Jenna Coleman - Trained for Flux
Description
FAQ
Comments (2)
hello, I've tried a few of your loras with bad results. so thought I'd ask, which version of flux, sampler, scheduler do you use for your samples.
I train all my models on the original fp16 Flux checkpoint. Make sure to include the trigger word in your prompt and set the lora strength between 1.2 and 1.5 for best results. If you're using an fp8 version of Flux or mixing LoRAs, I'd recommend a minimum strength of 1.5 for the character LoRA and don't go over 1.0 for styles.