I won’t waste too much time on technical details.
I started by training a photographic dataset using NLTYI Uncensored realism as the base to shift the focus back toward photorealism (NLTYI Uncensored realism was leaning more into an artistic direction).
Then I re-captioned my set with Booru tags and trained it using SDXL as the base.
I merged these two sets (at a 0.15 ratio), but ultimately I wasn’t satisfied.

So I ran the same Set again, this time using Lustify by Coyotte as the base model.
That version was then added to the earlier one (at a 0.1 ratio) as a stabilizer for the output.
Since the Dataset was captioned using Booru tags, both Booru tagging and the captioning style from versions X–XII work extremely well with Ragnarok.

NLTYI Uncensored realism has improved in many areas: photorealism, digital painting, poses, hands, feet, and much more.
That said, it’s still an SDXL model, and I don’t recommend comparing it to models like Flux, Reve, or Sora. For example, it still has limitations when it comes to text rendering or faces at a distance.

I recommend using it as part of a pipeline for your projects. Example setup:

And now, here are the recommended settings:

Recommended Settings(VAE is baked in):

Res: 832*1216 (For Portrait, but any SDXL Res will work fine)

Sampler: DPM++ 2M SDE

Steps: 30-40

CFG: 3-6 (less is a bit more realistic)

Negative: Start with no negative, and add afterwards the Stuff you don´t wanna see in that image.

VAE is already Baked In

HiRes: 4xNMKD-Siax_200k with 15 Steps and 0.3 Denoise + 1.5 Upscale