Neat! In my opinion those generated by generative AI with controlnet are even more impressive [0], albeit also reducing their likelihood of being correctly read by all.
Show HN: qrframe – generate beautiful qr codes with javascript code
41–50 of 81 posts
Re: Show HN: qrframe – generate beautiful qr codes with javascript code
#42Earlier quoted context omitted.
OP is being respectful and merely points out an issue: the codes generated don’t work with some devices. If you want a code that works with everything, you can’t do too much customization. It’s a very okay thing to point out and there isn’t much negativity in it. I agree that we should support projects like this, but being dismissive of real issues is not the way.
No; top comment: "don't use it" because "I tested on " and - you don't see the problem? It is OK to point out, but not to dissuade others from using, you see? Especially on a Show HN for a cool project where some good person has clearly poured their passion into it. You got to stand up for that stuff. Bring the positivity, man! Frame real issues in the correct (qualified and personal) light, not proclaiming a blanket…
The problem here is that it cannot possibly work robustly. It does what it does by trampling all over the error correction codes and using out of spec pixel shapes.
This is not the sort of problem that will go away once the technology matures. It is fragile by its very nature.
It is absolutely OK to recommend against it on that basis.
Re: Show HN: qrframe – generate beautiful qr codes with javascript code
#43Earlier quoted context omitted.
[flagged]
If you want to use something for production, you shouldn’t only test on the latest stuff you personally have at home but on the crappiest stuff you reasonably expect users to have. Otherwise, you get websites that lag and spin up the fans on everybody’s laptops while the developer is happy on his workstation with 32 cores and a gaming GPU.
No, thanks. I have no obligation to do that and if it lags you can always just close the page.
Re: Show HN: qrframe – generate beautiful qr codes with javascript code
#44This looks super cool but I could only get 10/21 to scan with the classic zxing scanner (>100M downloads), and 3 of those took a bunch of playing with camera angles and distance, so I'd recommend against using anything too fancy for real world barcodes.
Hmm, yes, I did not try these but I have noticed that my default phone app in my S22 is freakishly fast at detecting QR codes, something I never remember that zxing was.
Does anyone know if these are today implemented in neural networks or something? Specialized hardware? Or just very well optimized. Sometimes it feels like only 1 frame in the camera is enough, and distance, angle, lighting and focus is not important.
Re: Show HN: qrframe – generate beautiful qr codes with javascript code
#45This is cool
Re: Show HN: qrframe – generate beautiful qr codes with javascript code
#46Earlier quoted context omitted.
If you want to use something for production, you shouldn’t only test on the latest stuff you personally have at home but on the crappiest stuff you reasonably expect users to have. Otherwise, you get websites that lag and spin up the fans on everybody’s laptops while the developer is happy on his workstation with 32 cores and a gaming GPU.
No, thanks. I have no obligation to do that and if it lags you can always just close the page.
I didn’t mean „you’re mandated by law to do this“, I meant „if you don’t do this, people won’t use your stuff if they can avoid it“. If that’s ok for you, that’s awesome. We were talking about production use though, were adoption and user friendliness is normally a goal.
Re: Show HN: qrframe – generate beautiful qr codes with javascript code
#47This looks super cool but I could only get 10/21 to scan with the classic zxing scanner (>100M downloads), and 3 of those took a bunch of playing with camera angles and distance, so I'd recommend against using anything too fancy for real world barcodes.
Hmm, yes, I did not try these but I have noticed that my default phone app in my S22 is freakishly fast at detecting QR codes, something I never remember that zxing was. Does anyone know if these are today implemented in neural networks or something? Specialized hardware? Or just very well optimized. Sometimes it feels like only 1 frame in the camera is enough, and distance, angle, lighting and focus is not important…
I'm pretty sure the decoding is done on each frame individually, so as long as you get a single good frame, you scan it
Re: Show HN: qrframe – generate beautiful qr codes with javascript code
#48This looks super cool but I could only get 10/21 to scan with the classic zxing scanner (>100M downloads), and 3 of those took a bunch of playing with camera angles and distance, so I'd recommend against using anything too fancy for real world barcodes.
Thanks for trying it out! I definitely agree, the fancier it gets the harder it is to scan so these should not be used for real world stuff where reliability is important. I basically only tested these on my own phone, a pixel 7 with default camera app, so I'm not surprised people with different hardware/software are getting varied results. I think I might try to automate testing using different scanners if I can fig…
Pixel 8, about half of the codes work.
Re: Show HN: qrframe – generate beautiful qr codes with javascript code
#49Neat! In my opinion those generated by generative AI with controlnet are even more impressive [0], albeit also reducing their likelihood of being correctly read by all. [0] https://stable-diffusion-art.com/qr-code/
While I do agree, one can be done almost instantly in your browser, one takes hours of rendering.
Re: Show HN: qrframe – generate beautiful qr codes with javascript code
#50Earlier quoted context omitted.
I tested it with my phone and got similar results: I got only 9 out of the 21 QR codes to work. I don't know why you think such a finding should not be mentioned here.
I think it should be mentioned but without the "so I'd recommend against using", because that's wrong. Also, specify OS, surface, time, etc for better test!
> "so I'd recommend against using", because that's wrong.
Someone's opinion is not "wrong". It may be misinformed, it may be different than yours, but you can't just assign a "correctness" value to something subjective just because you disagree with it.