Getting the Best Out of My LoRAs

I’ve noticed something while watching people use my models through the Civitai online generator... sometimes the results don’t quite match what I get locally when testing. That made me feel a bit guilty, because I want everyone to experience the LoRAs at their full potential.

For transparency: I test everything locally in ComfyUI with a really barebones Text2Img workflow. No upscalers, no Adetailer, no heavy post-render enhancements. Just the LoRA and my base setup:

• Flux Dev (22GB)

• Zer0intVision ClipL

• Google T5-v1_1 XXL encoder (fp16 only)

That’s it. Clean and simple and the results you see on my page all come from this kind of minimal workflow.

Now, here’s the part I need to stress:
I don’t have any “real” technical knowledge. Everything I know comes from a huge pile of trial and error with settings I only half understand most of the time. I’m by no means a pro, or some kind of authority in the AI field. Since I rarely (if ever) use the on-site generator, I honestly can’t guide people very well on how to get the best out of it.

But because I want you to have the same (or better) results I get, I’ll be sharing (Later this week) my ComfyUI workflow file along with explanations of the settings I use. Hopefully it gives you a clear starting point... something easy to replicate without needing a crazy setup.

At the end of the day, my goal is to keep LoRAs accessible and fun. Load them, play with them, and see where they take you. Thanks again to everyone supporting, experimenting, and pushing my models further, you’re the ones that make this exciting. ❤️