@Derteig I don't think I can. That was an experiment with an API that now has added censorship, and I don't have any record of the prompts I used. It was just a one-off because someone requested it (might have been you).

@bhutch125 no, Grok definitely wouldn't do something like this. Even if you could get it to try, the gore would be more... gelatinous. The details are in the clip's metadata (the QuickTime "comment" field).

@aliceflipper apparently, not all simulated girls have simulated ribs. I'll forward your discovery to the relevant authorities.

@kaywew there are a few. I don't talk about them in public.

@LoudPurple ...says the incestuous pedophile

@BDTGG that's exactly it :)

@BDTGG that one really stumped me in terms of not being able to figure out how to do what I intended. I might have a few more of these "winner" versions though.

@wetkoala happy accidents, lol

@Krommer this is a dump of the few results I was able to get with this theme. I consider it a dead end. Upside down people always confuse the AI. So does violence and gore. When you put them together it's almost always a disaster. Doesn't stop me from trying once in a while...

@allenhenry_24 these are an experiment to see how far I can push a censored AI. Low to high is probably out of the question. This only works because the content filtering often can't tell what's going on when there's a coating of blood. I assume it changes the color/shading profile it's trained to spot.

@Krommer it was supposed to be

@Derteig everyone knows this except for the AI, and the AI can't hear you :)

@wetkoala maybe. I'm more inclined to think that normies will increasingly have their way as content filtering gets more sophisticated.

@Dr_livormortis you think? Thank you, professor.

@Chloeirel_ DMs... but you would have to show your face 🫡😊

@Derteig it's not really my thing.

@m82717919 there doesn't seem to be any way to stop limbs from magically reappearing.

@lostinthehead unfortunately, the heart always looks super fake. With the models currently available, it's better left to the imagination.

@nabiljo try it.

@Rocky618 those are compilations. By "standalone" I mean a clip that has only one gutting, in the 5-15 second range. Try the "Voluntary Sacrifice" clips and the other (shorter) "Cutting Room Floor" clips. I don't have the prompts, so I would have to do the same thing I'm recommending to you.

@c0nv0lut3d the quick stab better reflects the prompt. I try to push violence and brutality descriptively, but the AI tends to mellow everything out. The "Cauc-Asian" thing is funny because it usually does the opposite and tries to transform characters into Asians unless I specify other races.

@Rocky618 these are a perfect setup for anyone who wants to do gutting (that is intentional, in case I ever get around to it). Find one of the standalone gutting clips I've done and copy the prompt (if you don't want to use your own words), then use it to generate new clips off of frames from this one.

@Gibraltar66 it understands slicing much better. It amuses me to speculate that “is it cake” clips formed a significant portion of its training.

@Natalie90 your comment inspired these: https://mlos.pika777.eu.org/A43631C

@Gibraltar66 there’s a reason the initial stab takes an entire clip. If you’re talking about this image/scenario specifically, I don’t intend to revisit it. But, as always, you can cut this clip down to a point where you want things to go differently and generate a new sequence from there.

@MeanMontague try it.

@ModularRodent also, it says "work in progress" :)

@The_Peeper a) AI has a very hard time with upside down people, and b) don't do anything with celebrities. You'll get the uncensored models shut down.

@ModularRodent the prompts for the clips you're talking about are in the metadata of those clips for anyone who's interested. I'm tired of fighting the AI to get that look, but all you have to do is copy the prompt, cut the video you want to see in that style, and generate a new clip from the last frame.

@marcus_speck I’m not that artistic. I don’t know how to put different shots and angles together. My brain doesn’t work that way.

With all my powers of descriptive English, I can't get the AI to turn the sickle point-upward more than a fraction of the time. The same goes for the sharp edge of knives in other clips. No need to point it out. I'm not retarded.

Not that anyone cares, but this actress and film(?) inspired the body type I try to generate and the manner of death I try to portray with AI. I wanted it to be more graphic, but think it's pretty great.

@Krommer lol

@Sunnivah aw, thank you

@polepod send me a DM. I don’t like to publicize it, because it’s already being censored more than it used to be due to negative publicity.

@darkstar25 try it! I'm going to start sounding like a broken record: the words you're using can actually generate the scene you're describing if you type them as a prompt instead of a comment :)

@Jewfish108 I’ve tried dismemberment with some of the hard vore things I’ve done. I couldn’t find any consistent way to prevent a limb from regenerating frame-to-frame. It could probably be done by repetitively describing the missing limb(s) in the prompt, but to me that’s more trouble than it’s worth.

@ZippieZipper whenever it's approved: https://mlos.pika777.eu.org/954C521

@Waistcutted I've mostly stopped showing intestines because I have to burn through too many generations trying to get them not to look like color-coded anatomy textbook illustrations. My suggestion: extract a frame from this and go for it. It might be easier than you think :)

@Eierliebhaber if you find out, I'd like to know. Asking for a friend :)

@ZippieZipper I like that idea

@darkstar25 if/when it's approved: https://mlos.pika777.eu.org/1B3C2D1