Unity !Ω Factorial Omega: My Dystopian Robot Girlfriend [v0.87.11] [Incontinent Cell]

[Kaiki-tan]

Well if you couldn't tell I was mainly suggesting speech synthesis anyway. I even mentioned ChatGPT probably isn't the best idea yet.

Touché then lad . Though I'd like to add; while it might not be a bad idea, speech synthesis is still a dozy as well, not as impractical as AI dialogue, but most still sound jarring or void of any consistency/expression (just because the game is about a robot don't mean it should sound like Siri lol), at that point you'd be better off just hiring VAs on a finished script, it'd likely amount to the same price anyway for any decent synthesis..

 

[JakeTapper]

Touché then lad . Though I'd like to add; while it might not be a bad idea, speech synthesis is still a dozy as well, not as impractical as AI dialogue, but most still sound jarring or void of any consistency/expression (just because the game is about a robot don't mean it should sound like Siri lol), at that point you'd be better off just hiring VAs on a finished script, it'd likely amount to the same price anyway for any decent synthesis..

Have you seen how good some speech synths have become? This isn't Siri crap. There are even ones that can run on a local model so don't even require a net connection. Check the video I linked about the one they did with Skyrim.

 

[Incontinent Cell]

This game is starting to scream for additional chatgpt and or speech synthesis implementation. I made a Unity game that can do both (handles the speech synthesis part really well and can integrate with both Coqui and ElevenLabs). If the dev wants to reach out I really don't mind sharing my code with him.

I can understand where chatgpt features might not be the best idea yet (unpredictable, might break lore, needs jailbreak etc), but the speech synthesis stuff is worth looking into. Could easily run all of the pre-written dialogue through a service in a short amount of time with a script. Hell, you could even let users create/choose their own voices and re-render all of the voice lines (as long as they have their own API key). The best part is this is literally a robot so synthesized voicing fits really well.

I will happily add AI to my game once two conditions are met:
1. It's open source.
2. It can be run locally on an average device.

Which is still at least couple of years away.

Also to people waiting for new update. We've started prerelease testing phase with our patrons. It will probably be released to all our patrons in a couple of days.

 

[Kaiki-tan]

Have you seen how good some speech synths have become? This isn't Siri crap. There are even ones that can run on a local model so don't even require a net connection. Check the video I linked about the one they did with Skyrim.

I just saw it, you're right it's pretty decent . Still notably it is made much easier when you already have so many clear audio files of characters in Skyrim to use similar to what people are doing with voice deep fakes hence why it sounds quite accurate.

Though like I said it's not a bad idea at all, but it is definitely still a new technology that needs a couple more years on its record especially for good open source projects (as Incontinent mentioned as well).

 

[ZixSum]

This game is starting to scream for additional chatgpt and or speech synthesis implementation. I made a Unity game that can do both (handles the speech synthesis part really well and can integrate with both Coqui and ElevenLabs). If the dev wants to reach out I really don't mind sharing my code with him.

I can understand where chatgpt features might not be the best idea yet (unpredictable, might break lore, needs jailbreak etc), but the speech synthesis stuff is worth looking into. Could easily run all of the pre-written dialogue through a service in a short amount of time with a script. Hell, you could even let users create/choose their own voices and re-render all of the voice lines (as long as they have their own API key). The best part is this is literally a robot so synthesized voicing fits really well.

OpenAI, why? If you implement the functionality with a Local LLM then you wouldn't need an internet connection or an API Key.
WELL, I guess that's a legal gray area, all of the current Local LLMs are pre-trained on leaked code. I'd recommend just waiting for an open source model not based on leaked code to come out and implement one of those into the game if the DEV goes that route.

I will happily add AI to my game once two conditions are met:
1. It's open source.
2. It can be run locally on an average device.

Which is still at least couple of years away.

Also to people waiting for new update. We've started prerelease testing phase with our patrons. It will probably be released to all our patrons in a couple of days.

I should have read a few messages ahead.
It seems we agree, I didn't even need to type this message heheh.

Can't wait to play the new version! Weirdly enough I had a feeling to check the thread today and I'm not disappointed by this update.

 

[JakeTapper]

OpenAI, why? If you implement the functionality with a Local LLM then you wouldn't need an internet connection or an API Key.
WELL, I guess that's a legal gray area, all of the current Local LLMs are pre-trained on leaked code. I'd recommend just waiting for an open source model not based on leaked code to come out and implement one of those into the game if the DEV goes that route.

I was gonna mention that too but a lot of these models won't run on the average PC as they're pretty high in VRAM demands. They can also be pretty annoying for the typical end user to set up. I did manage to get Dalai Alpaca 7B working but have had trouble getting the higher node models to load (says something about bad magic). Is this what you were talking about or are there other local models worth looking into? I'm genuinely curious what else is available.

 

[DudeBro79]

All this discussion of running the AI directly on the user's machine...

I was under the impression that AI would just be used to generate voice lines that would simply be recorded and packaged in the game.

 

[sinclairl]

All this discussion of running the AI directly on the user's machine...

I was under the impression that AI would just be used to generate voice lines that would simply be recorded and packaged in the game.

But that would actually make sense.

 

[wetnoodle32]

if you want infinite dialogue, you need the ai running. if you just want voice lines, then just put in the voice lines. simple as.

 

[Kaiki-tan]

You know how when you interact with Jun she might say something to you and then there's a quick little back-and-forth between you two? Or how you have the option to "chat" with her which basically does the same thing, a quick back-and-forth, or simply "we chatted for a while"? Now it'll bring up chatbot. Jun could have new lines that weren't pre-written, or you could hold a meaningless and probably nonsensical conversation with her.

Personally I think the writers should be the ones writing and the focus should stick to telling the story that's already being told. And tits. Focus should be on tits.

Spoken like a true man of culture brother! Although I am more of an ass-man myself .
Still I think there's something cool about that sentiment I'd love to see in the future; is instead of relying on AI completely, you get writers creating the story, lore and majority of main dialogue of "X" game and feeding it all into an AI then seeing how it goes. At the end of the day, AI can't replace creative mediums as it is still just a bunch of machine learning meant for demanding tasks yet only uses a combination of content fed to it as it is incapable of creating original content on its own.

 

[ViviX12]

You know how when you interact with Jun she might say something to you and then there's a quick little back-and-forth between you two? Or how you have the option to "chat" with her which basically does the same thing, a quick back-and-forth, or simply "we chatted for a while"? Now it'll bring up chatbot. Jun could have new lines that weren't pre-written, or you could hold a meaningless and probably nonsensical conversation with her.

Personally I think the writers should be the ones writing and the focus should stick to telling the story that's already being told. And tits. Focus should be on tits.

yea I'd rather "We chatted for a while" than trying to wrangle JunGPT into saying something that makes sense in a conversation that wouldn't even impact anything about the game no matter how 'meaningful'

 

[JakeTapper]

You want information on the different AI models? I think I can catch you up.

In terms of setup I use oobabooga/text-generation-webui (API) and Cohee1207/SillyTavern (WebUI) in conjunction, there are lots of nice tutorials that cover how to set that stuff up but it's pretty straightforward.

The model I use the most because it's the most coherent and uncensored one is anon8231489123/gpt4-x-alpaca-13b-native-4bit-128g but if you don't care about censorship in your models anon8231489123/vicuna-13b-GPTQ-4bit-128g is also a great option. Only use 13b parameter models if you have more than 10gb of VRAM, if you have only 8gb of VRAM I'd recommend a 7b parameter model like usamakenway/WizardLM-7B-uncensored-GPTQ-4bit-128g or pygmalion-7b-4bit-128g-cuda.

gpt4-x-alpaca is a good general uncensored model, vicuna is a good general censored model. Pygmalion is only good at nsfw roleplay, shygmalion is EXTREMLY NSFW and WizardLM is another good general chat model.

Despite saying some of these models are "only good at" specific things they all pass the turing test for sure during general conversation.

I always try to use GPTQ models because they tend to use less VRAM and use less disk space.

If you want more information I actually have a pretty long list, I only explained the first few that came to mind hopefully this helps. If you need help setting stuff up we can chat more, I don't mind.
Have fun!

Thanks man this is some good stuff! Gonna check out those web UIs and the gpt4-x-alpaca-13b as I have a 4090 so should be able to handle it fine as long as I can get it installed right. When I played around with dalai I could only get vanila alpaca 7b to work for some reason but I suspect the install had some issues. So the web UI has an API I can use? That will be very helpful.

yea I'd rather "We chatted for a while" than trying to wrangle JunGPT into saying something that makes sense in a conversation that wouldn't even impact anything about the game no matter how 'meaningful'

Thing is a large portion of IRL conversations are meaningless as well . I agree that due to this, raw gpt chat itself is typically not good for a game setting (although certainly wouldn't hurt to make as an option) but the way you make stuff like this have meaning right now is by having it fill in for repetitive lines such as reactions to being touched. You can also have the game come up with radiant requests such as 'buy x item' and then use gpt to generate the dialogue for the request given the parameters of what it wants you to do.

There are definitely ways to implement this that feel meaningful and help alleviate the repetitive nature of cohabitation games, but regardless it's absolutely true the VRAM requirements and end user setup make this impractical as of now and would prob need to be a toggleable option.

 

[Astcit]

For fuck sake this game is fucking depressing as fuck... oh wait, no, my life is the depressing bullshit.
Nice game, I like to pat her head.

 

[ZixSum]

Thanks man this is some good stuff! Gonna check out those web UIs and the gpt4-x-alpaca-13b as I have a 4090 so should be able to handle it fine as long as I can get it installed right. When I played around with dalai I could only get vanila alpaca 7b to work for some reason but I suspect the install had some issues. So the web UI has an API I can use? That will be very helpful.


Thing is a large portion of IRL conversations are meaningless as well . I agree that due to this, raw gpt chat itself is typically not good for a game setting (although certainly wouldn't hurt to make as an option) but the way you make stuff like this have meaning right now is by having it fill in for repetitive lines such as reactions to being touched. You can also have the game come up with radiant requests such as 'buy x item' and then use gpt to generate the dialogue for the request given the parameters of what it wants you to do.

There are definitely ways to implement this that feel meaningful and help alleviate the repetitive nature of cohabitation games, but regardless it's absolutely true the VRAM requirements and end user setup make this impractical as of now and would prob need to be a toggleable option.

The examples you provided for improving the game was exactly what I was envisioning, just a way to change up repetitive dialogue, story related conversations could be made unique in composition, but it would make more sense for those to be hand written. Depending on the prompts given to the AI emotional state and relevant context could even be used for the AI to react to the unique situation it's in, something that would be much harder to implement with purely hand-written dialogue. The unique dialogue for radiant requests would be an easy way to add "infinite" content lol.

The text-generation-webui is a basic webui but it also has an API plugin that you can enable, SillyTavern is a secondary webui that can use that API connection and it focuses on adding extra features like long-term memory and saving and loading conversations and different backgrounds and displaying character portraits.

If you have a 4090 you could actually get away with running a 30b parameter model as long as it's a GPTQ one, I don't have enough VRAM to run a 30b parameter model so I don't know how much better they are but feel free to share your results if you end up testing one of them. If you have any problems feel free to msg me.

 

[AveryFlare]

Great game. I patted her head so much she gave me heart eyes. It... activated something in me and I couldn't stop myself. Lovely game 10 outta 10 devs know what they're doing.

 

[mr__person__]

Great game. I patted her head so much she gave me heart eyes. It... activated something in me and I couldn't stop myself. Lovely game 10 outta 10 devs know what they're doing.

Be careful, head pats are a gateway drug into way more deplorable acts, like nose boops or, god forbid, hand holding

 

[MEMEACCOUNT]

Be careful, head pats are a gateway drug into way more deplorable acts, like nose boops or, god forbid, hand holding

Please refrain from speaking of such deplorable degenerate acts...

 

[AveryFlare]

Be careful, head pats are a gateway drug into way more deplorable acts, like nose boops or, god forbid, hand holding

I need them. NOW!!!



Honestly though, this game is a lot more wholesome that I initially thought. I haven't played any other games that really struck the balance between the wholesomeness and the sexual stuff like this except for Life with a Slave and that was ages ago. Really enjoyed my time with it and cannot wait for even more content in the future for both the wholesome girlfriend part and also the other part.

 

[mr__person__]

I need them. NOW!!!

View attachment 2652127

Honestly though, this game is a lot more wholesome that I initially thought. I haven't played any other games that really struck the balance between the wholesomeness and the sexual stuff like this except for Life with a Slave and that was ages ago. Really enjoyed my time with it and cannot wait for even more content in the future for both the wholesome girlfriend part and also the other part.

I mean if you're looking for that while waiting for the next update here you could check out "Imouto! Life ~Fantasy~" and "Imouto!? Life ~Monochrome~"

 

[AveryFlare]

I mean if you're looking for that while waiting for the next update here you could check out "Imouto! Life ~Fantasy~" and "Imouto!? Life ~Monochrome~"

Ah those games. I remember checking them out a while ago but completely forgot about them after a play session. Not sure what happened really but, I don't mind giving them a try again. Thanks for the recommendation!