more rabbits, but importantly I have been testing that SDXL is very sensitive to any tags that you got away with just leaving in for 1.5 loras.
My mirko lora got a lot better when I removed these two tags because it was fitting those as a requirement to put a lot of details on the character (probably would have got similar results on my old lora by including these two tags in all my gens, but it isn't obvious to).
Jump in the discussion.
No email address required.
I honestly have no idea what most of the settings on automatic1111 do. Like the basics sure but what's textual inversion or loras.
Jump in the discussion.
No email address required.
Textual Inversion is ghetto lora basically
Jump in the discussion.
No email address required.
More options
Context
Loras are very important, text inversion irrelevant.
Loras you often make yourself and are fine tunes of models, ie you take 50-100 images of a character and use training scripts.
That is how @TouchFluffyTails can teach AI what marsey catgirl looks like
@TouchFluffyTails wanna be a janny
Jump in the discussion.
No email address required.
More options
Context
More options
Context
SDXL seems too be way more sensitive too a single tag fricking up the whole prompt, doubly so pony
@iStillMissEd wanna be a janny so @iStillMissEd can doxx, hunt, kill cook and eat all the jannies who left @iStillMissEd chudded for these long days
Jump in the discussion.
No email address required.
More options
Context
Why is one square blank?
Jump in the discussion.
No email address required.
More options
Context
Jump in the discussion.
No email address required.
More options
Context
Hey sexy, why are you posting so quickly? 😊 You almost forgot to include
i wanna be a jannyin your comment 😈. Slowww down and remember to posti wanna be a jannynext time 😉 if that doesn't make sense stop by sometime and we can talk about it for a while 🥵Jump in the discussion.
No email address required.
More options
Context