Fan Art AI Netorase Phone (Le Stag) - Fan Art Thread

[Flippie_86]

Think there's any chance Le Stag will include the possibility to add mods? I could 100% see some of this fan art/short stories fitting into the setting, even if they aren't all canon

On his discord he has mentioned he has some ideas for "what if" style side stories to the game at some point.

 

[aqd]

Since Le Stag posted a poll earlier on coherent and non-coherent pictures, I started to think: "How would it look if Laura used mirrors to take the pictures?", and, well, here's the result:

View attachment 4210976 View attachment 4210975 View attachment 4210974 View attachment 4210973 View attachment 4210972 View attachment 4210971 View attachment 4210970 View attachment 4210969 View attachment 4210968

Love the art style. These are some of the best images I've seen. Great work.

 

[JohnLange]

OK one more Halloween-themed one.

Ashley, you ruined my spell! I didn't want my boyfriend's FACE "like a horse"! It was supposed to be, um, something else...

View attachment 4146489

Let me just check... yeah that part's the same. (sigh....)

Does this image generation setup know how to do animal genitalia?

 

[aqd]

First attempt using Civit.





I just installed A1111. Will be attempting to get comfortable with that.

 

[aqd]

Who knew Laura had a little sister? (this was totally accidental)

 

[nickohoma]

As a unity developer, there's not actually much to it, you could reproduce the interface in a couple of hours or just grab an asset to do it for you, they're fairly common and inexpensive. You could also just decompile the entire game in about 30 seconds if you care to, C# decompiles cleanly and often even includes comments. What makes modding a total pita with this particular game is that for some godforsaken reason he's decided to use scriptable objects instead of a database or csv.

End of the day, it's really the writing + likable character designs and escalation that makes this special vs other implementations of the same basic mechanics.

Obligatory Laura:

She's really happy you let her work at that club.

Edit: I forgot a few, oops

I know jack squat about Unity so I'm certain you know better than me. I just remember earlier versions of the game having a few technical issues that the dev had to work through, so figured he put a fair amount of work into hammering out his interface. I was surprised when I didn't find a drop-in "phone simulator" Unity plugin. Last I checked there was a dude on Patreon saying he was gonna make one but no idea if that will happen.

Someone could probably do a VN side-story in RenPy easily enough though. Which might be good anyways to break out of the phone interface limitations.

 

[nickohoma]

Does this image generation setup know how to do animal genitalia?

It definitely does. Change two keywords in your prompt and your characters are suddenly anthro horses with the anatomy to match.

I've respectfully declined to make those pics though. Just not my bag.

 

[knightblade155]

I forgot my a few of my favorites, oops,

I know jack squat about Unity so I'm certain you know better than me. I just remember earlier versions of the game having a few technical issues that the dev had to work through, so figured he put a fair amount of work into hammering out his interface. I was surprised when I didn't find a drop-in "phone simulator" Unity plugin. Last I checked there was a dude on Patreon saying he was gonna make one but no idea if that will happen.

Someone could probably do a VN side-story in RenPy easily enough though. Which might be good anyways to break out of the phone interface limitations.

He's using Dream OS

 

[aqd]

Huh. first attempt at A1111 was not so successful. Though I got the checkpoints and Loras all in the right spots, but the style is very wrong, and it's not looking like Laura. Any advice for a total newb nickohoma ?

Used this image for the prompts with only very minor edits:

You must be registered to see the links

Spoiler: Not Our Laura

You don't have permission to view the spoiler content. Log in or register now.

Closer?

Spoiler: Still not Laura, but better?

You don't have permission to view the spoiler content. Log in or register now.

 

[nickohoma]

Huh. first attempt at A1111 was not so successful. Though I got the checkpoints and Loras all in the right spots, but the style is very wrong, and it's not looking like Laura. Any advice for a total newb nickohoma ?

Used this image for the prompts with only very minor edits:

You must be registered to see the links

Spoiler: Not Our Laura

You don't have permission to view the spoiler content. Log in or register now.

These are the most likely culprits.


Width and height will usually be either 1024x1024, 832x1216, or 1216x832. Those are the resolutions the model was trained on so things get weird if you stray from those. There are a few others that work but those are the most common.

In A1111, when you select your LoRA's it should put them in your prompt <like this:1>. The number after the colon is the strength, which you can adjust. But they will be in your prompt, unlike in Civitai. If they're not in your prompt they're not being applied. You can click them in the LoRAs tab to add them to your prompt though.

And just double-check you have the Pony model selected as your checkpoint. Sampler could make a difference too but not as big a difference as in your pic.

 

[knightblade155]

These are the most likely culprits.

Just a heads up, but A1111 is very outdated and shouldn't be used anymore. Forge has the same UI, compatibility with many of the same extensions and is massively faster (when used with the appropriate --xformers --cuda-malloc flags)

 

[knightblade155]

You must be registered to see the links

I trained a lora on Laura, actually made a full Pony finetune too accidentally but it's a bit too big to upload easily. Activation text is NTRSLaura. Do not use all the other loras with it, they're not needed. Works best on Pony or AutismMix, but should be fine on any pony derived finetune

I recommend a style prompt as follows:

You must be registered to see the links

 

[aqd]

These are the most likely culprits.
View attachment 4212791

Width and height will usually be either 1024x1024, 832x1216, or 1216x832. Those are the resolutions the model was trained on so things get weird if you stray from those. There are a few others that work but those are the most common.

In A1111, when you select your LoRA's it should put them in your prompt <like this:1>. The number after the colon is the strength, which you can adjust. But they will be in your prompt, unlike in Civitai. If they're not in your prompt they're not being applied. You can click them in the LoRAs tab to add them to your prompt though.

And just double-check you have the Pony model selected as your checkpoint. Sampler could make a difference too but not as big a difference as in your pic.

Ah, I was not calling out the LoRAs! Thank you. Also need to restart to get the last one (body type) to load.

Spoiler: Maybe Laura?

You don't have permission to view the spoiler content. Log in or register now.

 

[nickohoma]

Just a heads up, but A1111 is very outdated and shouldn't be used anymore. Forge has the same UI, compatibility with many of the same extensions and is massively faster (when used with the appropriate --xformers --cuda-malloc flags)

When I was choosing between A1111 and Forge, this announcement swayed me away from Forge

You must be registered to see the links

They may have since pulled back from that, or maybe it was much ado about nothing. But I went down the A1111 path. I had started with ComfyUI but couldn't figure out regionals/couple_mode in Comfy.

Whatever people want to use is cool with me. I just know A1111 better than the alternatives so that's what I use.

 

[nickohoma]

Ah, I was not calling out the LoRAs! Thank you. Also need to restart to get the last one (body type) to load.

Spoiler: Maybe Laura?

You don't have permission to view the spoiler content. Log in or register now.

Just FYI, the Body Type one is not in Le Stag's Lora combination. I just add that one at low strength because I like it. If you look at Le Stag's images though, I think he just uses the Disney Princess one, ExpressiveH, Detail Tweak, and Smooth Anime.

 

[Fleis]

These are the most likely culprits.
View attachment 4212791

Width and height will usually be either 1024x1024, 832x1216, or 1216x832. Those are the resolutions the model was trained on so things get weird if you stray from those. There are a few others that work but those are the most common.

In A1111, when you select your LoRA's it should put them in your prompt <like this:1>. The number after the colon is the strength, which you can adjust. But they will be in your prompt, unlike in Civitai. If they're not in your prompt they're not being applied. You can click them in the LoRAs tab to add them to your prompt though.

And just double-check you have the Pony model selected as your checkpoint. Sampler could make a difference too but not as big a difference as in your pic.

Thank you for coming to my TeD Talk!

Jokes aside, it's awesome how you're helping out everyone! Kudos to you, good sir (or lady)

 

[Baleygr]

Laura, Nathan and MC went to the beach.



After enjoying the beach for a while, they decided to have a different type of fun.



Together, they gave Laura what she wanted.



It had been a fun day. A day to remember.



Months has passed, and it turns out that it wasn't only memories that was created that day.



Wonder who the father is...

 

[nickohoma]

Wonder who the father is...

It's both! Heteropaternal superfecundation is a real thing, not just limited to greek myths.

 

[Cavalo 23]

‍☠

 

[aqd]

Trying to get images for a scene with Laura's little sister coming to visit, but damn am I now understanding how hard it is to make images with two (let alone three) people in the room. I need to go back and study the posts on making regions and inpainting and all. Because generating like 100 images and picking through for the one right one isn't it. The one with the right number of hands, arms that aren't merged, oh shoot now the Laura and her sister each got one eye of each color... and then keeping clothes and such consistent across a scene. Boy is that a lot of moving parts.

For instance, near identical prompts:


I've got some practice to do, clearly.

Spoiler: Realistic Laura?

You don't have permission to view the spoiler content. Log in or register now.