
Join The Tinkerer on Whop to get early releases, private pages, and the Tinkerer Discord role - all in one place. 👉 Join on Whop
CyberRealistic GenAether (SDXL) models are Textual Inversion embeddings built to create vivid, fictional AI characters. They're not real people. Each one is a crafted persona with its own look, personality, and backstory.
⚠️ Important Note:
SDXL TIs are not functioning correctly in the Civitai online generator. Please use them locally in A1111, Forge, or ComfyUI for proper results.
This limitation affects all CyberRealistic GenAether SDXL TIs.
Who is Astrid Holgren?
Astrid Holgren is a 26-year-old writer and fitness enthusiast from Gothenburg, Sweden. With a naturally athletic build and a spark of wanderlust in her eyes, she balances her days between early morning runs along forest trails and late-night writing sessions fueled by herbal tea and Nordic jazz. Passionate about both literature and life, Astrid is the author of several self-published novels, blending Scandinavian folklore with modern romance. When she's not working on her next book, she’s either reading on a sunlit balcony or exploring the outdoors with friends—always chasing new stories, whether on the page or in the world around her.
What is a TI?
A Textual Inversion (TI) is a small file (.safetensors) that teaches Stable Diffusion a new token (word).
These CyberRealistic GenAether TIs contain character information (age, gender (woman/girl), and facial/body features) so you can get a consistent look across your scenes.
Installation & Usage
Automatic1111 (A1111) / Forge
Place the
.safetensorsfile into:
stable-diffusion-webui/embeddings/Restart A1111/Forge.
In your prompt, type the TI name (filename without
.safetensors).
sdxl_astridholgrComfyUI
Copy the
.safetensorsfile into:
ComfyUI/embeddings/Add a Textual Inversion Embedding node, connect it to your CLIP Text Encode node.
Use the TI name in your prompt as normal.
Example Prompt
sdxl_astridholgr, walking along a narrow cobblestone street with hands in pockets, thoughtful downward gaze, fitted denim jacket over loose blouse, candid mid-step capture from behind at slight diagonal, cool overcast daylight diffused evenly, old European alleyway with stone walls and hanging lanterns, documentary street photography aesthetic with muted colorsPrompting Tips
Add age keywords to make the character appear younger or older.
Combine with a character LoRA (low weight, e.g. 0.3) to create alternate looks while keeping the same base character identity.
Description
Important Note: SDXL TIs are not functioning correctly in the Civitai online generator. Please use them locally in A1111, Forge, or ComfyUI for proper results.
FAQ
Comments (11)
I'm unable to get anything close to your example pictures using onsite generation. For this embedding and others in this collection. Even if I hit remix and then generate without changing anything. Could you share your local generation workflow please?
I use Forge
https://civitai.com/articles/15255/forge-setup-guide-cyberrealistic-models-windows
I don't really know why it doesn't work. You use a SDXL checkpoint?
@Cyberdelia Thanks, I'll look at that. As for onsite, no literally remixing your images without changing anything and I don't get anything close.
@SuperDragon57Â I have noticed it now also. Added a notice in the description. Need to contact Civitai why this is not working.
@Cyberdelia Not much luck using these locally on comfyui either. I guess I'll try out forge next.
I'm sorry but the results are totally different than the samples
@Hewytul you tried the Citiai generator? For some reason, Textual Inversions aren't working properly there. I tested several of mine, and they all look off. Everything works perfectly in local setups like A1111 or Forge though. Really strange—definitely need to look into it.
Dont make loras.
You mean I shouldn't create TIs (not LoRAs), and you're saying that because the result looks completely different from my examples? I'm noticing this too now — apparently, the Civitai generator has issues with my SDXL TIs. They work perfectly locally in A1111, Forge, etc.
@Cyberdelia Yes, exactly:) But not because it looks different, but in Forge i can save them with strength and prompts. Anyway, barely any people made embeds for XL, its an 1.5 legacy. I dont say its bad(its the opposite, very good), i just lamenting about the usage.
Nice composition and lights



















