Buy me a coffee: https://ko-fi.com/futureflix
This model was created intended to generate 3D line work.
Example of prompt: "monochrome, rear view, 1girl, long hair, nude, silent comic, LineWork"
Strenght: 0.6 to 1
The model worked best for this LoRA was SDXL 1.0
Description
FAQ
Comments (5)
cool let us know when it can do continuous lines as in "Franco Grignani lineart". it is truly impossible. I can't even do circles. :D
Pretty interesting idea!
@futureflix yeah getting that crispy grignani line is,, well its 100% impossible for me, no matter what I try. The lines always slide out of control in someway. Maybe there is some secret prompt or some secret way with controlnets, but nobody has managed to do it as far as I know. Maybe its a flaw in stable diffusion itself..
@yofoton174609 Well... maybe SD is not seeing the way we use to see, specially about abstract lines. We can recognize patterns that SD still doesn't. I believe also it could be a matter of approach. I'm curious to investigate it.
@futureflix yeah ive noticed some really strange behavior when it comes to lets call it "hard lineart" aka (Franco grignani and similar retro lineart). But I think franco grignani hits the perfect nerve to demonstrate the problem.
If you try to make the image it will always have some glitching "splitting hairs". But if you upscale it once 2x, some of the glitches seems to go away a little. And if you just zoom out the image so it seems smaller, the glitches seem to go away by sheer illusion.
It seems to be able to "fix" its own errors with some forms of operations even after the image has been rendered. It's almost as if its trying to follow the thread of the line, and then its "attentionspan" runs out. Its really strange. oh well, Let me know if you manage to fix the "franco grinani" bug :P . Its driving me nuts.
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.