dall-e mini Vore thread

[EquusMinor]

https://huggingface.co/spaces/dalle-mini/dalle-mini

I haven’t seen this here yet.

[HungryNacho]

I didn't think a robot could actually generate vore drawings, well you learn something every day, this kind of looks like how I see art when I'm high on some marijuana edibles or if I having a really weird dream and I'm viewing art in it lol

[EquusMinor]

Whenever I put “vore” in it spits out these amorphous belly creatures. It appears that’s what it thinks of us…

[EquusMinor]

More blob critters

[GeneticMess]

I messed with this AI a month or so ago, with the right combination of words - such as "[character] with a fat belly" - you can manipulate the thing into giving you vore-adjacent imagery. As well as fat fetish art, so BBW-lovers rejoice. Some of this stuff turns out pretty good, as long as you just... don't look too hard at the faces.

[EquusMinor]

It doesn’t like Elon very much. He always looks like this when you run his name.

[Brainfood]

I messed with this AI a month or so ago, with the right combination of words - such as "[character] with a fat belly" - you can manipulate the thing into giving you vore-adjacent imagery. As well as fat fetish art, so BBW-lovers rejoice. Some of this stuff turns out pretty good, as long as you just... don't look too hard at the faces.

I am enchanted with these results. I never would have imagined it'd come this close to making anything that was nice to look at. I love this fucking thing.

[GeneticMess]

I am enchanted with these results. I never would have imagined it'd come this close to making anything that was nice to look at. I love this fucking thing.

I've gotten better results than I thought I'd be able to, given the filters in place. But a fat gut is technically SFW, so the AI sees nothing wrong with making one. If only we could get our hands on one of the more advanced image-gen AI out there right now like dall-e 2 AND circumvent the no doubt stricter policies set in place to prevent more stuff like this.

[EquusMinor]

I’m wondering why it presented as a multi opening solution.

[Chessa]

There's a LOOOT of potential with AI and vore-centred content, we've already gone down the route of AI generated vore stories, but ai generated vore art is still a new one to really fill out.

Dall-E 2 and Dall-E mini have potential (Dall-E mini is Crayon), Google imagen will be live soon, but I dunno.

The others are Discord-based and can't be used in this manner I don't think.

[Chessa]

Oi, A bit of a heads up, peeps, Stable Diffusion is a newish bot that starting to come out that will be powerful and also open source, currently it can generate nudity as an example, and it would likely be very simpleish for an AI developer (Or someone with enough willingness to learn ML (Machine Learning) to modify it using the source code to generate vore content, it's about to enter closed beta, but a full released is planned at some point when it can be improved enough for the Stable team to be satisfied with the base model.

[Redicicle]

Probably some of the best fully AI-generated internal shots I've seen. Prompt engineering seems to be pretty important here. Small differences in wording, formatting, or order matter a lot.

[Redicicle]

Big belly
Thus far, attempts to get animals to eat people, mid-swallow shots, or translucent cutaway bellies with a coherant person inside have failed.

[Redicicle]

AI generated stomach Textures that might be useful to manip artists

[Redicicle]

The new diffusion models on Wombo Dream can also be used to generate vorish content. Though for some reason they are highly likely to fail completely when given NSFW inputs (basically it detects if it's about to generate realistic porn and nopes out) and take a long time and forcibly change artstyle to less realistic styles when given NSFW inputs combined with image prompts.

[Redicicle]

Wombo Dream images

[LightDragon]

It doesn’t like Elon very much. He always looks like this when you run his name.

I think this is intentional. The developers of Dall-e are well aware of the harm their AI could generate if someone used it to create deep fakes (like "Elon musk kills a baby with a gun").
And they have taken actions to prevent this kind of use. I think making the face unrecognizable is one of them.

Otherwise, the AI is promising, but at the moment too limited to actually do things, especially vore-wise.

[fieldmousse]

i got access to DALLE-2 the other day and tried generating some vore images. I didn't really have any success and quickly ran out of credits.
If anyone else tries to though note that if you put "X eaten Y" then it's pretty likely to combine the two like as in "you are what you eat"
Tried to do a simple "snake that has eaten a mouse" and it would give me a snake with a mouse head.