I was stunned by how far the fan art thread has progressed and how similar the images look like. They're basically identical. I wasn't aware how few prompts are needed to create Laura & friends when you use the right LoRa.
To be fair, creating fan art images, that look like in-game images, was never a problem. LeStag generates his stuff on Civitai.com and posts images of the girls there as well, with all the models and prompts visible to everyone. So any one of us there could just remix those images to create our own scenarios. And those who generate locally can create images with greater fidelity and have more control over generation: the banner of the fan art thread is four of the girls together in the same picture — good freaking luck creating something like that on Civitai.
Instead, the problem was always creating something complex, because a) LeStag generates in Civitai, so his tools are limited b) he uses Pony Diffusion as his checkpoint. Civitai doesn't have regional prompting to tell AI "here, this girl lying on her back should be toned, fit, with long pink curly hair and green eyes; while this girl licking her pussy should have black wavy hair in a ponytail, brown eyes and more thick body". It mixes these things up.
The only somewhat reliable way to make 2 specific girls together in the same pic on Civitai, as far as I know, is to take a good irl picture, modify it and then use image2image tool to create a similar image in LeStag's style.
Pony, unfortunately, is more prone to errors and artifacts, like extra fingers, than, say, Illustrious checkpoints. But those all have a very different style compared to the one used in the game. Only after
Vordana trained the models of the girls in Illustrious using LeStag's pics, did we get the ability to make Illustious pics that look like Pony.