Pretty sure that whoever gets close to the game renders, uses loras that they have trained from the game's images. There have been multiple discussions about this as well.
Fan Art My New Girlfriend [CircleGames] - Fan-art
- Thread starter Yougiblack
- Start date
- Jul 3, 2021
- 1,990
- 12,418
I don't have to "out" myself because I've been talking about it in the past. What you don't seem to get is that your posts almost seem like a threat. You're not making this thread a fun place to look at images for everyone, you're trying to hold it hostage. "You either put all the assets on the table, or you better stop posting alltogether".
If anyone wants to make a guide on how to create exact copies of ingame images, fine by me. If you're interested in my reasoning for not sharing what I was told a while ago, my posts are still up.
Have fun with your thread.
- Jan 14, 2020
- 143
- 267
My two cents: I agree that fan art isn’t necessarily meant to copy the original style (is there even really one?), and it can/should explore different artistic directions as long as the characters remain recognizable.
That being said, my personal preference leans toward styles that stick closely to the original, simply because it’s like bonus content—kind of like a poor man’s DLC! Expanding the game’s universe with, for example, short scenes like those from WhatsReal123 that fill in a cliffhanger or add depth to a situation is pretty cool.
I also understand that “gatekeeping” feeling. Clearly, the era of sharing for the greater good (like the pre-2000s internet) is long gone, and now everyone plays their own game. At least, that’s how it feels. Then again, it’s human nature. Let’s just say we’re no longer on an upward trend, but that’s just how the world is in 2025. Who’s to blame—the individual, society, the individual shaping society, or society shaping the individual? Plenty to debate there but not the place for…
In any case, I don’t think this is worth fighting over. If everyone does their best with what they have, that’s already great!
Personally, I love what I’ve seen in this thread—so much so that I’m adding a Fan Art gallery option to the main menu of my game in my next translation update. Players will be able to add their favorite images to a dedicated folder and view them in-game. I'll populate by default the folder with images from SymbioticLife, Ceagle, WhatsReal123, alex20152015 and a few others.
My current issue is figuring out how to credit the creators, but I don’t know how to solve it since users will have full control over the source folder and the images aren’t signed…
Anyway, just my opinion—not exactly decisive, but it is what it is...
That being said, my personal preference leans toward styles that stick closely to the original, simply because it’s like bonus content—kind of like a poor man’s DLC! Expanding the game’s universe with, for example, short scenes like those from WhatsReal123 that fill in a cliffhanger or add depth to a situation is pretty cool.
I also understand that “gatekeeping” feeling. Clearly, the era of sharing for the greater good (like the pre-2000s internet) is long gone, and now everyone plays their own game. At least, that’s how it feels. Then again, it’s human nature. Let’s just say we’re no longer on an upward trend, but that’s just how the world is in 2025. Who’s to blame—the individual, society, the individual shaping society, or society shaping the individual? Plenty to debate there but not the place for…
In any case, I don’t think this is worth fighting over. If everyone does their best with what they have, that’s already great!
Personally, I love what I’ve seen in this thread—so much so that I’m adding a Fan Art gallery option to the main menu of my game in my next translation update. Players will be able to add their favorite images to a dedicated folder and view them in-game. I'll populate by default the folder with images from SymbioticLife, Ceagle, WhatsReal123, alex20152015 and a few others.
My current issue is figuring out how to credit the creators, but I don’t know how to solve it since users will have full control over the source folder and the images aren’t signed…
Anyway, just my opinion—not exactly decisive, but it is what it is...
Thanks for sharing. Recently I asked someone to share which Lora's were used in a render, and the guy just kept it like a state secret. TBH I don't get this behavior, people act arrogant as they are real artists doing something special by himself only, but by the end of the day they are still using Loras based in someone else work to make NSFW renders from Disney or any other anime characters.
- Jul 3, 2021
- 1,990
- 12,418
Ever considered that some people may have spent countless hours to find and finetune loRA's that look somewhat similar to the game? You could've done the same, but I guess calling someone arrogant just because they didn't answer is easier.
And that ladies and gentlemen is how a good thread goes down the drain.
- Mar 17, 2019
- 336
- 3,244
I've always been quite open with what loras I use, in fact for SoA which I work on I made sure to advocate for us sharing all the various character loras we utilize - easily linked on the fan art page. I also have also pointed people in the direction of Melkor and Vintage, as long as I've known those two in many ways made up the "base" of the look for MNGF. That being said, I don't share out my homebrew loras always because they often utilize others work and I don't want to cheapen their labor (yes, I'm aware of the hypocrisy working in AI, but to my best knowlege Melkor isn't making a NTS VN or even an NTS comic so he's not direct competition) and also I often spent many hours refining those, setting up the prompts for the trainer and redoing them until I got good results.
I have helped at this point literally dozens of people over dozens of hours - some of them who ARE direct competitors in the AI NTS space - in learning anything from the basics to controlnets. I certainly feel no guilt for not sharing 100% of everything I learn with every single person.
For those who are polite and make work I have an interest in then sure, I'll volunteer some of my time. But I'm certainly not under any mandate to give up every last piece of information I have.
In fact I'll even give a small bit of insight into how I've been doing things lately, let's show my hopes for how the night ends with Ari. I create an image first using illustrious, which has superior posing and hands and... well everything. And then I use controlnets to move that to pony.
(Also, I'm aware Eric's a bit larger than normal here but I live by the adage that a dick is as big as it needs to be for the scene to be hot... That and I was too lazy to fix it , which I could have done with controlnets)
I have helped at this point literally dozens of people over dozens of hours - some of them who ARE direct competitors in the AI NTS space - in learning anything from the basics to controlnets. I certainly feel no guilt for not sharing 100% of everything I learn with every single person.
For those who are polite and make work I have an interest in then sure, I'll volunteer some of my time. But I'm certainly not under any mandate to give up every last piece of information I have.
In fact I'll even give a small bit of insight into how I've been doing things lately, let's show my hopes for how the night ends with Ari. I create an image first using illustrious, which has superior posing and hands and... well everything. And then I use controlnets to move that to pony.
(Also, I'm aware Eric's a bit larger than normal here but I live by the adage that a dick is as big as it needs to be for the scene to be hot... That and I was too lazy to fix it , which I could have done with controlnets)
- Jul 3, 2021
- 1,990
- 12,418
Totally reasonable and much appreciated input. I personally don't use Illustrious, but mostly do image searches on google and other sources for a base if I wanna use ControlNet with my image generations. Either running them straight through ControlNet using the lineart_anime_denoise preprocessor in combination with Xinsir's Union model on a rather low weight lately (finally figured out a nicely working setup), or as silly as it may sound- pasting the image into paint and editing it by hand, adding snippets of other images and combine them into one big abomination before running it through ControlNet, img2img and whatnot. I'm terrible with photoshop, so paint is really all I need. lol
What I really don't like is when people come here saying stuff like this;
That is so demotivating to me and really disrespectful to any of the great fanartists providing content here and in regards to the time they may have spent doing their research, getting better at AI image generation in general. Arrogant? No passion? No interest? Telling people to stop posting???? Like fuck, how entitled can a person be?
I was fortunate enough to receive a DM from a user who isn't active in this thread anymore, pointing me in the right direction last year after the release of ~Ch02. But the time I had already spend learning and researching, figuring out and finetuning loRA's&Co slowly getting closer to the game's artstyle...and then people come here shitting on others? Fuck no. It's like you said, I'd gladly help people out if I can. But serving everything on a silver platter when someone acts like the above? Again, fuck no. I should probably ignore it and just do what this thread is all about, but man reading some of the stuff is annoying as hell.
Anyway...I'm currently working on a little idea I've had recently. Might post it soonish. Here's just a random image for now!
- Mar 17, 2019
- 336
- 3,244
I use the same process, but with photoshop and layering. Illustrious serves for all simpler images, it's quite advanced compared to pony.
- Dec 4, 2022
- 162
- 168
I know that Automatic1111 saves its prompts & settings in the PNG's Text structure per the PNG spec. It's super helpful not just for the settings, but keywords and LoRA names with strength. I can literal do an IMG2IMG and click an Icon to load everything perfectly into A1111, then re-prompt as needed.
Question: Does Comfy UI and others front ends have that same kind of feature?
And like my sig's PSA mentions, F95 converts images to an AVIF format for saving. That 2MB PNG will reduce to 200KB -- and lose all that valuable PNG data. Please consider posting an archive or alt location for the original PNGs.
Question: Does Comfy UI and others front ends have that same kind of feature?
And like my sig's PSA mentions, F95 converts images to an AVIF format for saving. That 2MB PNG will reduce to 200KB -- and lose all that valuable PNG data. Please consider posting an archive or alt location for the original PNGs.
- Mar 17, 2019
- 336
- 3,244
- Feb 8, 2018
- 26
- 60
- Jan 24, 2025
- 14
- 39
- Mar 17, 2019
- 336
- 3,244
Wow, I'm incredibly surprised by the smoothness of the hair flip and the hand quality. I'll have to check out runway at some point...