<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/18d64277-a62f-4704-805d-54bff41ff582/original=true/18d64277-a62f-4704-805d-54bff41ff582.jpeg" />Join The Tinkerer on Whop to get early releases, private pages, and the Tinkerer Discord role - all in one place. 👉 <a target="_blank" rel="ugc" href="https://whop.com/joined/cyberdelia-ai-lab/products/the-tinkerer/">Join on Whop</a><hr />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.<hr />⚙️ Recommended Settings<pre><code>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</code></pre>⚙️ Recommended Settings DMD2<pre><code>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!</code></pre><hr />💡 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.<hr />🧠 PromptingExample negative prompt:<pre><code>cartoon, illustration, anime, painting, CGI, 3D render, low quality, watermark, logo, label</code></pre>or use <a target="_blank" rel="ugc" href="https://civarchive.com/models/1531979/cyberrealistic-negative-sdxl">CyberRealistic Negative</a><hr />✨ Optional Adetailer<pre><code>Adetailer model: face_yolov9c.pt
If you only want the main face being refined set 'Mask only the top k largest' to 1.</code></pre><hr />🔗 Links Backup location: <a target="_blank" rel="ugc" href="https://huggingface.co/cyberdelia/CyberRealisticXL">HuggingFace</a> <hr />☕ Fuel the Project If this model gave you stellar results, <a target="_blank" rel="ugc" href="https://ko-fi.com/cyberdelia">consider buying me a coffee</a>. Or two. Or, you know… a GPU.<hr />⚠️ 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).<hr />💬 Want Better Prompts? Use this custom ChatGPT to auto-generate polished prompts for CyberRealistic XL: 🔗 [<a target="_blank" rel="ugc" href="https://chatgpt.com/g/g-6834133e3ab881918a91b3ec6b9eb01f-cyberrealistic-prompt-helper">Try it now on ChatGPT</a>]

CyberRealistic XL

This is version 2.4 (internal) with some additional training from version 3.0. There were some issues with version 3.0.

cyberrealisticXL_v31.safetensors

v3.1

v10.0

v9.0

v8.0-Inpainting

v8.0

v7.0 DMD2

v7.0

v6.0

v5.8

v5.7

v5.6

v5.3-Inpainting

v5.3

v3.0

v2.2

v2.1

v2.0

v1.1 (VAE)

v1.0 Play (VAE)

Want to buy me coffee? CyberRealistic XLIt took a while, but here is the SDXL version of CyberRealistic. The criteria for this model are simplicity in usage and versatility. So, no complicated prompts (you can if you want it); simply unleash your imagination. Enjoy and have fun playing!Rest assured, both CyberRealistic and CyberRealistic Classic (based on 1.5) will continue to receive updates and support as per usual.NEW:This is version 2.4 (internal) with some additional training from version 3.0. There were some issues with version 3.0.

Files with SHA256: 95d17e744fba748e6d0c55126e60c794fc9c26e563dbf9c98feead78c3731e78