I DO NOT authorize PixAI & Yodayo (and other users) to transfer this model from Civitai.
PixAI users who wish to use this model can do it by using the uploaded one on my PixAI profile from the link below : https://pixai.art/@vxp/artworks/models
The model has been reworked to produce Anime-style images, but at the same time, by using prompts like "realistic, photorealistic" it is capable of generating realistic images, thereby expanding the range of creativity.
The following settings are suggested:
SAMPLER: DPM++ SDE
SCHEDULER: Karras
STEPS: 15 (12 ~ 20, as you prefer)
CFG: 4 (3.0 ~ 6.0)
Recommended quality prompts:
masterpiece, best quality, amazing quality, absurdres, very aesthetic
Recommended negative prompts:
bad quality, worst quality, worst detail, bad, low quality, bad anatomy, 6 fingers, bad hands, bad feet, bad hand, bad finger, bad fingers, extra finger, extra fingers, split finger, split fingers, extra digits, fused arms, fused hands
Description
some fine update on model, now character have more anime style with detailed background
FAQ
Comments (10)
Keep up the great work 👍 🎉
This new version is a true MASTERPIECE! If you can come up with a better version than this, I'd say you're a GOD! In fact, that's exactly what I was talking about with a friend the other day—the illustrious hasn't reached its peak yet, and I believe there's still so much potential. And you did an incredible job with this new version!
You are a little bit overhyped, there's not a single SDXL that can match SD1.5 backgrounds
Sorry for the noob question: what needs to happen for this checkpoint to be made available through onsite generation?
Unfortunately I have no idea how to make mine one available on Civitai, But on PixAI you can use them online
@VXP yes, I've been using your checkpoint in pixai. Thats why I like it so much!
I prefer the publish experience and community in civitai. Hopefully we will see your models live in the onsite generator soon :)
so im not sure if you’re able to give a good answer but i use the 1.88 version on pixai paired with a hyper SDXL LoRA to reduce credit use. but i want to use regular parameters now. so how could i get similar results? if it helps i use 7 steps with Euler A sample method and cfg scale at 1.1
It depends on which Hyper LoRA you're using — there are versions with 2, 4, 8 steps, and so on.
It also depends on the weight you're applying.
For example, if you're using a high weight, you should reduce the CFG (though this isn't always ideal, as it can weaken the effect of the negative prompt).
On the other hand, if you're using a low weight, like 0.3, you can reduce the steps while keeping the CFG around 3, which still allows negative prompts to be effective.
@VXP https://civitai.com/models/612920/hyper-sdxl-2steps-lora this is the lora i was using. typically at around 0.7 or 0.8 as the weight. and in the previous comment i shared my cfg scale and sample steps