Is it possible to create own JAV using AI technology?
- Thread starter Muz1234
- Start date
-
Akiba-Online is sponsored by FileJoker.
FileJoker is a required filehost for all new posts and content replies in the Direct Downloads subforums.
Failure to include FileJoker links for Direct Download posts will result in deletion of your posts or worse.
For more information see this thread.
BUDDY
I've had some time here and there to start getting this together. I now think I have the source images I need, next is to generate some video clips and see what comes of it all. I even found a nice (and free) Japanese pop style soundtrack to go with it. No dialogue, though. I'm not quite there yet and not sure I'd like to even put my feet into that yet.
Here is my working checklist for the upcoming release of my super-duper-AI-generated JAV trailer. Working title:
My Favorite Hospital, Smiling Nurses Doing Good Work, I Can't Believe My Eyes
Green is selected video clips. Orange is selected images, need video generations. Yellow is waiting on image selection.
My Favorite Hospital, Smiling Nurses Doing Good Work, I Can't Believe My Eyes
Green is selected video clips. Orange is selected images, need video generations. Yellow is waiting on image selection.
Well I guess since you need deeper clarification, what type of JAV will you make. You probably like stuff from 2003 like every other new person joining this site.
If I could make ai JAV, it would be based on those Incest hentai, with the x-ray cumshots
Last edited:
Of course! Here I am thinking most might make something simple or "standard" but it's AI for crying out loud. You can make anything you can think of! Do you have an example of that?
Intriguing. I would say it is technically possible but practicably impossible. I have tried using LTX desktop with a 16Gb Nvidia GPU. You start with an image, give the model a prompt and it generates a video from that point. For 1080p the max length is 5 seconds.
For my test, I started with this image.
and prompted for the man to slap her arse a couple of times whilst continuing to **** her.
Problem 1 - the model pretty quickly deleted the gentleman's ding dong, rendering him incapable of any further action.
A further problem is that the original image does not contain the mans face, so it is hallucinated...
I guess what this means in practical terms is that every 5 second snippet will have to contain pretty clear images of the participants faces.
On a side note, it is VERY difficult (if not impossible) to get the model to follow the prompt. In the example aboce, he did not once slap her arse.
I have also been working on this image ...
trying to get the man to grab the womans right ankle. I have tried about 20+ prompts so far with no success. At low resolutions clip generation takes about 20-30 seconds. If you get a good result and want to improve the resolution you need to use another tool as it is apparently impossible to get LTX to generate exactly the same video even with the same source and the same prompt.
Basically the statistical bias of the training data will outweigh the prompt almost every time. Because the model has not been trained on a man grabbing a womans ankle, all instructions fail. Instead, disappointingly, there is a lot of unprompted neck grabbing.
Overall the short term future does not offer much promise.
For my test, I started with this image.
and prompted for the man to slap her arse a couple of times whilst continuing to **** her.
Problem 1 - the model pretty quickly deleted the gentleman's ding dong, rendering him incapable of any further action.
A further problem is that the original image does not contain the mans face, so it is hallucinated...
I guess what this means in practical terms is that every 5 second snippet will have to contain pretty clear images of the participants faces.
On a side note, it is VERY difficult (if not impossible) to get the model to follow the prompt. In the example aboce, he did not once slap her arse.
I have also been working on this image ...
trying to get the man to grab the womans right ankle. I have tried about 20+ prompts so far with no success. At low resolutions clip generation takes about 20-30 seconds. If you get a good result and want to improve the resolution you need to use another tool as it is apparently impossible to get LTX to generate exactly the same video even with the same source and the same prompt.
Basically the statistical bias of the training data will outweigh the prompt almost every time. Because the model has not been trained on a man grabbing a womans ankle, all instructions fail. Instead, disappointingly, there is a lot of unprompted neck grabbing.
Overall the short term future does not offer much promise.
LOL sorry, that made me laugh!
Maybe it's the model you're using, is it a base LTX? There is another newer(ish) one I saw that is supposed to be better with NSFW because it was trained only on people. I think... Also, have you tried using a LoRA that's NSFW if there are any for LTX? I'm still using Wan. After I finish my trailer I'll take a crack at your images and see if Wan has any better luck.
