This LoRA is a conglomeration of multiple targets, 26 categories and styles, all with the intent to produce a modern art style appeal with the associated themes, some provided below...

1960s fashion, alien, condom, ejaculate, sex toys, glass, detailed hands, latex, oral sexual innuendo, erotic sculptures, dessert/cake/frosting, silicone, skin contrasts and situational peculiarities.

This LoRA, no matter the slight weight, is unlikely to produce a photo real experience but the typical "realistic" look, one expects from things like Illustrious, is definitely achievable since every one of the source images were produced using a refined model of that type.

It will produce SFW and NSFW content, most of the training data used was SFW, allowing for some innuendo and those images that were NSFW were chosen for their artistic value.

The Chroma version was trained with 8100 steps and included the following resolutions: 448, 512, 576, 640, 704, 768, 832, 896, 960, 1024, 1088, 1152, 1216, 1280, 1344

The Z-Image version was trained with a target of 1024 for 8,000 steps.

None of the example images were post processed or scaled.

Krea 2 version

I refined the dataset for this model target, removing some images, combined a couple of categories and added some images which may have the effect of leaning generations more towards NSFW.

Erartica is meant to augment your images that you generate. There's nothing real about the output, it's trained entirely on synthetic data and is intended to produce vibrant color separation, 3d visuals, graffiti type art (among other things) and has an overall effect of air brush quality imagery.

Many of the actual training images can be closely replicated, which I've verified, which makes this quite a versatile model since it was trained on 358 images in 21 categories of very different content.

It is not meant to produce "clean" SFW images, almost every data source had some allusion or direct relation to NSFW.