Novel AI for Vore Game Art

[iforgot]

https://cdn.discordapp.com/attachments/ ... 2/Xyla.rar
https://cdn.discordapp.com/attachments/ ... Kagari.rar
here is the correct link + another one

[cankoo]

How do you make more than one image with the same character? Even when I use the same reference image I often get wildly differing results. I tried messing with the sliders on the right and that doesnt help. I dont think Im using this software correctly. Any tips?

[SaioKisaragi]

How do you make more than one image with the same character? Even when I use the same reference image I often get wildly differing results. I tried messing with the sliders on the right and that doesnt help. I dont think Im using this software correctly. Any tips?

I'm not sure exactly what the problem is but if you're talking about rendering more than one image at a time then there is this slider.
This will allow you to render up 2 images at once. I hope this helps.

[cankoo]

How do you make more than one image with the same character? Even when I use the same reference image I often get wildly differing results. I tried messing with the sliders on the right and that doesnt help. I dont think Im using this software correctly. Any tips?

I'm not sure exactly what the problem is but if you're talking about rendering more than one image at a time then there is this slider.
This will allow you to render up 2 images at once. I hope this helps.

OP on the first page of this thread has proof of concept art of a character that has a consistent art style over several images. Even when I have the AI reference images that it made it still makes huge changes. Like, how do you make it so that is changes some things without completely rewriting everything?

[EnderTG]

Consistant descriptive tags and copying the seed should help.

[SaioKisaragi]

So I was testing the AI with the new furry diffusion and I think I have found a tag line that you can copy and use to get pretty reliable mawshots of any species you want.

drooling, open mouth, evil smile, front view, looking at viewer, head, tongue out,

You can add tags to the end to further customize what character you want. For example: "mouse, anthro, woman, brown hair" will give me a result like this:

Spoiler: show

You can also get some decent feral results if you leave out the tag "anthro".

Spoiler: show

Unfortunately, this tag line doesn't seem to work well with humans. Though I may experiment again and try to find a tag line that will work for humans. I stumbled on this by chance and experimented with it further.

I hope this helps! And feel free to check out some other mawshots I made while testing this.

Spoiler: show

[blyatman]

i have found that setting it too the furry mode instead of full anime helps better just make sure u set a "human" tag

[BasicNP]

Testing out some techniques

[FoxyVixen21]

I am currently creating an internal stomach LORA here are some examples, its not ready yet, but I think it is getting close.

Spoiler: show

[SaioKisaragi]

I am currently creating an internal stomach LORA here are some examples, its not ready yet, but I think it is getting close.

Spoiler: show

00141-1289351424-masterpiece, .png

00144-2857686687-masterpiece, .png

internalstomachgirl1.png

anti5.png

isg2.png

Wow those look pretty good! What tags did you use to get those results?

[FoxyVixen21]

Wow those look pretty good! What tags did you use to get those results?

I use these tags, but the are part of a model I am training, so you are unlikely to get the same results, but here:

masterpiece, best quality,
stomach internal, red theme, purple theme, organic, saliva, veins, tentacles, fluid-filled, dripping, chamber, acid, wet, transparent water, flesh, lumpy,
1girl, sitting, blue hair, green eyes, nude, crying,

and the negative prompts:
lowres, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, worst quality, low quality, normal quality, jpeg artifacts, signature, watermark, username, blurry, artist name,

other stats:
Steps: 25, Sampler: Euler a, CFG scale: 8, Seed: 3671157687, Size: 520x720, Model hash: 37a86dca58, Model: grapefruitHentaiModel_grapefruitv4, AddNet Enabled: True, AddNet Module 1: LoRA, AddNet Model 1: internalstomach3(2a84d3916636), AddNet Weight A 1: 0.8, AddNet Weight B 1: 0.8

[happy]

Hey how do you make your specifications in novel ai to make an image like it doesn't make sense to me. Is there a tutorial for this in vore

[FoxyVixen21]

Hey how do you make your specifications in novel ai to make an image like it doesn't make sense to me. Is there a tutorial for this in vore

I run a local copy of stable diffusion, I am also training my own LORA to make internal stomach pictures, which is how I get those results. I don't use the site novel ai, here a tutorial to step up your own:
https://rentry.org/voldy

[happy]

So what happens if your doing everything on your phone

[FoxyVixen21]

You can run it with Google Colab here are some ways to do it
https://rentry.org/sdgoldmine#browser

Google Colab is a cloud installed version

[ArtiArca]

So I was testing the AI with the new furry diffusion and I think I have found a tag line that you can copy and use to get pretty reliable mawshots of any species you want.

drooling, open mouth, evil smile, front view, looking at viewer, head, tongue out,

You can add tags to the end to further customize what character you want. For example: "mouse, anthro, woman, brown hair" will give me a result like this:

Spoiler: show

[ https://i.imgur.com/pvttQok.png ]

You can also get some decent feral results if you leave out the tag "anthro".

Spoiler: show

[ https://i.imgur.com/KJbATE3.png ]

Unfortunately, this tag line doesn't seem to work well with humans. Though I may experiment again and try to find a tag line that will work for humans. I stumbled on this by chance and experimented with it further.

I hope this helps! And feel free to check out some other mawshots I made while testing this.

Spoiler: show

[ https://i.imgur.com/jns1hVV.png ][ https://i.imgur.com/3xB7NtO.png ][ https://i.imgur.com/iBGvHuU.png ][ https://i.imgur.com/kuxFKbP.png ][ https://i.imgur.com/a2L3yZ8.png ][ https://i.imgur.com/Dk8p8wl.png ][ https://i.imgur.com/Hv2aAJo.png ][ https://i.imgur.com/AfmYY7C.png ][ https://i.imgur.com/cUpzU48.png ][ https://i.imgur.com/PEwFYx2.png ][ https://i.imgur.com/2hWiAmQ.png ][ https://i.imgur.com/qp3OcTZ.png ][ https://i.imgur.com/Zm6Xoj5.png ][ https://i.imgur.com/mKriwyd.png ][ https://i.imgur.com/MGGZ2a6.png ][ https://i.imgur.com/Zqotnag.png ][ https://i.imgur.com/j4Em5Kc.png ][ https://i.imgur.com/LCqBqa9.png ][ https://i.imgur.com/GplCP8j.png ][ https://i.imgur.com/V8FeWsB.png ][ https://i.imgur.com/iNjcyqi.png ]

This is really good stuff. I just wish they offered a trial of the image generator, even if it was just a couple of attempts. I don't want to throw down ten dollars and then discover I can't work it.

[hexcorsist]

This is really good stuff. I just wish they offered a trial of the image generator, even if it was just a couple of attempts. I don't want to throw down ten dollars and then discover I can't work it.

This is the exact reason I chose to use stable diffusion over Novel AI. It's a bit of an effort to install but the freedom of being able to run 100s of images and picking the best ones has made it well worth that investment

[Content2020]

This is my sucess rate

[SaioKisaragi]

Wow the bulges and expressions are really nice!

[Stompy]

[quote="Content2020"]This is my sucess rate

Looks great. Any chance you can send me the prompts and what model you used so I can try to replicate then.