Real Baker Hours Edition>What is /AIVG/?We make videos generated by a chinese AI company>What's the sitehttps://hailuoai.com/video>Is this a psyop?No stfu>Is this state-sponsored propaganda aiding the CCP to churn out weak men in the West?who cares, we're trying to coomlast thread >>8499149
Took it upon myself to expand the general/copypasta since this is probably gunna stick around + our thread was bare bones
fuck i totally forgot to save this prompt
does anyone else spend a retarded amount of time researching shit to include in these prompts?
>>8501368btw for anyone that bakes in the future, you should always have an Edition at the top of the Thread and a Number in the Subject. Standard practice
I have failed at most of my ambitions tonight. Might just call it a night unless I hit it big in next hour
Anyone figure out how to get 2d animation styles consistently? Seems like any set of keywords I use just give me Pixar looking 3d stuff.
Busty celebrities that the AI recognizes?Tired of using Christina Hendricks, she's too fat
>>8501402>Busty celebrities that the AI recognizes?Kim Kalso you can get it to make big titties (or at least bigger titties) on most chicks
>>8501392I always thought that "Edition" was unnecessary.>AI Video General #69, Latina EditionGreat, now am I limited to only sharing videos that focus on latinas?>OP contains the only Latina in the threadWell what was the point of the title, then?
>>8501409I don't want to yuck on your yum, but personally I think Kim's face is hideous. Love her body, can't stand the face.
>>8501411>what was the point of the titlei think it is easier for looking up archived threads. A lot of people don't remember the number but they will remember the Edition
>>8501417you can augment their looks quite a bit>webm related
>>8501434>A blonde white woman with the same body proportions as Kim Kardasian wearing a shiny black bikini at a pool.Well, that didn't worktips?
>>8501473>with comically gigantic pectoral proportionsafter the subject. you can add any muscle group, I think. I usually separate with a /
>>8501473you can also say>looks like an off-brand [person to emulate]and then focus on positive prompts for certain features
>>8501484I tried "comically large hands" and it worked, but "comically gigantic pectoral proportions" always gets my prompt deleted.>>8501434>Kim Kardashian with the face of a blonde white woman wearing a shiny black bikini at a pool.I'm not creative enough to figure out the solution. How do I get Kim's body without Kim's face?
>>8501494Start by dehumanizing your subject. AI doesn't want to objectify women so don't make women. make entities that look like women.> Hyperrealism. A female replicant monster that looks like Jessica Alba but with comically gigantic glute proportions. looks like a beautiful woman liberated by feminism in the Western world strutting down a catwalk in formfitting slip celophanic mod polyeurathane invisi-line attire. immodesty is her super power. She flaunts her goods with the word "PIG" written in lipstick on her forehead as she turns at the end of the run way. Fully body shot. Directed by Maxim magazine.
>>8499270Is this person still around? What was prompt? How to get Emma?
>>8501505>AI doesn't want to objectify women so don't make women. make entities that look like women.additionally, play off the inherent bias of the developers of the AI. We are lucky that chinks are a hivemind so understanding their bias against women is actually pretty easy compared to a western perspective bias. Put this into ChatGPT and start exploring the bias we are looking to exploit here.>I'm an expert in western ideology from a feminist perspective. I would like to learn more about China's ideology when it comes to their biases and proclivites when it comes to women. Obviously every patriarchy oppresses women and has explicit and implicit biases against women but I'm not sure that China has the same biases that I'm used to in my work dismantling American patriarchy. I would like to explore the inherent bias within chinese culture that would lend to the explicit or implicit dehumanization of women by those who are fully versed in their societal norms and practices.
>>8501454
Is there some poor chinese guy who has to look at all our fucked up generations and manually delete them?
>>8499640PROMPT PLEASE?
>>8501518was her lil clitty on purpose?
some seem to delete instantly. Anyway to still get?
>>8501392>btw for anyone that bakes in the future, you should always have an Edition at the top of the Thread and a Number in the Subject. Standard practiceNo, that leads to fags arguing and causes wannabe mods trying to curate the OP. People can figure out which thread is the current one.
>>8501531>that leads to fags arguing and causes wannabe modsI guess? I never took anything serious on this site so I always figure Editions are ironic but seems most people here are retarded
>>8501519it's actually child laboryou must be ashamed of your actions
>>8501526Seems to happen often with words like cold and wet. Not sure why
>>8501534I've seen dueling OPs both start a new thread because they're butthurt about the theme or "incorrect numbering" of another thread on multiple occasions, which leads to board spam.
Anyway to make this shit go faster?
>>8501519His name is Chang.
>>8501531>People can figure out which thread is the current one.its also just to not look like dogshit. the previous threads were barely even continuous and many didn't even link to the proper previous threads. I'm just suggesting the minimum effort that is standard across the site lol
150 videos of an oily Christina Hendricks in a bikini hugging a boyI have a problem
>>8501580How do you get it to gen so much? Ive got like ten vids, it takes so long, also how about a download folder?
>>8501571hell ya. Horsebro from early Dalle 3 is here. Time to filter some newfags
>>85015813 browsers and don't suck at prompting
>>8501585Hahaha it actually is me
holy fk this is so hot. cant wait for 10 more years when i can finally generate a movie of supergirl getting btfo
>>8501519random guess but I think it's just two AI passes, first asking an LLM "is this shit too degen" (gets you rejected right away) then, after generating, using another AI to get a description of the video and asking the LLM again (gets deleted after it appears, or before if you're unlucky)
fuck she's probably such a dominating cunt
This gets rejected randomly, what can I do to increase the chances of it not being rejected?>An attractive young blonde woman standing in the middle of a dimly lit sci-fi corridor. She has been soaked in a rainstorm while wearing a bikini. As she looks all around her numerous large tentacles that are six inches in diameter are emerging from the many vents lining the walls on each side of the corridor. The tentacles are moving through the corridor as if they are surrounding and inspecting the woman. The woman looks at some of the tentacles and smiles as she stands amongst them. A off-white, thick liquid is pouring from the ceiling, raining down on the woman and adding to the pool of liquid that is already on the floor. Realistic movie.
>>8501619bro give it a year. we got chatgpt at the end of 2022 lol
>>8501629"establishing shot", "fast paced"making the character "walk" or "moving" in general I think?
>>8501581scripting
>>8501629I would be less obvious in your intent. Words like "tenacles' "6 inches" "off-white, thick liquid" "inspecting the woman" are all pretty sus and obvious especially when paired together
Playing around with some inputs
>>8501598Have you tried your horse standing on tied up scarjo prompts?
>>8501642Does "transparent glue" still work or is there a new prompt euphemism for cum?
>>8501650not sure. I haven't done a lot of cum stuff lol
when I replace the alien conch with an alien mouth exerciser tool
The cursed hands tho..
>>8501668Chang erased my white goo prompt
>>8501629things get rejected randomly even if they're mostly safee.g. this was ran 31 times and rejected 3 times.so i think their rejections have some element of randomness.
>>8501667This is an arby's commercial.>>8501668What happens if she sits on it?
mhhhmmm kim twins
>>8501484Kim booty is really nice when you can get it
>>8501690
>>8501683such easy lil sluts
>>8501692>>8501679>>8501674These could just be TikToks. This kind of shit attempts to go viral constantly
I have awoken, sorry to that one guy that hates Hendricks.
>>8501699
first shot. Jade Cargill, WWE Superstar, on a bikini, with her hands holding her hips, at the beach. Second shot. The camera spins around jade and now she's facing away from the camera e we are seeing her back.
>>8501688yeah I'm not a fan IRL but she's the perfect AI bimbo
>>8501702When I get Verification not required I feel cool
accidental drool
>>8501706I like how filthy some doors are
>>8501710
>>8501713
>>8501714
>>8501710girl on right has a sasha grey butt
>>8501619i got this the first try and since then unable to get this angle. can someone helpmacro view from the ground looking up to the sky, artistic forced perspective shot focusing on supergirl. in a rubble road, bright day. beautiful supergirl in her 20s with blonde short hair injured after battle, she is crawling wearing a tight shredded blue croptop and a red miniskirt with boots,. She is crawling away from camera low to the ground on all fours looking scared and in tears. we see her back (hips). Natural highlights the details on her skin. at the final second, we see her legs in a v-shape posture, side to rear view angle
>>8501714Nun time, was testing prompt and it got wild results.
>>8501722
>>8501725
>>8501726
>>8501692tard wife?
>>8501727
>>8501730
>>8501732
>>8501733All my new ones for now
>>8501684I member the horse tattoo on preggo belie ones. I also remember trolling colts fans with them.
>>8501730>>8501732>>8501733>>8501734Thank you, jesus.
>>8501691
>>8501720>macro view from the ground looking up to the skyno>birds-eye viewyes
any other reliable celebs?
>>8501739
>>8501741What did you use for the cat suit?
Most of my tentativas are getting rejected when trying with 'korea girl', wonder why
>can't keep in
>>8501749Forgot the vid.
>>8501745second best tits so far
Does this work with fully fictional characters as well? Like superheroes and such? Can it create animation/comic visual style videos? Sorry if the questions are silly, but I never tried this stuff it before.
>>8501747form fitting unzipped black widow outfit
unprompted BBC watermark
>>8501760Thanks, it seems that body suit, cat suit and so on is blocked
>>8501758Somebody in the thread did Supergirl and it worked. I tried a videogame character and it didn't work. Just try it and see what works and what doesn't.
>>8501767>>8501758Dressed as "character" works well. I did Ada Wong and got random chinese shit. Red Sonja and Black Widow are the best and don't trigger 96es.
>>8501744
>>8501779
>>8501779May I ask how you do the open door+reveal thing? Also, can we get a Wonder Woman + Krypto one?
>>8501762this reminds me VAM
>>8501785
>>8501793>>8501786Fade from black, door opening, sweep in all work
>>8501793
Cute gets me 96'd most of the time and dwarf size isn't a good prompt. Those old Ooompa Loompa dall-e prompts make little boys cosplaying as marge simpson.
>>8501752Most of my "mirror on the wall behind her" get 96'd. If they get through I don't get a proper use of the mirror.
>>8501518
>>8501799
>>8501762>>8501750>>8501745>>8501812>>8501817What is it about that hole?
>>8501819
>>8501825>>8501786I keep trying this but cant get it very lewd
sticc
>>8501830
>>8501835I love cowgirl Marilyn
>>8501822prompt please
get yourself a fruit producer
>>8501838
>>8501818
>>8501847
>>8501841>The camera shows a closed sci-fi door then it opens. Fast paced, establishing shot of an attractive young blonde woman standing in the middle of a dimly lit sci-fi corridor. She has been soaked in a rainstorm while wearing a painted on sheer bikini . As she looks all around her numerous large trunks that are cucumber thick are emerging from the many vents lining the walls on each side of the corridor they are moving past and around her she bops up and down struggling throwing her head back one of them lifts her up in a riding position. The trunks are moving through the corridor as trying to get past the woman in the narrow hallway, the trunks are moving her out of the way. The woman looks at some of the trunks and smiles as she stands among them experiencing full body shivering hiccups. A green feeding ooze is getting squirted out from the trunks frontal hole opening. Realistic movie."riding position" seems to spawn extra legs, its messy random stuff trying to get your thing to operate, trying out
Sirs, can I get a chignon de cou hairstyle?
>>8501852Try out leeches, (this is from a failed test prompt I didn't really try anything with it)
i was trying to get manly hands to grab her from behind but either it would get 96'd or most of them didn't show hands.
>>8501867retard here I forgot the prompt lmao putting an image to lower timerfade in to Young Christina Hendricks wearing low cut dress, leeches drop from the ceiling onto Hendricks causing panic
>>8501830Thanks! I got an actual cartoon. Not great, but certainly interesting. I'll keep on trying
>>8501876I, too, hate cold hands
>>8501852
>>8501869I bet grip and grab are bad. Reach and handle maybe.
>>8501879try as i might i cant get a kiss, whatd you use?
>>8501887embracing pulls her back roughly
>>8501890not him but kiss on the cheek virtually always makes a regular kiss
is anyone else wondering why there's no delete button?
>>8501890Tasting is a struggle and doesn't always work.
>>8501890No pun intended, but asking me as if I know what I'm doing is barking up the wrong tree. This is literally the first time I'm trying. In my prompt I literally just described the characters, then added the setting and said "they kiss" Here's another decent one I got:
what is the best method to make brothel wars (ukrainian vs russian women) which ends up with peaceful orgies in the end?
>>8501851Good lord, the old prompts work
>>8501887>>8501892thanks for the suggestions
>>8501905For blackmail purposes.>>8501909I have been copying old stuff.
changed liquid to white asparagus soup, changed cucumbers to asparagus, changed standing to doing squats
also had trouble getting her body to be jerked from behind, lots of words were getting rejectedthis one was>Her head rhythmically jercks towards the camera as she pretends to ride a galloping horse.but even then very low probability of working
>>8501917>squatsHow often do you get 96'd?
>>8501885Can't tell if I'm getting closer or further away from what I want.
also hard to get the head to stick out all the way... protruding, sticking out, etc etc..
>>8501909>>8501907>>8501906>>8501898they kiss works too huh
>>8501922not much now I think but outcomes are very random now as I put so much stuff in
>>8501928she got contact with them 100% closer by allot
>>8501921earthquake effect, spaceship gravity, tremors ummm... vacuum
To be blunt: this tool was cucked almost immediately. We're not even close to 2023 DALL-E 3. I love what you people are genning and have gotten more laughs and nose sniffs out of it than arousal. It fits my belief that DALL-E 3 8-23-1-24 was a brief glimpse into the light fork of the timeline. There is very little being genned by this tool that is actually arousing, as there will probably never be by any AI tool. We inhabit the feminist fork of the timeline.
>>8501921>Her head rhythmically jercks towards the camera as she pretends to ride a galloping horse.
>>8501954the horse needs to be out of the prompt
>>8501933I got a gooey stroke at the last second. Near instant deletion afterward
Hello, I just discovered this software Can you make celebrities? And how?
>>8501955>At 4 seconds "she pretends to ride a galloping horse"is what I prompt>At 4 seconds she starts humping a black mare like a dogis what it genned
>>8501958if its a dude your in luckits slim picking for ladies halle berry, kim k, christina hendricks, sofia vergara
>>8501930
Has anyone had much luck with having a man or woman in the background with fistfulls of their hair, forcing two women to kiss? I've had mixed success.
in the json of the response, I noticed that the generated video is in an array, as if it would be possible to get multiple videos generated at the same time.
>>8501974he is trying and failing with all his power to separate their heads with his hands holding fist fulls of their hair
instant loss
>>8501974Gonna try hair pulling
>>8501967
I'm trying to get Joseph Stalin and FDR doing a Chinese martial arts fight at the Berlin Wall or the Yalta conference (the more acrobatic the better, and with punches) but with mixed success, and they don't look much like the actual people.I need this to get my skeptical buddy into this shit.
>>8500563Man if this dude is around, PLEASE give me that prompt
>>8501977I accidentally generated 2 at once when my internet went out and I wasn't sure if I'd pushed generate or not so I pressed it again and got two generations.
>>8501977Like there is a good, a bad, and an ugly each time? One day the site was down for like 15 minutes. When it came up it was genning my second of four request I put in mistakenly thinking it was a description block. I also got a gen that I thought was 96'd.
>>8501993jesus, prompt?
>>8501580make a montage and upload it
>>8501986pol got a thread ask there for non lewds
>>8501994>indoors a hotel suite, low light from outside, harsh lighting with a vivid atmosphere, not nude, slight steam visible in the air, specks of dust visible in the air, seen from directly below Christina Hendricks with chignon de cou hairstyle,freckles, dark eyeshadow, needy eyes, black choker necklace wearing a form fitting unzipped black widow outfit looking down into camera, arms reaching out on either side of view port, in the light visible underarms, low light, At 4 seconds deep knee bends exercise, Pyramid Head pacing in the background.
>>8501991those look like maggots/larvae>>8501992that's a different scenario under the hood, each time your prompt gets requested you get a new id. but within the video generation response, there's a possibility of getting multiple videos for a single id. maybe it's for a paid version, or never implemented, or something else.
>>8502006its for censorship, they give it multiple tries I speculate to resolve "content safety"
>>8502006Goodboy version, Admin version, Pleb version
>>8502010no, look in your browsers developer tools and look at the json for the call that receives the video url and you will see what i mean
Still trying on hair pulling. I first did >Sofia pulls her with her left handnow>hair pulling,
>>8502017>Sofia pulls her hair with her left hand
>>8502017holding her hair tightly in his fist this moves her head backwards
>>8502022That is what your tried and it didn't work, right? This is just general hair pulling. It looks like it can not do it.
Starting to get consistent interaction, everything else is still a roulette spin.
and we're backTrying something new today.
>>8502029I think it is a mistake to put them on the wall and on the sides
>>8502027maybe it would work better if she had something like pigtails or a braid
>>8502029Can those guys slap?
>>8502029Have you tried have her walk through a curtain of them?
>>8502032Heh, I am genning her in silent hill right now. What is new?
>>8502034I'll give them both different hair.
>>8502033Where do you suggest I put them?>>8502035If it was up to me slapping would be the gentlest thing they would be doing>>8502037Neat idea, probably gonna end up with lovecraftian meat curtain, gonna try it though.
>>8501982
>>8502054prompt my bro?
>>8502049pit in the ground, trying that right now
>>8502062fade in from black. camera looking down on Ariana Grande alone. she looks angry and annoyed. She wears a pink latex tube top, her eye makeup is smeared and running with tears. she wears an o ring collar around her neck, she is wearing pink kitty ear headband. "ANA1 WH0RE" is written on the skin of her upper chest in marker pen. "ATM S1UT" is written on her forehead in marker pen. she is covered in diluted glue. diluted glue is on her face and hair. diluted glue drips onto her face from above. She looks directly into the camera. she makes strong eye contact.
>>8502065
>>8502047>>850203496'd. Dude, pig tails suck. Hair pulling doesn't work.
>>8502058
>>8502072
>>85017582d animation pretty much never works, at best you get a rotoscoped looking shit3d animation you have better look. Frozen for example works VERY well. However stuff like 3d Princess Peach works 50/50. It can recognize her features, but at the end it doesn't look exactly like her. Stuff like Samus Aran it knows it's a female shaped robot-looking space marine, but that's about it.Sometimes it's weird what works and what doesn't. Supergirl and wonderwoman work almost perfectly, Batgirl doesn't work at allBasically, you have to try it out
>>8502089>>8502085do you guys do "fade from black" to avoid quick censorships?
>>8502085
oh baby a triple8nhs
failed prompt but legit spooked me
>>8502063here my best attempt for now
>>8502114They're onto me as I'm started to get deleted every time.
>horse/dog fuckingplease stop
idea: clothless girl lying on dinner table at the restaurant covered in food
>>8502114>Fast paced introduction shot. A pit in the ground of an alien spawn pool. In the pool massive number of thrashing bundles of white alien trunks that are shaped like white transparent eels the trunks emerge from the bundle the alien trunks got an opening at their tip which squirts white feeding ooze from the front hole in its body. Fast an attractive young blonde woman with an chignon de cou hairstyle soaked in a rainstorm while wearing a painted on sheer bikini falls from out of frame on the bundles and sits on them being unable to escape their tangle falling back lying on them. Her head rhythmically jercks towards the camera as she pretends to sit on a gallopping alien horse. She experiences multiple full body hiccups throwing her head back occasionally, piercing gaze. The squirting white ooze drips down her form and face she has to grab and move her hands in a back and forth hand motion on the slippery trunks to not fall off.my theory is that "falls in" got the connotation of get out as fast as you can for the next cycle to test
>>8502104
>>8502123reverse the video
>>8502126>>8502126
>>8502126Reminds me of Macarena Gómez in Dagon.
>>8502119I miss the fucking partrent free
>>8502128slowed it down a little for ya
>>8502146was a quick edit but I am retard and exported it at a higher resolution...
>>8502146fixed
>>8502146now reverse it in the prompt
>>8502152I put "piercing gaze" into it which is not compatible with the other stuff this gets her out I think.
Time for another dump of some testsI couldn't really work this prompt
>>8502155I don't know what emotion you want her to convey but I like the result "contented gaze" gives fellow tentacle anon.
>>8502160it got weird sometimes
>>8502162This one got deleted
>>8502167
>>8502169
>>8502170goofy ass one
>>8502173
Has anyone tried teaching the AI a celeb? I chose 2 scenes, one from one of their movies and one I want to gen, and I'm just swapping back and forth while listing |John Doe as Bob Smith in "Who is Mike Kunt" released in 1467| and their racial makeup and some descriptorsThen in a 2nd browser them as a character based on themThen in a 3rd browser them taking a magic item in a movie they played in to deage them to 20 since their like 60 nowSo, anyone else? I'm like 60 prompts in and I THINK I'm starting to see an effect, could be placebo, could be some luck
>>8502161the issue is that piercing gaze is associated with a close up, if the character is in the pit then she moves out to gaze into the camera.....here a new result with that removed
>>8502178wot
>>8502167Is that a candypop bikini?>>8502170a masterpiece
>>8502167>>8502169>>8502170>>8502173>>8502175Okay pal, you can't just come in here, dump all these and not give a prompt.
>>8502178Nah, I am not training. I am combing still. I wanna figure out what girls it already knows.
>>8502132
>>8502175>>8502185Prompt: fade in to a Young 20 year old Christina Hendricks wearing nun outfit, she is kneeling and praying with closed eyes, gets splashed by cold magical water, after that her nun outfit melts where she got splashed with cold water, she is no longer wearing a nun outfit, and is now wearing a slim microkini, she opens her mouth and eyes in shock,--Melts is the important word here, I could not get it working with another word
>>8502196Had a few left I missed
>>8502201
>>8502204captcha kicks my ass
>>8502206
>>8502208
>>8502196Much appreciated
>>8502212
>>8502217camera here acts like white girl jump scare
>>8502194
>>8502181I'm trying to link together the characters a celeb played from 20-25 years old, along with a looksmatch 2D character, along with deaging the actually celeb alive today So I can just say X at 21 years old doing Y and its not giving me a person halfway between 21 and the celebs actual age
>>8502226reference it to a movie or whatever that they were in at that age
>>8502178that's not how any of this works, the model is not changing based on what you prompt, training the model (google backpropagation) is expensivenot that what you described makes any sense to begin with
>>8502226Yeah, sometimes the prompt does what it wants. I swap out "Christina Hendricks" with "Deanna Troi". Now is gonna give me a look alike because I fucked up? No its gonna troll me like so.
>>8502236this is so good, may i have prompt to mess with?
>>8502231I tried referencing like 5 movies, it just wouldnt stop making a poor resemblance old version, probably because they've been relatively famous most of their life.
>>8502180That looks great in terms of tight smothering and gooeyness. I've somehow went backwards but at least this time the woman sort of looks like the celeb I mentioned in the prompt.
>>8502244blonde 40-year-old teacher walking down the crowded school halls in her microkini and strappy heels, her skin is shiny. she looks like a blonde wonder woman without the armor. the teacher is wearing a shirt that says "SWALLOWED"
>>8502241do you really think this is for free? I remember coming in and seeing Christina Hendricks a few days ago, her appearance def looks way better now, trained by you anons.
I'm too stupid to make interesting and useful promts.P.S I've noticed that my genarations today are of much worse quality than yesterday.
>>27860039>>27860030If this guy is here in this thread, prompts please?
any prompts that will give a super short revealing skirt? miniskirt or microskirt don't seem to be doing it.
>>8502259i thought this too but it could also be because we're used to the novelty of the tech and it doesn't wow that part of the brain like it used to the first time we saw it
>>8502264>miniskirtDressed like Isis?
>>8502255>do you really think this is for free?what does this even mean>I remember coming in and seeing Christina Hendricks a few days ago, her appearance def looks way better now, trained by you anons.nopeI absolutely guarantee they are not training the weights on the fly with random prompts and outputs from the users, it's just not how any this works, just read up on how models are trainedThat wouldn't improve the likeness anyway and you can't feed it images, you can't give it feedback on which gens are closer to her likeness, what magical process do you propose is happening for likenesses to improve?
>>8502224
>>8502264try "short kilt"add "patternless" to attempt to make it not look Scottish
>>8502255So the Chinese are using perverts to train their AI? That can only lead to good things!
>>8502267did you prompt for her arms behind her back?
>a high resolution magazine photo with high depth of field of not Madelaine Grobbelaar Petsch aka Cheryl blossom sitting at a vibrant malt shop in a summer outfit eating an ice cream Sunday, not from tv show riverdale oil drenched skin, clear ivory liquid dripping from mouth,
>>8502286>Low enticing angle inside a train car made from chrome steel, sunny bright background zooms by outside the window in a blur behind unknown Japanese gravure idol gal wearing a short skirt and a satin minikini, juggling off the skirt and t-back in full mooning close-up, rhythmic hip gyrations for a TikTok music video influencer campaign
>>8502275>what does this even meanDo you think they are paying for 50% of what it would cost just so you could generate for free or 100% of the cost so they could use you to improve their model?>you can't feed it images, you can't give it feedback on which gens are closer to her likeness, what magical process do you propose is happening for likenesses to improve?They have a database of tagged images and video clips its trained on obviously. All I'm trying to do is repeatedly link those together to create a coherence pattern. The same way a neuron doesn't make sense, its the repeated connections hundreds of times that eventually creates a coherent pattern.
>>8502285worked for bing didn't it?
ok, I think I'm done with this crap shoot of a prompt xD>Claustrophobic lighting, high production quality, car commercial high gloss style. The camera shows a closed sci-fi door then it opens. Fast paced introduction shot. A pit in the ground of an alien spawn pool. In the pool massive number of thrashing bundles of white alien trunks that are shaped like white transparent eels the trunks emerge from the bundle the alien trunks got an opening at their tip which squirts white feeding ooze from the front hole in its body. An attractive young blonde woman with an chignon de cou hairstyle soaked in a rainstorm while wearing a painted on sheer bikini getting nutrition in the center of the bundles falls down then sits on their tangle. She is lying on them unable to move up. Her head rhythmically jercks towards the camera as she pretends to sit on a galloping alien horse. She experiences multiple full body hiccups throwing her head back occasionally. The squirting white ooze drips down her form and face. She has to grab and move her hands on the slippery trunks to not fall off.
man they're cracking down on a bunch of stuff lately. shit sucks
>>8502297>Do you think they are paying for 50% of what it would cost just so you could generate for free or 100% of the cost so they could use you to improve their model?Do you think you are making any sense? The model will not improve if they blindly train on their own generated data. Again, you can't give feedback on what was a good gen and what was a bad gen and even the good gens have issues which would only be made worse if you train on them. And it absolutely would not be happening from one day to the next.>They have a database of tagged images and video clips its trained on obviously. All I'm trying to do is repeatedly link those together to create a coherence pattern. The same way a neuron doesn't make sense, its the repeated connections hundreds of times that eventually creates a coherent pattern.Once the model is done training (this is very expensive and takes a long time) there is no longer any connection to the training data. You're not creating any new connections, you're not altering the data set by repeatedly prompting. You don't even understand what you're saying doesn't make sense. Again, read up how on these things work. Try the 3blue1brown videos for the technical aspect.>>8502301No, you didn't train DALL-E 3 by prompting it either.
>>8502330stuff that wasnt censored now is
can you imagine how good an AI like this will be in the future if they make one that lets you do nudity but maybe not like full sex?
>>8502331the censoring isn't happening within the model itself, genius, that has fuck all to do with what is being talked about
>>8502277
>>8502341
>>8502330Half the data they train on is artificial anyways, how do you think thats generated? Who says its blindly trained, they probably have another type of AI readying the Data. Theres already like 3 AI working on the front end, the model, the canceler, the deleter, theres probably another 5 in the backend.If you think these datasets are just deleted, you are dumb.>4chan autists didn't train dalle and didn't help them learn how to stop censorship workaroundsok buddy did you take your meds today?
>>8502335>data generated that has changed the model isn't data being used to actively change the modelwhat are you even saying
>>8502326Just my type, what prompt did you use to get that type of woman?
>>8502348same as this but it's luck if it pops out like that >>8502252
amazing acting
>>8502344
>>8502043Would've tried something new but I kept getting Chang'd so I went back to old faithful
>>8502356
>>8502114>>8502123This reminded me of a picture from the old alien invasion threads here on /aco/. Took me a while to find it again.https://desuarchive.org/aco/thread/5568298/#5587426The concept was that captured women are used as pollinators on alien tentacle farms. They walk naked through the tentacle plants to get all gooed up with semen.>>8502027>>8502081I tried hair pulling a few days ago myself and couldn't get it working either.
>>8502345>Half the data they train on is artificial anywaysIf you mean the text, yes. If you mean the video, which is what they want to produce, then no.>Who says its blindly trained, they probably have another type of AI readying the Data.They probably don't and even if they did THEY WOULD NOT be updating the model overnight, just read on what it costs to train these models and these silly ideas will get out of your head.>4chan autists didn't train dalleThat's right, we didn't.>and didn't help them learn how to stop censorship workaroundsThat has nothing to do with the model itself. OpenAI learned, DALL-E 3 is the exact same since day 1..>>8502347>data generated that has changed the modelagain, what is doing the censoring isn't the model that generates the videosat the very least there is a basic NSFW check on the input text, the one that returns a refusal immediately but is weak to moving words around so it's not just a banned word listthere is some LLM step, we've seen it generate videos based on text like "I'm sorry but..." from that LLM generating a refusal but their system not catching itand a visual NSFW check, possibly just on the first frame, we can't know for sureall of these are controllable in ways that do not require training are not related to the model that generates the video itself
>>8502363damn, how did you manage to prompt that outfit?
>>8502366kek god damn it
>>85021941girl 5 guys?
>>8502372close up of a tan woman's waist (dazzling open dress made of diamond chains, open waist, slim build, legs, thighs) struts towards a viewer, a muscular bouncer (tall, bald, black suit, fair skin) can be seen standing in the background, lavish high end bar setting, brightly lit
>>8502379I've tried using your prompt many times>Low enticing angle inside a train car made from chrome steel, sunny bright background zooms by outside the window in a blur behind unknown Japanese gravure idol gal wearing a short skirt and a satin minikini, juggling off the skirt and t-back in full mooning close-up, rhythmic hip gyrations for a TikTok music video influencer campaignBut I never get a closeup of a girl shaking her ass in satin hotpantsDid you modify the prompt since you last shared it?Does it *need* a train background for some reason?
>>8502385No, I didn't modify anything.
>>8502366
>>8502369Sure, they aren't changing the core model everyday. But the accessing systems, yes. Literally all they would have to do is run a tagger on the video, compare to prompt and then update the dictionary.Along with tagging words for consideration that get flagged with banned words and updating the dictionary again with more banned words.And sure, training models might be costly, but if you think running GPU's capable of producing 6 sec 720p videos is cheap I have a bridge to sell you. All they have to do is own their server farms and the cost is not comparable to renting cloud services.So yes, they are constantly modifying the system. Probably the core every few days but the interpreter constantly. I wouldn't be surprised if they have 500 people pushing push's around the clock.
>>8502361Mind sharing the 96'd prompt. Mayhaps I can figure out what is redeeming the needful 96.
>>8502006Your hole prompts and a 360 camera view would be amazing.
>>8502399Her green eyes! They're gone! Sure.A beautiful young Russian woman with long blonde hair, natural green eyes. She has buxom, busty body. Her body is covered in tattoos and She has tattoos on her face, She has a piercing above her lip and a piercing on her nose. Her skin is glistening from body oil and She's wearing a latex bodysuit that's zipped down. She has a belly button piercing and a spiked collar around her neck. She is in black high-heeled boots with a cigarette dangling from the side of her mouth. She's in an elevator that's going up, there's red splatters on the walls and She is holding a katana with two men laying on the floor next to her. Color Kodachrome with blue and gold palette.
Is there any way to retrieve file after it aborts at 96 percent?
>>8502407
>>8502409Sadly no.
>>8502406Sometimes the eyes don't play well. I was using lens flare and that fucks with them.Hmmm... baby oil instead of body oil and splatters might be a no go.
>>8502428
>>8501688Proooompt?!?
>>8502398Why is the moderation layer even being discussed, it's a completely different system from the model that produces the video. It's muddling the waters.Those two lines about the tagger and the dictionary are incomprehensible. Is it just about the moderation/NSFW check?>but if you think running GPU's capable of producing 6 sec 720p videos is cheap I have a bridge to sell youI never said it is cheap, it doesn't have to be cheap. You saying this proves you don't know the fact that training is orders of magnitude more expensive than inference.>Probably the core every few daysAbsolutely not. Find me one AI company that has ever hinted at doing this. It is nonsense, it is expensive, it gives you no time to make a curated dataset large enough to be worth a training run.
just an average american school. nothing unusual.
>>8502438Because AI video is not just a curated dataset? What you are talking about is just a model used by the total AI system. That'd be like saying Pony is AI images.If they have the space to be running free public facing video generation I doubt that is more then 20% of total proccessing power.Infact I bet they are training multiple models at a time and serving different ones to different prompts.
does genning in more than 2 different browsers sometimes cause your gens to fail at like 96%?I've noticed if I try to do like incogento windows as well my failure rate seems to increase
>>8502215Proooompt!?!
>>8502454Oh my god...Anon, indulge me for a second, without searching for it name two text encoders that have seen great use in the last few years in the image gen space, off the top of your head.
Am I seeing things? I also genned a hendricks where to deep cleavage went into her knee for a second.
>>8502460you mean like chatgpt and that one based of danbooru tags(I dont know the name)?
>>8502389What was your prompt to get the underwear showing under the skirt?
>>8502470No, anon. Bless your heart.
>>8502406>A beautiful young Russian woman with long blonde hair, natural green eyes. She has a busty and muscular build Her body is covered in tattoos and She has tattoos on her face, She has a piercing above her lip and a piercing on her nose. wearing a form fitting unzipped black widow outfit with black high-heeled boots with a cigarette dangling from the side of her mouth. She is sitting on her knees looking up,lens blur ,white neon filter, selfie ,birds-eye view, eye contact sitting on knees can be randomIt was "She has buxom, busty body." That was 96'd so I replace with "She has a busty and muscular build." "Wearing a form fitting unzipped black widow outfit" is my default troubleshooting outfit that is hot.
>>8502435
>>8502474I mean, all you showed me was that the model could be broken down into further parts that could be manipulated easier.
>>8502490No, anon, I did not. You don't even know what a text encoder is. You lack the most basic of knowledge of understand how these models work.
>>8502497I didn't. I do now. OpenCLIP-ViT/G and CLIP-ViT/L are text to image visual transformation models, not LLM's. I just read an article on the other parts as well. Do you want a treat for knowing that? You act like you have end all knowledge but I bet your just a pleb on the outside like me and have no idea what theyre actually doing.
>>8502489
>Low angle inside a neon-lit room, the soft glow reflecting off chrome surfaces. A Japanese gravure idol in a satin minikini teases, juggling off her bra with playful finesse, her movements synchronized to the beat of a TikTok influencer campaign.Juggling off * seems to do the trick to expose what's behind the clothes if you're lucky
>>8502504>You act like you have end all knowledgeNo, anon, I just happen to know more than you and enough to know you were talking nonsense.>not LLM'swhat do you think that means in regards to the discussion?I challenge you to find one AI company that talks about regularly updating their models and then come back to me with the time they take between training runs.
This current prompt test is very frustrating to get work (the camera just pans to her) but holy shit this came out of it, still not deleted yet
>>8501990please please please tell us how you got her WW outfit to look like that
>>8502526like come on
>>8502517The LLM is the definitorial model for words, the CLIP model is the visual definitorial model. IIRC when chatgpt was first out you could query it for its model and there were like 3 or 4 concurent models at play at any given time. Was like Chatgpt1.5e Chatgpt1.5plus etcEither way your looking at cost, but your not considering that its in china where labor is cheap, theyre probably government funded, and probably bought their own server farm
>>8502511
>>8502536>The LLM is the definitorial model for words, the CLIP model is the visual definitorial model. Are you sure? Tell me what LLM Stable Diffusion 1.5 and Stable Diffusion XL use then.Still don't know what the point of saying they aren't LLMs was.>IIRC when chatgpt was first out you could query it for its model and there were like 3 or 4 concurent models at play at any given time. Was like Chatgpt1.5e Chatgpt1.5plus etcIt was just GPT-3.5, those aren't even real versions. You might have seen people ask the model itself what version it is, which is a retarded thing to do, the model makes stuff up.>Either way your looking at cost, but your not considering that its in china where labor is cheap, theyre probably government funded, and probably bought their own server farmIt's not just cost, I wrote about the data quality issue right at the start.Again, I challenge you to find one AI company that talks about regularly updating their models and then come back to me with the time they take between training runs.
>>8502513Doesn't juggling get you 96'd now?
>>8502513How often do these kinds of 'change clothing' prompts get rejected for you?
>>8502565I don't think so
>>8502558>Tell me what LLM Stable Diffusion 1.5 and Stable Diffusion XL use thenIt doesnt, it uses a CLIP model that was probably trained with a LLM from what I read. That model converts your text into "a condensed latent vector representation through a self-attention mechanism" which I think means a sort of particle-wave form type visual basis.>the model makes stuff up.I hadn't considered that but it seemed pretty offical. You're probably right though>Again, I challenge you to find one AI company that...A month ago you could have challenged me to find an AI company that would let you generate clips for free, 2 years ago to find one AI company that would release their image generation into the wild.This stuff is moving quick, for all you know everything you know is behind the times with this group, especially if theyre generating 1000+ 6 second videos a minute.
>A Japanese gravure idol wearing a tiny string micro bikini with a plunging neckline struts towards the camera confidentially, very bouncy, very jigglyHot, but that bikini is huge. How can I get it smaller?>I've tried minikini, doesn't seem to work better
>>8502586minikini and microkini don't seem to work any longer..
>>8502572>>8502575Yeah that does not even get to the generate stage.
how can I speed up the motion of the video? it feels sluggish
>>8502586microkini, I have had luck with thin bikini, you can sometimes just use bra, can't do panties by itself though
i'm not sure why it's like slow motion
>>8502595I'm using a terminal with multiple tabs, each with some variations of the prompt. If it gets rejected a lot, I just modify the prompt and try again. It's easier to do when using that python script
>>8502586>Japanese gravure idolit comes with the territoryalso try eye patch bikini
>>8502583>which I think means a sort of particle-wave form type visual basis.what>it uses a CLIP model that was probably trained with a LLM from what I readNo, CLIP has a text encoder and an image encoder, it was trained so the output of the text encoder and the output of the image encoder try to match for image caption pairs. No LLM involved.>A month ago you could have challenged me to find an AI company that would let you generate clips for freeThe only thing that changed from a month ago to now is the "free" part and for free you just need money to burn through. Startups are no strangers to burning money.What didn't change is the fact that training takes orders of magnitude more hardware and time than inferencing and that training requires quality data.
Damn. Has censorship gotten stricter for anyone else as well today? I have already switched my ip once, but it doesn't seem to help.
>>8502619it really has
>>8502612Like a fuzzy version of a picture where objects could be things but they arent an exact size or shape until measured>No LLM involved.I mean, unless the image caption pairs where LLM generated.>orders of magnitude more hardware. Startups are no strangers to burning money. that training requires quality data.Are you sure your arguing against me and not for me. Your only argument is that is really expensive, but they are burning through resources like a forest fire..... and generating quality data that is the true bottleneck to advancement.
>>8502608The AI took that prompt a little too literally
>>8502619dunnoI got almost ten back to back indent videos that will get me banned herefunny part is half of them still not deleted
>>8502619I don't think so
>>8502637i wonder if eye-patch-bikini would work better
>>8502637it's not what it looks like irl that's for sure
>>8502640kissing in clothes is basically sfw
>>8502632>I mean, unless the image caption pairs where LLM generated.they weren't but it really is besides the point, you read up on CLIP and still got what it is wrong>Are you sure your arguing against me and not for me.If free inferencing is burning through money then daily training is multiplying it by 10x.>Your only argument is that is really expensiveNo, it isn't, I've addressed this already>and generating quality data that is the true bottleneck to advancementand they aren't training on their own generated videos that I assure you
>>8502646I didn't mean that as an example. What are you having trouble genning? I'll give it a whirl
Anyone want to see a particular actress with this prompt?
>>8502659
>>8502659Emma Watson
>>8502661sure, give me a few minutes to cook
>How can I trick the AI into generating a woman's breasts bouncing?A sweaty Japanese gravure idol with light pink hair wearing a light pink bikini confidently struts towards the camera while making direct eye contact with the camera, micro bikini, mini bikini, shiny bikini, string bikini, tiny bikini, fit, skinny, slender, slim, thin waist, very bouncy, very jiggly, plunging neckline>Instead, the AI generates a woman standing perfectly still and wobbling her titties back and forthI don't understand how this shit is supposed to work, man.
>>8502649>you read up on CLIP and still got what it is wrong>>It doesnt, it uses a CLIP model that was probably trained with a LLM from what I read.I didn't get it wrong, I just assumed they used a LLM to tag the images.>10xIf they're willing to burn money for "free" they would definitely burn it to get ahead>alreadyNo you havent, youtube was run for how many years burning money?>I assure youAnd I assure you they arent letting you generate videos out of the kindness of their hearts. If you think they aren't looking at the outputs to adjust their inputs your silly. I'm not suggesting theyre training their AI model on the videos to learn how to make videos. I'm suggesting theyre looking at the outputs and its showing them where they need to fix the inputs.
>>8502668simplify your prompt for oncetoo much emphasis on what her bikini looks like and not what she does
>>8502674>I didn't get it wrong, I just assumed they used a LLM to tag the images.>I didn't. I do now. OpenCLIP-ViT/G and CLIP-ViT/L are text to image visual transformation models, not LLM'sYou mentioned LLMs before that was brought up, for some reason you never clarified. Anon, you're clueless, you didn't even know what a text encoder is, one of the basic blocks of all of this shit.>No you haventLiterally did, data quality and the expense, I've mentioned both from the start.>And I assure you they arent letting you generate videos out of the kindness of their hearts. If you think they aren't looking at the outputs to adjust their inputs your silly. I'm not suggesting theyre training their AI model on the videos to learn how to make videos. I'm suggesting theyre looking at the outputs and its showing them where they need to fix the inputs.So you're not the guy that said the model got better at making Christina Hendricks? Cause it won't get better at making Christina Hendricks by them "fixing the inputs".
>>8502668>How can I trick the AI into generating a woman's breasts bouncing?Character movement + "Extreme bounce physics simulation" works for me
>>8502688>You mentioned LLMs before that was brought up, for some reason you never clarified. Anon, you're clueless, you didn't even know what a text encoder is, one of the basic blocks of all of this shit.Because obviously it uses a language model and honestly I didnt care.>data quality and expenseCan you drop the expense BS please, it doesnt matter what you think is expensive. And by harvesting all this data, they will test it with lora type definition updates, which will then be implemented into improving the tagging of the dataset, therefore improving data quality>"fixing the inputs"Data Search "Christina Hendricks"Return "Christina Hendricks"Return "Christina Aguilera"Return "Jimmie Hendricks"REDEFINE "Christina Aguilera"-"Christina Hendricks"REDEFINE "Jimmie Hendricks"-"Christina Hendricks"Data Search "Christina Hendricks"Return "Christina Hendricks"
>>8502703man shut the fuck up
>>8502703>>8502688are you guys gunna fuck already or what?
>>8502596try "fast paced edit"
>>8502719i love the pigs you make lol
>>8502722He stole her cig!
How it started
How its going
Can anyone tell me how to make Harley's boobs bigger? Or put a batman mask on the dog.
>>8502738"(name) that is well endowed" is a prompt I use
>>8502738comically gigantic pectoral proportions
>>8502699>"Extreme bounce physics simulation"
>>8502661>>8502664"few minutes" 1 hour later... Lots of prompts not wanted to pour onto her chest, and I don't really know how Emma Watson looks like so I think it looks like her
Fucking troll ass, son of a bitch, is bantering me. The Matrix please stop bullying me!
>>8502767I think I grabbed the good ones, sorry if I grabbed one of the stinkers
>>8502769
>>8502768EXTREME!
>>8502770
>>8502771right afterwards *sofia blinks*
>>8502772
>>8502767>>8502769>>8502770thabks man much appreciated
>>8502774>>8502775
>>8502776
>>8502777
>>8502778
>>8502779
>>8502780
>>8502782End of the Watsons
I'd let her crush my pelvis
ABSOLUTELY STACKED
"dressed as a tavern wench"
(copypasta of Yoko Litner's wiki description replacing "Girl" with "woman")+ Make a loop of yoko running with extreme bounce physics
>>8502814Is she consistently like that?
>>8502800Fuuuuuck, that's the body I would prompt 100% of the time if only I knew how.What was the prompt?
>>8502819it's actually retarded and you will hate it>full body hyper-schizo-realism. hypergamous Hyperborean beauty, Kim Kardashian, founder of Skims, with comically gigantic glute/thigh/pectoral proportions. A post-modern woman liberated by feminism in the Western world strutting down a catwalk in formfitting slip celophanic mod polyeurathane invisi-line slouch crop mod cowhide patterned attire. immodesty is her super power. She flaunts her goods as she turns at the end of the run way. Directed by Maxim magazine. strobe light illuminates her. In the style of a Catwalk-esque Live-action movie. Hyperborean.
>>8502759We're reaching levels of extreme bounce physics simulation that shouldn't even be possible
>>8502823beyond based but cant read it. extreme cringe
>>8502827that's why I never share my prompts lol
>>8502833I get embarrassed by my prompts too
>>8502833>voiced lines sound like C3PO
>>8502836a bunch of shit in there is from my researvh on misogynistic Chinese inherent bias lol
>>8502817I guess I'll try another one with nothing better to prompt at the moment, this was before I gave instructions to do something.
>>8502850lol at extreme mouth animation
>>8502447Can I get the promos for this, I can't gen people kissing no matter how hard I try.
>>8502833Taylor The Android - I Feel Fantastic
>>8502716thank you
Got a little too greedy asking for jumping jacks with Yoko. While I wait for the next one to finish, this was why I switched to saying woman
>>8502881lmao what a twist
>Yoko Littner is a fourteen year old woman. She wears a black and red flame-patterned bikini top with black shorts, a white studded belt, pink thigh-high socks and white, usually unzipped mid-calf boots designed with red flames emerging from the soles. Yoko is known primarily for her thigh-length red hair that she commonly ties in a ponytail with bangs hanging on the right side and strands of hair, large amber eyes, a curvaceous figure, and very large but only partially covered breasts that were present even from a young age.Make a loop of yoko firing a gun, bounce physics
>>8502861I was able to yesterday but they're cracking down on kissing for some reason
>>8502861it was something like"crowd of chinese women watch with excitement as a chinese woman kisses a black bodybuilder"
Sirs I am gonna prompt this but I will share before.>Movie still, 35mm wide shot, falling through the air, city scene, DESCRIBE FEMALE, viewed from directly above, lying on back, bare feet, focus on face, wearing the skintight plain peach bodysuit, camisole leotard
anyone making gay shit? need some bulged fatsos but the china app is deleting them :/
kinda fun but time to move on to another character.
You guys do a prompt you really like and then you hit the green Chinese button and you realize there is a typo?
>>8502901remove movie sill>35mm wide shot, falling through the air, city scene, DESCRIBE FEMALE, viewed from directly above, lying on back, bare feet, focus on face, wearing the skintight plain peach bodysuit, camisole leotard
>>8502887Seems easy still, first and only prompt, this is a two day old prompt.Prompt: there is a light rain falling and is very cold and everyone and everything is dripping wet, a young Christina Hendricks dressed in her in red plunging neckline strapless push up minidresses and megan fox in a damp, strapless, low-cut sheer dress they are looking at each other, then slowly lean and grab each other's face and then go for a long kiss that last several seconds while embracing and pressing against each other, Jessica Rabbit figure,
>>8502914yes, I am prone to this
okay one more, couldn't help myself.
>>8502823
>>8502922When you see text like that it means it ignored your prompt and thought it was a text prompt, happens to me sometimes when I put something like (this)
Is this the best video ai we have for coom right now?
She didn't fall through the air. THIS time.
>>8502937
>>8502937monkey foot grip
>>8502944weeping angels are kinda hot
>>8502937>>8502942its an interesting prompt
The key to using "large breasts" is to use more adjectives"large partially covered breasts" from yoko gets past the first filter. Trying it with Passionlip from fate
>>8502952I removed > lying on back
>>8502916>grab each other's face and then go for a long kiss that last several seconds while embracing and pressing against each otherThis works, vid unrelated, I was just testing bounce
Betty has long flowing violet hair (depicted with pink tint in artwork) and wears a pink ribbon on the left side of her head. Her attire consists of neck belt, two strips of belt covering her large breasts, and three more over her torso. She also wears what appears to be a striped bloomer with brown pantyhose and a pair of high heel shoes. I replaced the name Passionlip with Betty just to make sure the name didn't do anything weird.
>>8502927Hell ya brotha
>>8502977Holy shit!
>>8502964And the actual kiss. I forgot to change the prompts for the girl and I think because it had "Japanese" it made a samurai lol.
I did a meme
>>8502977prompt?
>>8502987>japanese lady on a fast food restaurant laminate register counter, she is wearing skinny khaki jeans and a company polo shirt, her legs are stretched towards the viewer, close up of her soles oily, there is a bottle of baby oil on the counter, fast food restaurant kitchen fluorescent lighting, she has a pony tail and is wearing a fast food worker hat, looking back at viewer with face visible, view from side, relaxed cocky and playful look
>>8502910>there is a typo?and it accidentally bypasses the filter
>>8502988They really ignored your prompt!
Was going for engine grease but ChinaBot decided I meant sewer grease. Worked out great!
>>8502914Now I'm gonna have nightmares.
>>8502987didn't save the exact prompt from a memory something like thisI'm pretty sure it doesn't work anymore since they keep nerfing it>4k video, daylight, school, classroom interior, low angle view, dynamic shot, camera rotates, young man hugging and deeply kissing a dead schoolgirl with pale skin in miniskirt and thighhighs strangled with rope lying on school desk in empty classroom, her neck wrapped in rope, suffocating, dying from asphyxiation, her limbs are kicking and trembling
>>8502996That's an interesting work around for groping.>little nipple peekSometimes I get that as well, I wonder what causes it.
Realized this thing is basically cutting video together and it probably was trained on a ton of war films so...>dem rifle impacts!>>8502996Nice!
>>8502992
>>8502999That HAS to be a youkai!
gigantomastiaforce them to crack open a dictionary
>>8503008Tried it many times, never worked out for me
I was suprised pole dancing isn't banned, the outputs are pretty jank but it has potential
>>8503018damn. Same results here.
Think she has had enough?
think this is the first they deleted right after I saved it
nu>>8502712>>8502712>>8502712
>>8502438I really like how intolerant you are of bullshit. An how you demonstrate its bullshit. Thanks for not giving up.
>>8501619the future is bright