Video ai website

Forgot the link https://lumalabs.ai/dream-machine/creations

This seems interesting. And scary? Like The potential of this sends shivers down my spine.

it has an NSFW filter, which got quite upset with me. Couldn't get much out of it, because it mostly just ignored my text prompts/images.

>>25217
Same. Still looking for a good free image to video ai

found these on /g/.

>>25391
>Female in the background spontaneously transformed into a fat woman
>Female in background spontaneously impregnated 9 months or gained the equivalent in belly fat.
I think everyone would be too terrified to be disgusted.

>>25391
I love going to the Golden Corral and dosing.

It's rather stifling with its filter, but it IS free, even if it takes days to get the generations done.

It works best with images, don't even bother not to do something without a base frame.

This is pretty stupid but also kind of neat (AI in a nutshell)

i wonder if someone made a tutorial about how to make inflation content on luma. Maybe i could even make a remake of my bowsette video

i been doing some test on luma using an ai generated image and morph i made of headcannon bowsette and some sequences of Hellresident, some of them looks odd, i think the bowsette's one are the best looking ones

works pretty good with a start and end point attached.

>>25980
oh yeah the future is now

>>25956
what prompt did you use for those blueberries, especially the cop?

>>25986
no prompt, just hellresident's morph sequences, or you mean the "video prompt" i just use stuff like "woman bloating up" or something like that, clear stuff, or else Luma may get confused and do weird stuff

let this thread rolling.


This is interesting, I tested it on pairs of generated images in SD.
It would be very promising if it were possible to extend an existing video by inserting an image as the final frame to guide the process and maintain consistency rather than letting it float randomly.

And the generation speed is terrible. Most of the videos were in queue for 2-8 hours, crazy.

I'm even a little surprised that this thread hasn't been spawned already.

Oh baby we got a stew going.

Here are my results:

Loading times are ridiculous.

Nighttime downloads are faster.

>>26241
OMG - what I wouldn't give to see her coming down the hall towards me like that, or to get behind her on the stairs, or to be in an elelvator with her. Of course that wouldn't last long because I'd be fired foe sexual harassment the first day on the job.

>>26270💀💀💀

Is there a way to use this site without being signed into Google?

>>26503
I've been skeptical about this ai seeing some of the other videos posted here but this right here, THIS is the future, I gotta try this out for myself at some point

>>26529
JFC! That is amazing! Please keep those coming.

>>26478
I've been trying to find a way with no success.

>>26534
You should be aware that video generation wirh it is hit or miss. I've done 150 runs so far with five "meh" results.

>>26585

Just make a burner google account with a random name, then download an offbrand browser (edge or mozilla, etc) to sign in when you use it (so you don't have to log in and out each time).

>>26529
The stuff of dreams. This is the only type of AI that should exist.

I’ve been trying for weeks to get it to work with blueberry inflation images/morphs but the results have all been nightmare fuel. I’ve tried with descriptive prompts and very basic and can’t seem to get consistent results. If anyone can get anything with it please share!!

>>26614
Luma is worthless for prompt-based image generation. You have to generate the starting image elsewhere and upload it.

>>26173
Damn!! Finally a real benefit to AI! If we can use its power to remake Kelligrl’s content and bring her into the 21st century what else could we do??

Way wicked work tho! Pioneer god tier. Thank you.keep up the good work!

Wicked work! God tier pioneer!!!

Ive been in queue for 14 hours this website is worthless.

>>26707

I've been in queue for two days now. Not only has it been 4 per day limit, you can only submit 2 of those for back to back. Haven't tested how long you have to wait to be able to submit again after the 2 back to back limit.

>>26707
>>26720

Meh...

We must suffer for our art

Have we found any similar sites for ai video?
Everything I've come across looks kinda sketchy.

here's what I managed over the course of like a whole week :p

>>26758
Hands are clearly the most complex, inexplicable, utterly unknowable things in the universe to AI. If AI has a god, it's a human hand.

>>26758
Vidu, it uses SORAs architecture, and is free. Gives you 80 free credits per account per month, gens 4 second clips.

seems to be kess traffic. last few only took 10-12 hours to complete

Here's a few I made from some AI-BBW progressions. Cool program

>>26758

Hapier is decent, and you can use 3 photos. I actually made a couple with it too. Not nearly as impressive as luna though

Whoops this is the other one

>>26866
How is the nsfw filter on this one?
is it easy to prompt around it?

>>26868

No filter from what I can tell. Can also do 10 creations a day with no monthly limits or queues. Not nearly as good as Luma at interpolating between pics. Here's a few more I made.

>>26868

Do you have any prompts to get around the nudity filter in luma?

Made a few more in hapier. Find that with before and afters it works before to use the before as Pic 1 and the after as Pic 2 and 3 so that it doesn't try to figure out the middle on its own

>>26920
What kind of prompts are you using with Haiper? When I do long ones it gets confused. When I do short ones it fills in.

The result is always Cronenberg monsters that generate hands out of bodies and weird mutating transitions.

What was your process to make that?

Damn I'm gonna have to try Luma. I've seen some gens on deviantart watermarked with kling ai, has anyone used that?

>>27150
Kling ain't too bad. I'm still playing around with it. Though youll have to generate images elsewhere as it doesn't seem to go beyond small BBW range from my short experience. Same deal with text to video option. Image to video has kept the video consistent with the image as far as the subject is concerned. Also seems less restrictions than Luma.

It can make high quality images for like 0.20 credits each. Video for 10 and image to video for 20. Daily credit refresh is about 60. Text to video seems to complete within 15 minutes but the other can take all day.

>>27172

I actually did try running some images from Stable Diffusion through Kling but the results were messy. I barely tried with the prompts though, and any tips would be appreciated.

Does he seem tough to get around its filters though.

>>26860

I've spent an embarrassing amount of time playing around with luma since i posted these lol. This one is by far the best one luma has made for me. Got a few more from luma and hapier I'll post later

>>27182

Out of curiosity are you doing to get that effect? I assume it's a start and end picture, but mine end up horrible

>>27182
>>27183

Would love to hear more about your process as well, this is fantastic

>>27182
Consider me third. What are you using to make your start and end frames?

>>27198
>>27188
>>27183

I use before and afters. Either pre-made or I make the before image myself using a bunch of free programs.

>>27182 for this one I made the before using open ai and reference images, then slapped her face and arms on in an app. The after is a pic from AI-BBW (his stuff is great). I removed a few objects in the original including the face (had a bif slice of pizza in the way). I removed the original girl and reused the background for the before Pic so that everything was smoother. I also cropped them slightly different so that the frame would slide (seems to help things move more naturally).

After that I upscaled them and ran them in Luma. Somehow it came out perfect the first time, haven't gotten anything nearly as good from the same images and prompt since.

That's the biggest thing: trial and error. Usually takes around 3-5 tries to get something worthwhile.

Anyways here's a few more I've made. Please share any good B&A pics you can think of.

And a few more

And a few more

Whoops double post

>>27230
Do you mind sharing how different your before and afters look? I was wondering how you were getting camera work with consistent backgrounds.

I have mixed results with Luma. A couple have been really good, then three times in a row it just kinda zoomed in on the model's warped AI face.

Is there a good upscale online that's actually free or is that something where I really need a local install to process?

>>27252

I try to make them as consistent as possible. When making a before I use the same background by removing the woman from the after using the ai retouch app. Sometimes I'll outpaint if the picture is too zoomed in.

Here are the apps I use:
Ai retouch - remove or modify the woman in Pic

Image upscale (I tend to do 4x so I can use the face upscaler)
https://imageupscaler.com/upscale-image-4x/

Outpainter:
https://www.promeai.pro/outpainting-expand-images

This is the best free outpainter I've found, but it only let's you save the results in low res, so I use the upscale website after.

Keep in mind you need a Google account or email for the sites, and there are limited tokens. I made like 15 gmails to try to get around this lol.

I'll attach some of my before and afters here. These ones I made the before using open ai

This kind of shows my janky ass process lol started with an after. Removed the table and pizza. Fixed her face in ai retouch. Made a before in open ai using reference images (Google search for pose, after Pic for face and style) then messed around with it in ai retouch again.

The befores don't have to be perfect. Just close enough.

Keep in mind I have literally zero experience with anything graphic design or art related. So anyone can do this. Just takes time, trial and error, etc

>>27257
Shit, this is helpful. Is AI Retouch just an Apple app?

>>27260

I have it for Android. Probably an apple version too. I also use it to do my before pic background removal. They have a pro background removal thats better and you get 3 free tries on. But you can just uninstall and reinstall the app to get more tries.

I decided to try and make a few in Kling. The censor is more restrictive than luma which is annoying, but the results so far have been pretty good. Just takes forever to get results.

BTW does anyone know any other ai image creators that allow face, body, and style reference for free? I've run out of open ai credit.

>>27263
Do you know what the logo/creator listed is? I just didn't see it on the Google store.

>>27264
I've been playing with Stuffer AI but it might now be sophisticated enough for this.

I like the results I get from Kling better than Luma, Kling takes forever and Luma let's you add extensions.

>>27270

Logo looks like this. The new image references for stuffer ai are really good for poses.

>>27270

I've been fucking around more with the Pic retouch app. It's really smart considering it's a free app. Been making my before images simply using the object erase function. Here's a few examples using AI-BBW pics as the afters.most of these befores took less than 15 minutes to make.

One more

should have kept the backgrounds consistent but still the best i've gotten out of haiper so far

>>27348

This is really good. Who's the source for the pics? I'll have to post my hapier stuff.

>>27300
>>27340
How much do they actually give you for free before they start shaking you down for a credit card?

https://hailuoai.com/video

Here's another site that does video generation. It's pretty fast. Don't use the mobile version of this site, browser only.

I'd post examples of generation but it outputs mp4's and I'm too lazy to convert them to webm.

>>27361
bretty gud

>>27361
does it have restrictions?

>>27366

This is quality. What prompts are you using here?

>>27372
very basic. there's likely an llm on the frontend restructuring the prompt like bing. here's one from the first:

Scene: a busy hospital. An extremely obese woman in white bikini stands on a scale, being weighed by Pierce Brosnan. The scale reads 326 pounds.

>>27366
how do you evade the restrictions?

>>27366
also you think it may be able to make expansion videos, or only fat videos?

>>27376
try it and let us know.

>>27378
Doesnt ;-;, atleast for what i tried. The character remains skinny no matter what, or cancells my video due being nsfw or smt. Always something wents wrong, dang it.
"Busy lab, a worried skinny woman wearing bikini, her body bloats and swells like a fat balloon, then she suddenly is a extremely obese woman"

Someone knows how to fix this?

>>27340
What's the source on the girl in pink sweats?

>>27404

They're all from AI-BBW on devart

>>27407
Oh, she was paywalled so I scrolled right past her.

Kling really doesn't want to take a straightforward approach to this

>>27428
how do you make your sequences? i mean the two images

>>27430
Scroll up, they weren't my before and after images.

>>27430

I grab the afters from artists and make the befores in the ai retouch app. Here's a few more I made.

Takes zero talent or skill. Just need to get creative with it.

>>27428

I've had this issue as well. Try moving the creativity/relevance slider to 0.85. It won't freelance as much.

Hailuoai video generator

Hailuoai

>>27450
>>27453
Wow those two are really good. Sadlt i dont know how to make expansion videos

Has Luma been weirder than usual for you guys? I keep getting weird results.

>>27349

Those are just gens I made myself while messing around with ponyxl

>>27465
Not bad. How about this prompt?

600 pound 20-year-old woman, in a white tank top and blue jeans, waddles through the suburban neighborhood.

>>25189 (OP)
Much happier with the video results than I am from the image generating AI, very surprised by how bad those image generators look in comparison

>>27506
Few more

i tried to make some expansion videos on Hailluoai and im near to nail the method...But for a strange reason the inflatees always spin, someone knows why this happens?
"fantasy movie, a woman drinking a glass of water, her body swells and bloats and blows up into a huge extremely obese body"

>>27507 that's really sweet.
How about this prompt?
Happy 400 pound beautiful woman in a tight white blouse and sandals, round belly grows bigger, pop buttons, being feed a by nurse.

>>27524
>>27486
You can try it yourself. You don't have to type it here, the website is free to use.

>>27525
I just realized it. Last time I tried it, nothing was working. Thank God it's working now.

What was your prompt for the lady in the red top?

Been running my image sets through kling. Results have been better than luma by far. Here are a few of the better ones I got with this set. Will post more when I'm not busy

>>27574
How do you make the 2 images? I been struggling to figure out how to make 2 frames. Leonardo AI gives me too different results and other inpaiting AIs give me too similar results

>>27584

I run through my process here: >>27257

>>27584

Also here: >>27340

Just for fun, generated by hailuoai.
I like the first one, she seems excited, like it was a programmed date.
"Miss, lets start your feeding program"

More generated by Hailuo ai.
my favorite ones, specially the one on the left.
The action and expression generated, and the following gorging of the food.

more pieces

Someone have manager to make inflation videos with haillouai? It looks promising but i havent be able to make even a single inflation video

>>27585
ai retouch is only for android? Also yo use that one for the big changes, or you use prome ai for the big changes?

>>25189 (OP)
Another couple years and AI porn is going to be the new main. holy fucking shit.

>>27506
>>27593
moving images are easier to hide imperfections with
just make everything blurry and in motion (either the subject or the camera moving/zooming), add a bit of video compression and it obscures most of the AI funkyness.
the movements and such are still hella fucky tho. they have this weird, jerky, ungainly motion, weird postures and finer movements such as hands, feet, talking, chewing, blinking, turning one's head, turning one's upper or lower half and hair all seem to be things it still struggles with and i'm not convinced it'll be easily solved. particularly when it even in those videos the hands still look terrifying and body parts still continue to randomly expand/shrink as AI still struggles with perspective and foreshortening.
that last video in particular has genuinely horrible hands; with the hand closest to the camera/viewer growing additional digits only for most of them to disappear over and over while the hand furthest away seems to be twisted, distorted and suffering from a severe fungal infection in the nails.
i am not an anti-AI luddite but considering the processing power, data and the amount of rejects one gets from this i'm very much concerned that once the techbro cashflow stops that most of these services will be too expensive for the average person to utilize.

>>27340

I use prome to outpaint if the picture is cropped, then upscale it. Here's an example. I pretty much exclusively use ai retouch to make befores now. Depending on the picture/pose I can make a before in like 10 min.

>>27592

Here's a few befores I've made recently almost exclusively in ai retouch

A few more

This before was legit 2 steps. Erased 95% of the body and it filled in the rest itself. I've gotten some wild results for air and water inflation prompts. Even added a hose for some lol

Also someone knows a prompt for make inflation videos on Hailuoai?

>>27574
Damn, what do your prompts in Kling look like for something like this?

>>27602 I assume those last two reference water balloons, right?

>>27600
>>27601
Are you willing to share the videos you made with these?

Looks like Kling just rolled out a new model. Don't know how the new generation compares, but it doesn't do end frames.

1.0 now has motion path tools. I can't try them until my credits refresh, though.