Version 3.0 of the Realism Engine
In this version we improved the skin realism, the eyes. We also improved the Male anatomy.
Keep in mind that this is not a NSFW model. For that I recommend SDXXXL.
Introducing Version 2.0 of the Realism Engine
We are excited to announce that Version 2.0 of the Realism Engine is now available, featuring significant enhancements.
This updated model exhibits superior prompt responsiveness and offers markedly improved overall coherence, including more facial expressions, a greater variety of faces, more poses, and improved hands.
Additionally, we have addressed several quirks from the previous version, such as the challenges in generating nighttime scenes and the issue where increasing the CFG Scale enhanced color vividness without proportionally adjusting contrast. While being capable of producing NSFW content it is less likely to produce it without asking.
This version include the VAE to help newcomers with their setup.
Happy Prompting
Good news for the fans of the original Realism Engine
Realism Engine SDXL is here
Version 1.0 is officially out.
Here's the recommended setting for Auto1111
Sampler: DPM++ 2S a,
CFG scale range: 5-9,
Hires sampler: DPM++ SDE Karras,
Hires upscaler: ESRGAN_4x,
Refiner switch at: 0.9 (ON THE FIRST PASS ONLY)

Look for this in Setting -> Stable Diffusion
You can also add hires_fix_refiner_pass to you quick list setting to have it on top of you Ui
While this model hit some of the key goals I was reaching for, it will continue to be trained to fix the weaknesses.
I appreciate all the good feedback from the community.
Description
Better eyes, even better realistic look
FAQ
Comments (16)
welcome back and thanks for all the hard work!
Is it possible to drop an inpainting version of this?
13Gb?! Is this officially the largest model on Civitai? :(
Can the full model be used as a base for training a dreambooth model?
is this just a merge or did you train onto SDXL ?
Nice model, thank you.
Tried hires. fix 3x, tried upscale from img2img. Pictures are still blurry. But right before completion , at around 97% the image is pretty clear tho
Looks very nice but it's too big, doesn't even load with 12GB video card :(
Very nice! I am new to this, but does anybody know how to use this from HuggingFace? I do not use a webUI but pure python calls and HF is my go to in that case.
I've no idea what the "No refiner needed" crowd has done, but is that possible for your model as well?
How big was your imageset?
is the VAE baked in? and do we use sdxl_refiner_1.0.0.9vae for the refiner?
I recently attempted to replicate one of your tasks by using the same prompts and parameters you provided for generating a specific photo. However, I found that the skin details in the images I produced are not as realistic as in your examples. Could you possibly assist me in understanding why this might be happening? Is there a particular aspect or setting that I might be missing or need to adjust to achieve similar high-quality results?
please make an inpainting version
where 2.1
I get the error message
URLError: <urlopen error [SSL: WRONG_VERSION_NUMBER] wrong version number (_ssl.c:1007)>
What can i do to fix this?
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.