This LoRA got a training even more rough than the other ones, manipulating settings I didn't even know that existed. I've been doing multiple testings and I didn't encounter any issue.
This style was trained using images of a friend of mine (Skati) so I'll give her proper credits. That's why it's called "SkatiStyle".
You can remix the images, but DON'T use my characters without giving credits to me.
𝐒𝐮𝐠𝐠𝐞𝐬𝐭𝐢𝐨𝐧𝐬 (𝐢𝐦𝐩𝐨𝐫𝐭𝐚𝐧𝐭):
Recommended model: WAI-Illustrious v17.
Use DPM++ 2M Karras for more detailed/modern details, use Euler a (with Normal Scheduler) for more sketchy/rough shading and lines.
I'll be sharing two versions of some pictures, using Karras and Euler a.
The ones on the left were made using Karras and the ones on the right used Euler.
The recommended strength is 0.9.
Recommended CFG: 4.0
Recommended steps: 25-30 for most of images, and 40+ for wide/wallpaper formats.
AVOID using generic prompts as "ultra detailed" or "high resolution" since they will "fix" the rough lines and shading.
If you notice the pictures are coming out monochromatic or too grayish, add "vivir colors" to the prompt and add "monochrome" or "monochromatic" to the negative prompts.
