-GPTCap

v2-JoyCap

V1-JoyCap

V1-SingleCap

Experimental&nbsp;Used CivitAi Lora-Training for Flux1.D on Photo-Dataset tagged in natural LanguageStrength: 3 (seems to kick in)

Files with SHA256: 336187ed0456d9bd9f689b7ae4f35737635001234881bfb54a05f30a8769804b