This is IMO truly fascinating. Sure, plenty of the images are clearly "part QR code, part image". But others are... unless I make them really small, I would have have 0 clue it's a QR code. For...
This is IMO truly fascinating. Sure, plenty of the images are clearly "part QR code, part image". But others are... unless I make them really small, I would have have 0 clue it's a QR code.
For example, my Fairphone 4 had 0 qualms scanning this image. And honestly at full size, that's definitely not a QR code if you were to ask me.
Love this. Might need to print myself a nice WiFi QR Code for my home!
I can see it being a double-edged sword depending on the use case. For commercial and social uses you'd really want the audience to recognize the shape of a QR code for them to know to scan it, so...
Sure, plenty of the images are clearly "part QR code, part image". But others are... unless I make them really small, I would have have 0 clue it's a QR code.
I can see it being a double-edged sword depending on the use case.
For commercial and social uses you'd really want the audience to recognize the shape of a QR code for them to know to scan it, so having distinguishing features (the three alignment dots, checkerboard-ish patterns) is probably more of a positive.
On the other hand, for ARGs or CTFs or just sneaky easter eggs, the less obvious ones are a great fit. Like you'd never think to look for a QR code in this Lego castle, for example. I know it's there and I still don't see it at full size.
I stumbled upon this on a recent episode of Two Minute Papers. It seems to be using Stable Diffusion's image-guided prompting to create vibrant images that still function as scannable QR codes....
I stumbled upon this on a recent episode of Two Minute Papers. It seems to be using Stable Diffusion's image-guided prompting to create vibrant images that still function as scannable QR codes. The documentation is in Chinese, so if anyone could tell more details about it, that would be much appreciated.
You can try it with your phone, but some scanners work better than others: my Samsung's built in one really struggled to pick them up, while Google Lens worked on most of them. I suspect the showcase has been validated against WeChat's scanner.
There has been a trend of sacrificing some of the QR code area to fit an image into it thanks to the error correction built into the encoding, and this project seems like a possible evolution of that. Do you think it will catch on? Could this be the one of the first widespread applications of generative art?
They work better when they're a bit smaller. Downloading the image and scanning them did not work for me. But through my camera from a distance they all got picked up very easily.
They work better when they're a bit smaller. Downloading the image and scanning them did not work for me. But through my camera from a distance they all got picked up very easily.
Something about the combination of my laptop and my phone makes Samsung's scanner very inconsistent. Maybe my screen isn't bright/constrasty enough for it to easily pick up the patterns. I know...
Something about the combination of my laptop and my phone makes Samsung's scanner very inconsistent. Maybe my screen isn't bright/constrasty enough for it to easily pick up the patterns. I know it's not a hardware problem because Google Lens handles it fine.
This is IMO truly fascinating. Sure, plenty of the images are clearly "part QR code, part image". But others are... unless I make them really small, I would have have 0 clue it's a QR code.
For example, my Fairphone 4 had 0 qualms scanning this image. And honestly at full size, that's definitely not a QR code if you were to ask me.
Love this. Might need to print myself a nice WiFi QR Code for my home!
I can see it being a double-edged sword depending on the use case.
For commercial and social uses you'd really want the audience to recognize the shape of a QR code for them to know to scan it, so having distinguishing features (the three alignment dots, checkerboard-ish patterns) is probably more of a positive.
On the other hand, for ARGs or CTFs or just sneaky easter eggs, the less obvious ones are a great fit. Like you'd never think to look for a QR code in this Lego castle, for example. I know it's there and I still don't see it at full size.
I stumbled upon this on a recent episode of Two Minute Papers. It seems to be using Stable Diffusion's image-guided prompting to create vibrant images that still function as scannable QR codes. The documentation is in Chinese, so if anyone could tell more details about it, that would be much appreciated.
You can try it with your phone, but some scanners work better than others: my Samsung's built in one really struggled to pick them up, while Google Lens worked on most of them. I suspect the showcase has been validated against WeChat's scanner.
There has been a trend of sacrificing some of the QR code area to fit an image into it thanks to the error correction built into the encoding, and this project seems like a possible evolution of that. Do you think it will catch on? Could this be the one of the first widespread applications of generative art?
They work better when they're a bit smaller. Downloading the image and scanning them did not work for me. But through my camera from a distance they all got picked up very easily.
Something about the combination of my laptop and my phone makes Samsung's scanner very inconsistent. Maybe my screen isn't bright/constrasty enough for it to easily pick up the patterns. I know it's not a hardware problem because Google Lens handles it fine.