It makes everything like vector art.
Or just more like a cool poster.
Created as described in the tutorial of konyconi but with some own tweakings: https://civarchive.com/models/52697/tutorial-konyconi-style-lora
If you like my LoRA i would really appreciate a review with :
⭐⭐⭐⭐⭐
Have fun!
If you want, support me here:
https://displate.com/dervlex?art=5d91f0dbbcb29
Description
FAQ
Comments (12)
580 Downloads and only one review :D
Guys xD
That´s not how it should work :D
I need feedback! :D
Pretty normal, people don't know us, we're just "starting" in CivitAI, etc. I will do some tries with the LORA when I'm at home but overall I think it's pretty good work. Proud!
Very nice style and Lora :) But i struggle to make it work out with long prompts for some reason, like with some prompts describing "complex scenes" it just looks like the lora wasn't active, still works perfect for simple scenes when applied to a single object/character :)
Now i also got question about your training, what were the tweaks you made from the kyconi tuto ? Im struggling to train good style lora, i think its captionning problem, so can you give an exemple of a caption you would have written for the joker image you posted ? (if it was one of the training images of the dataset)
Hello. I'm not Dervlex but I think I may help you? Is the prompt > 75 words or > 75 words, that's a thing that seems to affect most of LORA's in Automatic1111. Furthermore if its < 75 and something off it's going probably the problem are the syntax of the prompt in terms styles conflicting or some words fighting others, etc. Can you put here an example prompt so we can test? Cheers.
You could also try to prompt the trigger word on the beginning "in the style of 0mib, ......."
Because of the training.
I tweak the Dim and Alpha and some other settings.
But it depends on all training data sets, so just trial and error :x
Don t Work
Sorry but Yes. It works.
https://www.youtube.com/watch?v=MEnUv9d1xKE&pp=ygUwU3R5bGl6ZWQgVmVjdG9yIEFydCB1c2luZyBTdGFibGUgRGlmZnVzaW9uIFdlYnVp
Some guy even made a video of the LORA in action and it's clearly working... If not working for you, check file name of the LORA or see the examples to see what's happening. The problem is not on the LORA mate. Can you tell us the the prompt, model, etc. so we can see what's going on? Cheers
As a hint: For upscaling I use 0.7 of LoRA strength.
I give it a try it today and it's not working , and i download the same model rev and the same , when i add the lora it's typing <lora:0mib3(gut auf 1):1> code like this i try with this code and without it the same .
You tried the same prompt ? With the trigger word "in the style of 0mib"?
And you can try to rename it. Rename it to "0mib".
And prompt then as try " <lora:0mib:1> A glass in the style of 0mib"
It is working as you can see in the Reviews.