Join The Tinkerer on Whop. Membership gets you early releases, private tools and a bunch of extra stuff.
👉 Join on Whop
💬 Join the community for support, free tools and early news on Discord
This is the model you reach for when you want super clean, photo-realistic results.
It’s sharp, powerful, and somehow just nails gorgeous images every single time.
⚙️ Recommended Settings
Sampling method: DPM++ 2M SDE Karras
VAE: is already Baked In
Sampling steps: 30+ Steps
Resolution: 832x1216 / 896x1152
CFG: 3-5
Clip Skip: 1
Upscale: 1.5 upscale (0.3 - 0.35 denoise)
Upscalers: 4x_NickelbackFS_72000_G / 4x_foolhardy_Remacri / 4x_NMKD-Siax_200k⚙️ Recommended Settings DMD2
Sampler: LCM
Schedule Type: Exponential
Sampling Steps: 8/12 (12 for extra detail)
CFG Scale: 1-1.5
⚡ These settings are not the same as the standard SDXL workflow — be sure to adjust your pipeline!💡 What’s the difference between FP16 and FP32 checkpoints?
FP16 (half precision) uses 16-bit floating-point weights. It’s smaller in size, loads faster, and uses less VRAM - perfect for normal image generation. You won’t notice any visual difference for inference, so it’s ideal for everyday use.
FP32 (full precision) uses 32-bit weights, which means it’s larger and heavier on resources. It doesn’t improve generation quality, but it’s better if you’re training, merging, or fine-tuning models, since it keeps the exact weight data without rounding loss.
🧠 Prompting
Example negative prompt:
cartoon, illustration, anime, painting, CGI, 3D render, low quality, watermark, logo, labelor use CyberRealistic Negative
✨ Optional Adetailer
Adetailer model: face_yolov9c.pt
If you only want the main face being refined set 'Mask only the top k largest' to 1.🔗 Links
Backup location: HuggingFace
☕ Fuel the Project
If this model gave you stellar results, consider buying me a coffee. Or two. Or, you know… a GPU.
⚠️ Heads-Up
This model can make mature or sensitive content. You’re fully responsible for what you create, so use it responsibly (or if you can’t resist, at least be artistically irresponsible).
💬 Want Better Prompts?
Use this custom ChatGPT to auto-generate polished prompts for CyberRealistic XL:
🔗 [Try it now on ChatGPT]
Description
Specially developed SDXL inpaint checkpoint for inpainting and outpainting.
In general, you can stick to the same settings you'd use with the main model. I usually go for inpainting with DPM++ 2M Karras and at least 30 steps to get good results. That said, you often crank the steps way higher - sometimes up to 50 - but that’s totally optional and depends on your workflow.
For CFG, somewhere between 2 and 6 tends to work well.
As for denoising strength, I usually start around 0.8. If the model starts getting weird or too abstract, dropping it a bit can help. But sometimes, bumping it closer to 0.9 actually improves the image.
Denoising really changes how the final image turns out. So if you’re not happy with what you’re getting, try tweaking just that one setting before messing with anything else. It’s usually the one that makes the biggest difference.
FAQ
Comments (11)
I love you! Hey using the fp16 8.0v with any clipskip generates black images. Removing CS make it work as it should. I have some lora's that require CS. I don't know what to do. Maybe you have an idea on why it's happening. Comfyui.
if you're using comfyui its clipskip -2, if you're using a1111 make sure you're changing the sdxl clip skip (there's two settings)
This is weird or maybe its me but the regular model seems to get better inpainting results than the inpainting model?
I usually use the regular model for inpainting, but when it comes to adding elements (like clothing), the inpaint model tends to work better.
@Cyberdelia Ahhhh yeah see, I definitely wasn't adding clothing
Hey my dude, I am struggling to find an inpainting workflow that really works well. I am also interested in...uhm..not adding cothing. Would you mind sharing your WF and models/Loras used?
@SaburoDio I use Forge Neo - everything is built in :)
@Cyberdelia and here I was hoping for a christmas miracle haha Thank you anyways!
@SaburoDio fooocus
Merry Christmas!
Please add Fp8/BF16 versions