Like Ra's Naughty Forum

@Like Ra, I hear you like Emma Watson in latex? 😁 

Ok ok, I'll stop now.

Figuring out the right tags for certain effects by browsing https://lexica.art/ is quite interesting.

But sometimes, there's a third leg out of nowhere.

Brilliant! Keep’em coming!

I installed Stable Diffusion on my new work laptop, it has an NVIDIA Quadro card with 4 gigs of memory and I get out of memory errors 

That's on Ubuntu 22.04, it has intell graphics as well so maybe there is a way to run everything on the Intell graphics and grant the Nvidia to Stabel Diffusion. At least I hope so, couldn't find anything online.

(01 Sep 2022, 19:04 )Anne Wrote: [ -> ]I installed Stable Diffusion on my new work laptop, it has an NVIDIA Quadro card with 4 gigs of memory and I get out of memory errors 

That's on Ubuntu 22.04, it has intell graphics as well so maybe there is a way to run everything on the Intell graphics and grant the Nvidia to Stabel Diffusion. At least I hope so, couldn't find anything online.

Hard to tell if you will be successful with a laptop. There are not many Quadro cards with enough memory from Nvidia. From the 8 Quadro RTX Mobile cards the 5000, 6000 and A5000 should work. Rest misses the memory requirements.
As per Stable Diffusion Github page:
The model is relatively lightweight and runs on a GPU with at least 10GB VRAM.

Stable Diffusion model takes 6.5GB VRAM. I’m going to buy a new graphic card, and I’m thinking about 6700XT with 12GB. 6600 has only 8GB. My current one has 2!!!

Then it will have to wait until I upgrade my pc. So in a year or 3...

I'll keep posting some of the weirder outcomes as well.

One of the most obvious limitations/ways to screw it up right now seems to be that whenever you choose image dimensions different from "square", the AI tends to throw in duplications/repetitions of limbs or even facial features.

Some variartions of

Quote:Jessica Rabbit, red Latex, Dominatrix

[attachment=55896][attachment=55897]
[attachment=55898][attachment=55899]

Quote:Cyberpunk nuns. latex

[attachment=55900][attachment=55901]
[attachment=55902][attachment=55903]

The AI does not deal well with "bondage" or "ballgag" in any combination or variation thereof.
I just hope some Kinkster will take the time and effort to train a model set by running a bunch of fetish sites through the AI. We all know that porn/adult themes have been the driving force behind all great inventions like VHS and internet streaming tech. Get crackin'.

[attachment=55912][attachment=55913]


And some random ones I don't quite remember the prompts for, but they kinda look cool.
Interestingly enough, the fully nude ones were not prompted for, the AI just went there.

[attachment=55910][attachment=55911]
[attachment=55904][attachment=55905]
[attachment=55906][attachment=55907]
[attachment=55908][attachment=55909]

Not exactly a fetish (well, maybe someone's fetish), but playing around with this bronze statue prompt just blew me away. I'm sure there's someone out there having a thing for muscular women.

SD is pretty good at anatomy.

[attachment=55917][attachment=55918]
[attachment=55919][attachment=55920]

Well, most of the time.

[attachment=55921]

Bonus: Muscular Men? Wolverine?

[attachment=55922]

Weird Wolverine?

[attachment=55923]

Quote:But Stable Diffusion has also been used for less savory purposes. On the infamous discussion board 4chan, where the model leaked early, several threads are dedicated to AI-generated art of nude celebrities and other forms of generated pornography.

Emad Mostaque, the CEO of Stability AI, called it “unfortunate” that the model leaked on 4chan and stressed that the company was working with “leading ethicists and technologies” on safety and other mechanisms around responsible release. One of these mechanisms is an adjustable AI tool, Safety Classifier, included in the overall Stable Diffusion software package that attempts to detect and block offensive or undesirable images.

However, Safety Classifier — while on by default — can be disabled.

Hmmm... How?