Merge via several models including my BestComix. This merge has more volume in picture and handles style lora's very well
Description
FAQ
Comments (6)
I'm not usually a fan of "Comic" style models. However, I think that your model provides both versatility and prompt adherence. I'm uploading some benchmark images (very comic-style) for others to evaluate this.
waiting for 4 step lightning version
I'm sorry, i don't know for sure what it is. I've seen 4 and 8 step versions of other models, but frankly, I couldn't distinguish the difference (on some models looks absolutely different, on other - same). Besides my model is a merge, so maybe it is easier to use lora for you?
Not sure if the file is broken but this keeps happening at the last step. It becomes all grey and unsaturated with weird white artifacts all over it. I've tried re-downloading but it doesn't fix it.
You probably need to use a VAE (or a different one).
Doesn't look as file problem. Vae might be the case, I use sdxl-vae on all my renders. Also check prompt and correct work of all extensions (if any). I had this problem at the very start working with automatic111 and found out that prompt was recurrent — instead of random generation it mixed all words in one (one step - one keyword). Try very simple prompt and disable all extensions, enabling it one by one.