Others Multiic [v0.5.22] [Stvoler]

[Creator13]

Hello, so for windows version I have to or don´t have download something?
for example, in help is written that for win the "kobold" is included, but it continue with recommendation of models to download. so, I have to download something and put it somewhere? :O for example "deepcogito" has 18,5GB - that I should download and then what?
Thank you

Edit: Ok, there is maybe some error:
Cannot find text model file: F:\Multic\resources\koboldcpp\saiga_nemo_12b.Q5_K_M
but the "model" is there...

Edit2: OK, maybe now I get it What model should I download for images. A lot of them are SFW / censored. what is best?

Edit3: hmm, so..it is working, but it generate only one image...and nothing more...

 

[HungLow247]

You can decide what you are before it starts, the game let's you set a name and gender for the AIs to interract with you

oh good , then ill try

 

[newasian]

I've been watching this game for a while but only started trying it out today. I really like it so far for the potential and infinite possibilities.

I have a question first. What model and settings did you use to generate the default images? I really like to achieve similar styles when using local SD generation.


I Also have a few suggestions for improvement.
- Allow users to add any chatbot API that uses OPENAI API standard. (3 inputs: A Base URL; API key; Model Name）
- More Character statues tracking, currently I only see orgasm counts. (pregnancy and children system maybe？Now, I do not know your implementation for chatbot RAG, but I do have some experience on my own, I usually write the reply format in json in their system prompt, and tell them strictly reply in json. So that I can easily parse the parts. ）
- Can you give the player the ability to modify the *hints* before sending as prompt? Most of the time, I find the hints are helpful as a starting base for my own prompt but I want to make some changes.

Again, really good work and thanks for the game.

 

[Blaziin]

Android download has virus attached to it. Make sure you scan it.

 

[AMSatis]

Is there any way to use a custom third-party api? I have a subscription on chub which supports OpenAI's chat completion endpoint, but I have no idea how to use it with this precious game. Can you help, pls?

 

[sehdal]

I hate ai designs

 

[fongfic]

Can any one suggest me how to play this game?

 

[IsaacOtto]

Can any one suggest me how to play this game?

It depends if you're using local models or online
If you use local models you can start them up from the settings. You can check the console up top to see when it's up and running.
And then you can start with your journey and have fun

 

[jbomb]

I hate ai designs

Then don't click them? There is actually an "exclude tag" Option, you can just exclude the tag "Ai cg"

 

[lowekz]

Why do I keep getting the error "kobold: failed to fetch" and "no choices in response data" when using the local model?

Tried changing models, tried changing the config, I always receive this error when I give a prompt. What am I doing wrong?

Edit: My game opens and the "loading model" never ends because of the error "Cannot find text model file", although the file is inside the folder already...

 

[PickingFool]

And how many of those 3TB games use AI with a local model installed?

I mean, come on, that's like saying "why is this car so big? I have a bunch of bikes that can carry me where I want and they're much smaller than a car. I expect a car to be as small as a bike because they serve the same purpose."

See the problem there?

I have to say, I agree. Go File is fast and SSDa are pretty cheap nowdays.

 

[Yeahboi82]

What lora weights should we use to generate images that are the same style as those already in the game? Also, mine keeps re-generating the same image over and over. Can we get an output to show what image gen prompt it's using?

 

[capii]

Let me try to give some tips on this game
For image generation, I read that it "reads" the last two message interactions, but for me it never really worked that well

What I do is change the prompt every 4 text messages so I can generate what I want, I will share some of the images I got from it

NEGATIVE PROMPT:
lowres, (bad quality, worst quality:1.2), bad anatomy, blurry, watermark, sketch, jpeg artifacts, ugly, poorly drawn, (censored), deformed, disfigured, missing arms, too many fingers, fused fingers, missing fingers, extra fingers, worst detail, futanari, clone , unimpressed, very displeasing, boring, normal quality, extra digits, low quality, bad, text, error, undetailed, distorted, extra, disembodied, extra digits, fewer digits, signature, bar censor, anthro, animal, multiple view, transparent background, simple background, motion lines, mosaic censor, bad hands, score_6, score_5, score_4, score_3, low score

score_9, score_8_up, score_7_up, masterpiece, best quality, amazing quality, hyper-detailed, raytracing, perfect eyes BREAK, 1girl, 1guy white tank top, hotpants, ponytail, casual, long legs, happy, standing, jerk off, big penis, pre-cum, kissing, she is kissing him and masturbating his penis, naked male

Notice after BREAK i input what I want for the current scene

I use LORA and disable image generation on changing places

I generally use
20-22 steps
cfg between 4 and 6
clip-skip 2
generate every 4

For the Loras
I Usually do
MoriiMee between 0.2 and 0.4
ponyv6 between 0.3 and 0.6


But this setting we have to play around more

 

[jager20150]

Let me try to give some tips on this game
For image generation, I read that it "reads" the last two message interactions, but for me it never really worked that well

What I do is change the prompt every 4 text messages so I can generate what I want, I will share some of the images I got from it




Notice after BREAK i input what I want for the current scene

I use LORA and disable image generation on changing places

I generally use
20-22 steps
cfg between 4 and 6
clip-skip 2
generate every 4

For the Loras
I Usually do
MoriiMee between 0.2 and 0.4
ponyv6 between 0.3 and 0.6


But this setting we have to play around more

thanks i now know how to use this

 

[HungLow247]

Not sure i want to allow the network connection considering the negatives. no boobs on the first chick . pass

 

[newasian]

And this is why this game is so fun.

 

[lowekz]

I managed to make the game work, but why is it so laggy after a few prompts?

I'm running it with 32gb RAM and nvidia rtx 5060.

 

[newasian]

I managed to make the game work, but why is it so laggy after a few prompts?

I'm running it with 32gb RAM and nvidia rtx 5060.

When running AI app, VRAM is what matters (and cuda core count). Basically you are running chatbot locally, even a watered down version. 5060 has only 8GB vram, it is absolutely shit piece for AI. when vram is full, the inference process will start moving data to cpu and ram which has a only few core compare to GPU's thousands of cuda cores.

I recommend use online API for chatbot and local SD for image gen. The online models runs usually a lot smarter than what you can run locally. I use deepseek v3(paid version),its good enough for porn writing and a lot cheaper than any other. You are rich you can go for claude or gemini 2.5.

 

[FatYoda]

Does this have a story mode or is it just pick a girl and do whatever you want with her?

 

[IsaacOtto]

Does this have a story mode or is it just pick a girl and do whatever you want with her?

Both