This is my first try on a merge of a realistic SDXL model. I tried to give it a bit of a Flux like look. It is somewhat capable of producing NSFW pictures, but can also make good looking SFW pictures.
Since this is a SDXL model, only natural language prompting is supported. Booru tags might work to some degree, but the results will be unstable,
Update Version 2.0
I added 2 additional models to the merge and increased the overall quality of detail. Effects like lighting, shadows or fog now look better then in V1 and the skin details are also a bit more detailed.
I also found out about 2 new samplers that produce nice results with this model:
Euler Smea + Simple (20 steps) = it gives the picture a bit studio like effect (e.g. artificial lighting)
DPM++ 2s pyramid + Align Your Steps (16 steps) = it gives the picture a bit more realistic look with saturated colors
In order to use these samplers you need to use SD Forge and install below additional samplers as an extension:
https://github.com/MisterChief95/sd-forge-extra-samplers
Recommended settings:
Sampler/Scheduler: DPM++ 2M SDE with Karras / DPM++ SDE with Karras / DPM++ 2S a with Karras
CFG: 2-5 (you can also try higher values, but did not really test it)
Steps: 15-25
Highres fix depending on use case.
VAE is included
Quality tags:
Positive
masterpiece, 4k, ray tracing, intricate details, highly-detailed, hyper-realistic, 8k RAW Editorial Photo, cinematic light, dramatic light, shallow depth of field, high budget, bokeh, cinemascopeNegative
cartoon, cgi, render, illustration, painting, drawingmerged models
Acorn Is Boning XL3
epicrealismXL_vxviLastfameRealism
pornCraftByStableYogi_v10FP16 (in V2)
gonzalomoXLFluxPony_v40UnityXLDMD (in V2)
+ styling lora
Permission: no commercial use, no selling, use only on Civitai (please also check the sources for their respective permissions)
Description
Version 2.0
I added 2 additional models to the merge and increased the overall quality of detail. Effects like lighting, shadows or fog now look better then in V1 and the skin details are also a bit more detailed.
I also found out about 2 new samplers that produce nice results with this model:
Euler Smea + Simple (20 steps) = it gives the picture a bit studio like effect (e.g. artificial lighting)
DPM++ 2s pyramid + Align Your Steps (16 steps) = it gives the picture a bit more realistic look with saturated colors
FAQ
Comments (5)
Installed from the given github link but do not see the additional sampler 'DPM++ 2s pyramid' nor the scheduler 'Align Your Steps (16 steps)'.
I installed from link via extensions method, did I do anything wrong? Using Forge.
With any sampler, the last step is covered in smudges and glitches. Lcm sampler removes them, but everything is blurred and a halo around the body appears. Ноw to use? I tried all the checkpoints you used in the merge, they work fine. I use Automatic. You have the most normal garter belt stockings.
Please try using Forge with my recommended settings.
A1111 seems to be discontinued and is maybe too old by now.
The prompt adherence and custom character lora likeness is incredible. No idea why this is underrated. Keep up the good work!
Had to comment once again after using it the whole day!!! Best SDXL NSFW model. It listens prompts in a single tag or sentence. The lora likeness is unbelievable. Thanks for the Model again. I'm using the 2.0 version!
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.





