Handsome Lad

Handsome Lad | Flux.d | SD3 - clips_t5xx medium

I trained this LoRA first for SD3 medium.

Then I decided to carry on with Flux. In both cases this is the first LoRA I trained for those models. Flux version works better - as it is very versatile for many types of images. It tends to render realistic-esque style, which may not always be the case I know, but it's good for Schnell to escape from the usual CG basic style it usually renders.

• For Flux, works on Dev and Schnell, its effects start even at 0.1, but something around 0.6 to 0.8 is a good amount especially when using other LoRAs.

• For SD3 use only medium, or maybe full. Only seems to work on SD3 medium base model.
  

I'm also adding a small dataset which were whether sample images during training or images generated with the LoRA.

For SD3 medium, it is sort of limited.