SDXL

XAVIER

#Furry
XAVIER - AI Model cover image
Завантажено
5 січ. 2026 р., 18:59
Використань
204k
Відгуки
Відмінно  (17)
Кроки вибірки

30

Метод семплінгу

DPM++ 2M SDE Karras

Масштаб CFG

3

Негативні

lowres, worst quality, low quality, large head, extra digits, bad eyes, monochrome, bad hands, bad feet, fewer digits, extra digits, deformed, mutant, signature, artist name, ugly face, mutated hands, blurry face, blurry, jpeg_artifacts, gif_artifacts, grainy, wip, uncolored, censored, scan

Дозволи
  • Allow image generation & sharing
  • Дозволити користувачам завантажувати вашу модель
  • Комерційне використання

Recommended LoRAs

Опис

XAVIER is a Checkpoint Merge of three excellent Checkpoints:Hyphoria (v0.01)HomoSimile XL (v4.0 [NAI-XL EPS-v1.1])Mol_Keun Mix (DeepBlue V2 EPS)Having used both Hyphoria and HomoSimile extensively, I sought to produce something with a different base style that maintained the general excellency of those two models. I settled on Mol_Keun Mix to help me achieve it, and I believe I've accomplished my goal.This checkpoint should be considered experimental, because it is. As such, feedback is highly welcome and appreciated.Lastly if you do generate with this checkpoint, all I ask in return is that you consider posting what you make with it (remember to add it to your list of resources if uploading to the site).You can see a showcase of Gen 1 Pokemon made with this checkpoint here: https://civitai.com/articles/16208Recommended Settings (coloring outside the lines encouraged):CFG: 3.0Steps: 30Sampler: DPM++ 2M SDE, GPU SeededScheduler: KarrasClip Skip: 2Suggested Upscaler:RealESRGAN_x4Plus Anime 6BStrength: 1.5Suggested Tweaks (for local gens):Perturbed-Attention Guidance Scale (PAG): 2.5Set alongside CFG: 3.0This will add just a hair of processing time per image; but it seems to further improve prompt adherence (especially when using more obscure tags), and naturally refines the base style a bit.May negatively influence loras.Checkpoint currently unavailable for onsite generation? Try it on Tensor: ********

Коментарі1

Створено XAVIER