Tell me if this works... I had to back the camera pretty far back.
Discussion
Didn’t work for me
WOS is not picking it up
Its a URL not a lightning invoice, lol. Should've said that. Use your normal phone camera.
Doesn't work for me
The Dark Side of AI?
(I hate url shorteners.)
That's what I saw. I was just listening to it too, but I need to take notes.
Trying to figure out the best way to do that because the more information there is the harder it is to get one that scans. Works far better with less info so URL shorteners held alot.
Trying to figure out a good way around it though because it becomes dependent on a redirect which I don't like and feels stupid when it's not necessary. (that said I generally like URL shorteners, 😆)
Worked for me.
How do you generate this qr codes?
This was generated using Autonmatic1111 interface on top of stable Diffusion running locally on my Macbook Pro M1
Took tons of tweaking to get it to be readable, but the combination of using the original QR code in both the img2img and ControlNet inputs you can produce a working result with a little work.
I had to darken the midtones in Kirta afterward to get it to scan better, also works best if you put your phone further back.
The QR code was generated here with 30% error correction:
I was looking at this project last night. Definitely some cool images being used as QR's... IF they work, lol.
I'm surprised to not find more such QR-codes.
?w=768&ssl=1
is pretty slick, scans perfectly here and only gives away that it is indeed a QR-code via the calibration squares. Yours still looks blockier but smoother than theirs is definitely possible, too.
I specifically made mine around tiny stone blocks in the prompt to try to start from something that scans well and then work backward. Just trying out different weights and ranges with ControlNet and different models.
No code detected
Looks cool but No such luck.
