Assuming Direct Control: AI Art Thread

Now using the Toilet concept LORA:
Early on I start with something like this:
straight on view. One (30yo:1.3) woman with (messy:0.3) (long) (dark red hair) and (light blue eyes) toilet bondage, <lora:toiletGirl:1.2>. She has a look of (utter despair:0.3) and is (sobbing:0.3). Her entire body is in the frame. (torn clothes:0.1) (red dress:1.1). (Tiny pupils:0.1)
(high quality), cartoon, colorful, (detailed faces), explicit, wearing a (detailed) (slave collar:1.3)
00905-1407578785.png

00909-1110450761.png

00930-1782320351.png

00935-2753255865.png

00938-1151631507.png

00943-1606413937.png

00953-1044877561.png


By the end it is this:
straight on view. One (30yo:1.3) woman with (messy:1.3) (long) (dark red hair) and (light blue eyes) toilet bondage, <lora:toiletGirl:1.2>. She has a look of (utter despair:1.3) and is (sobbing:1.3). Her entire body is in the frame. (torn clothes:1.8) (red dress:1.5). (Tiny pupils:1.3), (cum:1.2), (cumpool:0.5), (after rape:1.2)
(high quality), cartoon, colorful, (detailed faces), explicit, wearing a (detailed) (red) (slave collar:1.3), nipples

I start with a few defaults in the Negative Prompts:

(body out of frame) bad art, watermark, cgi, (bad anatomy), (extra limbs), (extra legs), (extra arms), disfigured, (deformed:1.3), (ugly:1.3), poorly drawn hands, poorly drawn feet, poorly drawn face, out of frame, mutation, mutated, blurry, text, multicolored hair, ponytail, cut-in, (worst quality, low quality:1.4),
 
So been experimenting with some more Non-Human (furry) AI Art. Even with Models that focus on Furry Content, it do be a struggle. Still, can get some nice images occasionally.


00023-3828566079.png
00006-2579587699.png
00076-1005501241.png
00079-1005501244.png
00010-2579587704.png

00035-1957085566.png

00034-1957085565.png

00139-3717461348.png

00132-3064614193.png
 
Impregmaniac said:
Is this all still Nightcafe? Because goddamn those are good!
Click to expand...
Nah, I mostly abandoned Nightcafe due to the rampant censorship. (Tiddies deserve to be free!).

I still steal some of their themed prompts though if I want a particular style.

I have Stable Diffusion with about 12 different Models loaded onto it. With the right model, LORA(think like a mini-model to refine an image a specific way), and time(high resolution images and animations take potentially *hours*)...I can make almost anything I want
 
BJ.png

This one has a lot of flaws but a year or two and it will be easy to generate photorealistic porn. Sometimes AI is frightening
 
You must log in or register to reply here.
Back
Top