Hypernetwork for generating images stylized as waven chibi characters. Actually produces good results even in different proportions of strength.
Trained with danbooru tags.
If you like my work and want to buy me a coffee :')
(updated link to ko-fi)
There two versions "1.0a" and "1.0b", they are parallel results of same dataset. Both will produce good looking images and it's final version of this hypernetwork.
A little tip, use 'turn pale' in negative and clip skip 2.
Description
Trained with clip skip 2.
A little unstable but can get really good results
FAQ
Comments (16)
can you please share the settings you used for your training, and just generally how it works, your results are amazing and when i tried to train a LORA i failed miserably lol
Hey, what kind of settings? Asking because i don't think that hypernetwork have the same process as LoRA :)
my bad, for some reason i thought this was a lora, i still wouldn't mind the settings for the hypernetwork training
@ggthewg
There a lot of good information, i'm using it as reference
https://github.com/AUTOMATIC1111/stable-diffusion-webui/discussions/2670?sort=new
is a safetensor version possible?
can anyone explain me in detail how to use this?
for the AUTOMATIC interface, you have to go to Settings, scroll down to 'Hypernetwork', select the file and apply the change
Hello please how do i use the hypernetwork? i enabled it from the settings but it seems not to work, do u use sd1.5 or 2.1?
please how to use this file ? do you have a tutorial and where to put it? i know lora and simple models (checkpoints models)
Put it in hypernetworks folder and activate it like any lora, hypernetwork folder might be close to base models folder or lora folder in ur setup
(for last version of ai at last)
多撒范德萨发生
这个不错，适合生成Q版头像
Your work is very good.maybe，alipay is also a good choice to get your coffee.paypal is not used that wide in some places.
I am sry but i have followed every steps according to the comments and tutorial on other website, the result is just totally different style. Is there any thing i missed? I am using the webui colab version.
ok i figured out i have to use anythingv3 model (for those who struggled, see here)
love it！🥰