Mutual oral pleasure? Let's go. Front, side, and back 😊

If you ever tried to prompt this pose you probably know how bad models are at it, so this should make it way easier.

Update 26.08.2026: Anima LoRA is in the correct file group now. Should work with onsite gen. thx @PhantomF4H

Krea2

is a bitch it seems. ass view is hit or miss. The other two views work fine tho, after now 4500 steps..

Anima

can almost do it without LoRA 😉

six9, 69face, 69, 1boy, 1girl, {oral, fellatio|handjob}, solo focus, girl on top, penis, testicles

six9, 69side, 69, 1boy, 1girl, {oral, fellatio|handjob}, hetero, from side, girl on top, lying, groping

six9, 69ass, 69, 1boy, 1girl, male pov, from behind, on all four, reverse fellatio, girl on top, hanging breasts, pussy, anus, {oral, fellatio|handjob}, {cunnilingus|groping}

Illu V2.0

First time training with different aspect ratios and actual buckets, and it turned out pretty good: This version knows all perspectives and even horizontal and vertical image ratios.

Start at 0.5 strength for horizontal, more for vertical.

Trained on 3:4, 1:1, 4:3.

I abandoned the specialized perspective tags so it learns and works with natural tags. Just use the main trigger and tag as you would, like:

(front:)

six9, 69, 1boy, 1girl, oral, {fellatio|handjob}, solo_focus, penis, testicles

(side:)

six9, 69, 1boy, 1girl, oral, fellatio, hetero, from_side, girl_on_top, lying

(back:)

six9, 69, 1boy, 1girl, from_behind, reverse_fellatio, pussy, anus, oral, fellatio, {cunnilingus|groping}

V1.0

Examples:

six9, 69face, 1boy, 1girl, fellatio, hetero, (reverse_blowjob:0.8), close-up, solo focus, breasts pressed, depth of field

six9, 69side, side view, girl on top, blowjob, 1boy, 1girl, oral, fellatio

six9, 69ass, fellatio, oral, 1boy pov, 1girl, rear view, low view, pussy juice, anus, spreading another's pussy, hanging breasts, girl on top, kneeling

V1.side

I re-trained the side part only. This version knows only

69side

although it can sway to front and back. Use "side view" to be sure.

And it was trained exclusively on 4:3 (1152x896)! Which just makes sense in this case 😊

0.6 to 0.7 strength.