CREATE YOUR AI CUM SLUT ON CANDY.AI TRY FOR FREE
x

UnceliusM

Member
Jan 10, 2020
118
119
Well unfortunately I think none of these fit ^^ Quite overthetop without nailing the emotions we are going for
Slaving along, here i tried to have relaxed eyes, closed-smile, euphoric/dreamy eyes and different seeds
You don't have permission to view the spoiler content. Log in or register now.
Getting any warmer?

EDIT: Just had a thought. What if we could take advantage of this heavy leg work and use these Helen emotions as bases for generating the emotions for the other characters. Could be doable.
 
Last edited:
Jun 28, 2020
188
94
Or you're not opening the pictures with a gif opener (y)

In game they are read as animation :KappaPride:
Tried to open them with a gif opener, didn't work, tried to open them in the browser, neither and in game there is no animation, I have seen some gifs work fine in the game but these aren't animated. That's why I think I have the wrong media pack.
You don't have permission to view the spoiler content. Log in or register now.
The pics
You don't have permission to view the spoiler content. Log in or register now.
You don't have permission to view the spoiler content. Log in or register now.
You don't have permission to view the spoiler content. Log in or register now.
You don't have permission to view the spoiler content. Log in or register now.
You don't have permission to view the spoiler content. Log in or register now.
You don't have permission to view the spoiler content. Log in or register now.
 

qwertyu12359

Jack-o-nine-tails
Game Developer
Aug 1, 2017
1,638
1,823
Slaving along, here i tried to have relaxed eyes, closed-smile, euphoric/dreamy eyes and different seeds
You don't have permission to view the spoiler content. Log in or register now.
Getting any warmer?

EDIT: Just had a thought. What if we could take advantage of this heavy leg work and use these Helen emotions as bases for generating the emotions for the other characters. Could be doable.
A lot of them would fit for "pleased"

But then let's compare my favorites one here (3, 10, 13, 14) and it seems like the face completely changed and became younger. So that's a hit for consistency and "milfness".
pleased.png 00365-770277594-euphoric_brown_eyes_closed-smile_euphoric_eyes_relaxed_eyes_lips.png


Therefore I reiterate, the original Euphoric and Ecstatic were a lot closer to hit the mark (y)
 

ImperatorAugustusTertius

Engaged Member
Sep 12, 2020
2,213
861
Tried to open them with a gif opener, didn't work, tried to open them in the browser, neither and in game there is no animation, I have seen some gifs work fine in the game but these aren't animated. That's why I think I have the wrong media pack.
You're right, not animated. I thought one of those scenes was animated in the past but we must have replaced it. I see the same images as you.
 

qwertyu12359

Jack-o-nine-tails
Game Developer
Aug 1, 2017
1,638
1,823
Tried to open them with a gif opener, didn't work, tried to open them in the browser, neither and in game there is no animation, I have seen some gifs work fine in the game but these aren't animated. That's why I think I have the wrong media pack.
You don't have permission to view the spoiler content. Log in or register now.
The pics
You don't have permission to view the spoiler content. Log in or register now.
You don't have permission to view the spoiler content. Log in or register now.
You don't have permission to view the spoiler content. Log in or register now.
You don't have permission to view the spoiler content. Log in or register now.
You don't have permission to view the spoiler content. Log in or register now.
You don't have permission to view the spoiler content. Log in or register now.
I'll atleast replace bj_purple, image is pixelated and girl has blue hair
 

UnceliusM

Member
Jan 10, 2020
118
119
A lot of them would fit for "pleased"

But then let's compare my favorites one here (3, 10, 13, 14) and it seems like the face completely changed and became younger. So that's a hit for consistency and "milfness".
View attachment 2142586 View attachment 2142588


Therefore I reiterate, the original Euphoric and Ecstatic were a lot closer to hit the mark (y)
I'll take those two then and run variations on them. Which parts of those do you not like? I could try changing mouths/eyes/nose separately.
 

30eruxa

Newbie
Mar 14, 2018
33
3
is enforced rule on a rule the slave wants to break only reducing pride by 1? I read from the wiki it's supposed to be reduced by 3 but it only ever reduce by 1 for me in game
 

UnceliusM

Member
Jan 10, 2020
118
119
You don't have permission to view the spoiler content. Log in or register now.
You don't have permission to view the spoiler content. Log in or register now.
You don't have permission to view the spoiler content. Log in or register now.
You don't have permission to view the spoiler content. Log in or register now.
 

ImperatorAugustusTertius

Engaged Member
Sep 12, 2020
2,213
861
is enforced rule on a rule the slave wants to break only reducing pride by 1? I read from the wiki it's supposed to be reduced by 3 but it only ever reduce by 1 for me in game
Depends on the rule. It's not the forcing that reduces pride, it's an effect of the rule itself, and not all rules reduce pride. Forcing a rule raises despair and awareness and habit, lowers devotion.
 

qwertyu12359

Jack-o-nine-tails
Game Developer
Aug 1, 2017
1,638
1,823
You don't have permission to view the spoiler content. Log in or register now.
You don't have permission to view the spoiler content. Log in or register now.
You don't have permission to view the spoiler content. Log in or register now.
You don't have permission to view the spoiler content. Log in or register now.
Euphoric alternate, with eyelashes for the win!

Last Ecstatic picture shocked me at the beginning, but with a bit of open-mindedness, I think it's so far the best. (y)

Praying pepe made me laugh ^^
 
  • Yay, update!
Reactions: UnceliusM

UnceliusM

Member
Jan 10, 2020
118
119
One thought I've had in terms of how to improve our results is to train the AI.
We make a shared data-set folder in google docs somewhere. Start collecting images of Successful Interpretations(the prompts accurately match the output) alongside each image's generation data(settings, seed, etc). People who are playing around with this software and are interested can contribute to this ever-growing dataset. In time we could have a dataset large enough to do some training.
There is also the matter of all those "niche" scenes and all sexual activities. In addition to downloading more images from the web we could perhaps utilize some of the art that is already compiled into image packs for the game to lead this AI deeper into the pits of depravity.
 
  • Like
Reactions: qwertyu12359
Jun 28, 2020
188
94
One thought I've had in terms of how to improve our results is to train the AI.
We make a shared data-set folder in google docs somewhere. Start collecting images of Successful Interpretations(the prompts accurately match the output) alongside each image's generation data(settings, seed, etc). People who are playing around with this software and are interested can contribute to this ever-growing dataset. In time we could have a dataset large enough to do some training.
There is also the matter of all those "niche" scenes and all sexual activities. In addition to downloading more images from the web we could perhaps utilize some of the art that is already compiled into image packs for the game to lead this AI deeper into the pits of depravity.
I don't know much about training, how many images are needed? eventually in the future that could be useful
 

UnceliusM

Member
Jan 10, 2020
118
119
I don't know much about training, how many images are needed? eventually in the future that could be useful
The more the better i guess. I'm not an expert on that myself, i've only ever done a couple of facial recognition bots.
Should do more research. There is also a thing called bias. Too much training on a single dataset will make the AI generate very closely to the dataset instead of being creative. There's also model combinations you can do in the web ui. You can merge models together. If we could perhaps find a more sex driven model out there somewhere, we could merge it with our current models and see what comes out.
 
Jun 28, 2020
188
94
The more the better i guess. I'm not an expert on that myself, i've only ever done a couple of facial recognition bots.
Should do more research. There is also a thing called bias. Too much training on a single dataset will make the AI generate very closely to the dataset instead of being creative. There's also model combinations you can do in the web ui. You can merge models together. If we could perhaps find a more sex driven model out there somewhere, we could merge it with our current models and see what comes out.
More sex driven? do you feel like the model lacks NSFW content?... or you mean more fetish oriented? I'm more interested in precision with the positioning of tattoos and piercings, also a way to make lamias more consistent, for those things merging a model sounds nice, an improvement over the current model, although without being able to extract the tattoos or piercings by means of differentiation between images to automate things, that, may be to huge of a manual task and a waste of time in terms of what is doable right now.
For niche scenes... haven't tried it yet but aren't those doable? the handcuffs aren't I already found out, but haven't tried torture and such things yet.
 

UnceliusM

Member
Jan 10, 2020
118
119
More sex driven? do you feel like the model lacks NSFW content?... or you mean more fetish oriented? I'm more interested in precision with the positioning of tattoos and piercings, also a way to make lamias more consistent, for those things merging a model sounds nice, an improvement over the current model, although without being able to extract the tattoos or piercings by means of differentiation between images to automate things, that, may be to huge of a manual task and a waste of time in terms of what is doable right now.
For niche scenes... haven't tried it yet but aren't those doable? the handcuffs aren't I already found out, but haven't tried torture and such things yet.
I haven't been able to produce anything really compelling. The AI knows what Cowgirl position is, sometimes knows what Missionary is and Doggystyle. Blowjobs and boobjobs that look good are rare. Anything that involves multiple characters however becomes quickly problematic.
I don't think the current model is capable of covering a majority of the sexual content we need.
 

qwertyu12359

Jack-o-nine-tails
Game Developer
Aug 1, 2017
1,638
1,823
One thought I've had in terms of how to improve our results is to train the AI.
We make a shared data-set folder in google docs somewhere. Start collecting images of Successful Interpretations(the prompts accurately match the output) alongside each image's generation data(settings, seed, etc). People who are playing around with this software and are interested can contribute to this ever-growing dataset. In time we could have a dataset large enough to do some training.
There is also the matter of all those "niche" scenes and all sexual activities. In addition to downloading more images from the web we could perhaps utilize some of the art that is already compiled into image packs for the game to lead this AI deeper into the pits of depravity.
Google docs though?

With google robots scanning so much content on their cloud, be careful, they might delete anything one day because "extreme" or "patch" content doesn't abide by their conditions of use
 
  • Like
Reactions: UnceliusM
4.00 star(s) 63 Votes