KNX OVA NSFW IllusXL
Designed by Ktiseos Nyx's T3RR4!
We're working hard at keeping up with what we enjoyed in SD 1.5 and beyond - the more we develop our models the more joy we see! If you're interested in supporting us in our long standing AI journey please scroll down for ways to connect & sponsor us!
Vibrance & VAE Note
There is SDXL Basic vae baked in, however if you prefer VIBRANCE, we can re-release a copy with LunaXL or Neptunia baked in. Please note, this is a NSFW model, so results may vary and instability and stability are not guaranteed.
✨ Recommended Settings
To get the best out of ANY of our models, try these settings:
Samplers: Euler or Euler Ancestral
Scheduler: Karras
Negative Prompt: Generally not needed! The model is designed to perform well without one, but feel free to experiment based on your preferences.
HiRes Fix:
Steps: 15
Denoise: 0.3
Upscale: 2x
Upscalers: Rex, 4x_Remacri, or your personal favorite.
ADetailer: Optional. Due to the strong aesthetic and baked-in VAE, it might occasionally produce unexpected artifacts like grey squares. Use only if necessary to correct faces or intricate details.
Embeddings: While no stylistic embeddings are strictly required to achieve the core look, we highly recommend using Ktiseos Nyx's custom embeddings for optimal control over specific character types, angles, and poses shown in our examples.
LoRA Compatibility: Chances are good that most LoRAs will work with this. However, be aware that while they will function, the result might not perfectly align with your expectations if the LoRA heavily influences style. It's a "What You See Is What You Get" situation with the base model's strong aesthetic – it'll likely produce a good image, just perhaps not the one you had in mind with a very specific LoRA.
🛠️ Merge & Training Details
The backbone of any of our mixes are built upon numerous LoRAs, many of which were trained in-house by Ktiseos Nyx. We're always experimenting and creating! Due to the wonderfully rapid pace of Ktiseos Nyx's LoRA development, we confess that sometimes the exact lineage of merged checkpoints becomes a delightful mystery. We believe in training as much as we can ourselves, but the spirit of community and shared creation is vital!
Trainer Tool Announcement:
We are excited to announce our upcoming Ktiseos Nyx Trainer tool! We aim to make training accessible and efficient for everyone. Check out the repository for updates and future releases:
https://github.com/Ktiseos-Nyx/Ktiseos-Nyx-Trainer/
🙏 Acknowledgements
A huge thank you to the incredible community members whose work and inspiration contribute so much. Special mentions to:
Arthemy, Wai, Holostrawberry, FallenIncursio, RichyRich, JustTNP, Hoseki, CitronLegacy, Samael1976, Kairen92, ShadowxArt, Flyx3, YeiYeiart, LazyVisionary, and so many more! We try to train as much as we can ourselves, but the community's creativity is a constant source of inspiration.
🌟 About Ktiseos Nyx & Connect With Us!
Ktiseos Nyx is available for commissions! Find us and more of our creations here:
Discord: https://discord.gg/HhBSvM9gBY
Arc En Ciel: https://arcenciel.io/users/77
X/Twitter: https://x.com/KtiseosNyx_AI
Sponsor Models & Commissions: https://ko-fi.com/duskfallcrew/commissions
Model Backups: https://huggingface.co/EarthnDusk/
TensorArt Space: https://tensor.art/u/611011406535381539
✨ Licence and Details
Outputs cannot be sold without further editing beyond color grading and basic resizing.
Commissions must go through KNX's main tech nerd - Duskfallcrew/
Monetary gain from outputs should only be used if further re-drawing has occurred.
I am not and nor is Civitai, nor is Ktiseos Nyx as a whole responsible for any outputs that go against anything.
Do not re-sell or commercialize ANY outputs without first contacting Ktiseos nyx
Description
Changing to KNX OVA NSFW so Terra and i dont have to clog model cards D:
FAQ
Comments (6)
⭐ Excellent model! ⭐ It feels like using SD1.5 again, which I really enjoy. Thank you for creating this checkpoint 🤗
Thank you <3 Kind words are amazing I just wish i wasnt' facing extreme burnout pushing models out xD -- been doing this for three years and i wish i was gonna say i was going to be fine for the 4th year ---
But uh ._. i'm strugglin to wanna download more of my models and even publish them lol
@ktiseos_nyx Take a break if you need it, and I'm sure your inspiration to creating / publishing new models will return. Take care 🤗
I like the style of your showcased pics but i m a bit concerned after looking at your prompts. 🤣
Admittedly, tags placed first receive more attention, but that said, there is also a recommended method for proceeding with regard to the order of tags in a prompt.
Also, and most importantly, you seem to use some non-Danbooru tags. If your checkpoint was a trained one i could understand and guess that thoses have been trained during the checkpoint creation. But your checkpoint isn't a trained one, this is a merge, so it doesn't respect that logic. And natural language with any ILXL checkpoint is like like flipping a coin and praying it lands on the right side: not accurate at all.
So would you care to explain from where does those non Danbooru tags come from and share a list of those that you re 100% sure that they have been trained among checkpoints that you use for your merge?
Because if they haven't been trained then they are just ignored and only impact the resulting image because of the change in the prompt tokens weight.
I fly off the handle and do whatever works XD I should HOWEVER BE mindful in testing and prompting because you are ENTIRELY RIGHT: Danbooru = STABILITY "BE NICE TO ME ILLUSTRIOUS YOU SUCK" = It'll give me the sucky sucky poses i didn't want.
This is why i'm still not entirely happy with a lot of the stuff i've been making and i'm facin extreme gen-ai burnout ;_;
@ktiseos_nyx Makes sense. Anyway this is a really nice checkpoint so, once again, thanks for your work on it. 👍