Will delete if asked!
PondBs style, I hope i did a good job on it! be sure to react and leave input if you have any!
Description
Comments (7)
Sadly, it's not a style; it forces scenes into place you don't even describe. Maybe try it with a v2 and fix a lot of the text and the multiple-panel issues.
Thank you! I will give that a try!
I'm still kinda new to lora training so I appreciate the feedback.
@futaguru Glad to hear. As far as I know, to train a style, you should try to have at least around 30-50 pictures and describe them as well as possible. You also need different poses, characters, body types, and so on, so that the LORA knows what is what, so to speak. But I have no experience in it, and its only something I read, because I wanted to know how it works. If you manage to nail this style, it would be awesome!
@anotherstabledif1714 ah ok! 30 - 50 sounds waaay better than what I was working with! I'll do more research and I'll get this style down!
I am terrible at making LORAs too, but I think this one is for sure overcooked. It usually tries to recreate 1:1 scenes from your references. I think you trained it with too many steps.
Ah ok! Yeah I think imma retry making this one again and I'll be sure to try it with way less steps.