画質 高画質

. 『腹筋が六つに割れたチアリーダー』
(Cheerleader with six-pack abs)

0 0

. 『silver-haired girl, hard abs, very long ponytail, purple eyes, large breasts, hair ribbon』

0 0

. 『silver-haired girl, hard abs, very long ponytail, purple eyes, large breasts, hair ribbon』

1 1

. 『silver-haired girl, hard abs, very long ponytail, purple eyes, large breasts, hair ribbon』

1 3

. 『silver-haired girl, hard abs, very long ponytail, purple eyes, large breasts, hair ribbon』

0 1

skipping the clamping-by-%ile and just denormalizing CFG20's latents to CFG7.5's abs().max() is very similar to reducing cond_scale, but not quite.

think it works out something like:
(uncond + (cond - uncond) * 20)/(20/7.5)
versus
uncond + (cond - uncond) * (20/(20/7.5))

0 0

99.99999999999%ile dynthresh
I think this shows that "clamping out n%ile outliers" is only important when you have excessive outliers. the rest of the battle is "what range of values do you span". hence denormalizing the latents to CFG7.5's abs().max(), gives us a safer range.

0 2

each step: compute CFG7.5. for each channel: flatten, center on mean, grab abs().max()
compute CFG20. each channel: flatten, center on mean, compute 99.x%ile of abs(), pick larger of %ile or of CFG7.5's max. clamp channel by that. normalize. multiply by CFG7.5's max.
code coming

0 6

. 『silver-haired girl, hard abs, very long ponytail, purple eyes, large breasts, hair ribbon』

0 1

. 『silver-haired girl, hard abs, very long ponytail, purple eyes, large breasts, hair ribbon』

0 2

ninja boy ,abs+年齢、18歳から入力してみたが、やはりだめで、下げすぎると男の娘忍になる(画像片方は14歳指定)。

6 7

. 『blonde-haired girl, hard abs, very long ponytail, blue eyes, large breasts, red hair ribbon』

0 1

. 『silver-haired girl, hard abs, very long ponytail, purple eyes, large breasts, hair ribbon』

0 1

Commission done by
Haojie lets his enemies beat his abs presumptuously. Although it seems to be painful, he enjoys it.

100 614