This is an amazingly beautiful hack. I'm kind of surprised we never saw something like this done by hand before the advent of SD -- plenty of QR code art, but none of it like this. I'm impressed.
I seem to remember some NY artist trying to sort of vaguely claim/intimate invention (falsely) by having an exhibition too. Ah yes, here https://art-qrcode.com/
As much as I like to distrust ML algorithms to give exact, reliable results, here I actually think that a random picture that aims to be a QR code that is also drawn manually will have much lower chance of QR code recognition code than ML-generated one, because it's hard to manually restrict the work so that it is reliably recognizable by many QR implementations while also "integrating" QR code features with art features so that they are not obviously two independent layers one laid on top of another.
It's costly to make such images, that's about it. Other way around you'd have to decide on the QR, wait for the commission, pay a substantial amount of money. Probably wouldn't even get such subtle codes either, this seems like straight up CV-trickery.
Something I learned recently: the phrasing "QR Code" is a trademark of a company called Denso Wave, who invented them. While the method of producing and reading QR codes is free for all, (Denso Wave waived their rights to the patents) they still protect their trademark. This has lead many Japanese companies that use QR codes, to call them publicly something else, such as "2D codes", in fear of getting sued.
Huh. Turns out it's a subsidiary of the same Denso that's a massive automotive supplier and largely owned by Toyota. Asian conglomerates are so interesting.
Vibrators. Hitachi is also known for their powerful, corded vibrators (it's not a dirty word): the Magic Wand.
Calling these 'marital aids' in 2023 seems a little outdated. Vibrators are devices used for sexual stimulation, which obviously makes them extremely useful for solo masturbation as well, in addition to being a sex toy someone can use on another person.
All magic wands are amazing (not just the hitachi ones, but try to find a decent brand like hitachi, bedgeek). Trust me guys, they're so good in many spots; treat yourself!
Vibrators have a lot of interesting heavy industrial uses. At massive scale, things like paint mixers are massive vibrators. I had a college friend whose job title on LinkedIn for many years was "Vibrator Engineer" due to specialization in that industrial field. I don't think any of the vibrators they worked on were "personal sized" and I doubt any of the companies on their resume had any plans to sell sex toys, but that was an industrial/mechanical engineering specialization important enough to those companies to be reflected even in job titles.
Relatedly, my favorite Japanese product line enigma remains Yamaha realizing that all the finely specialized industrial equipment for building organs and pianos could probably be multi-tasked to making motorcycles.
Interesting bit of info, thanks for sharing. Sadly for them, I think the genie is out of the bottle on this one; while I've seen a few people use alternate names, most call them QR codes.
In the Chinese world, it's still called the equivalent of 2D code in Chinese(二維碼/二维码). The original variant QR Code(QR碼/QR码) are used but to the best of my knowledge they are quite mixed together in terms of usage.
I think the mainland Chinese are more inclined to 2D code by the interactions with them in the past.
A QR code on a yoghurt that I bought couple of months ago came with some interstitial ad due to an unpaid fee. It made me double check the yoghurt's expiration date.
Fun to imagine in the abstract. I used to take classes in a building designed to have a certain appearance from above, and the user experience was definitely the worse for it
Error correction doesn't make it impossible to change a message by flipping bits, you just have to change enough bits, and know which bits to flip. I wouldn't rule out, that if you are being smart about it, you might be able to change it into a different valid url, without having to change too many of the squares. You might even be able to use the error correction to your advantage to an extend.
You could use an unnecessarily big QR code for extra correction bits to mitigate this somewhat (up to the maximum correction standard (30%?) even if that means 41 cell code instead of a 33 cell one). For URLs another size hack is to include meaningless extra data (a qstring parameter or anchor that won't be interpreted). This doesn't increase the error protection in itself, but gives the fancier more data to work with so it can produce better visual results without losing more of the signal correction strength. Using the generator at https://qrbtf.com/ someone linked above, try the “C2” style with https://qrbtf.com/ and https://qrbtf.com/#some-extra-data-to-increase-the-image-siz... to see the difference this makes.
The issue with artsy QR codes often isn't the individual bit flips, but that they often include large areas of all black or all white which can make it harder for the reader to pick up the actually correct bits if the viewing angle and lighting aren't perfect. The standard includes 8 mask patterns that you can freely choose from, the intention being that you pick the one which gives most black/white variation in the final result, so there is as few low-contrast areas as possible. With improvements in cameras and processing built into QR scanners these days this is becoming less of a concern, but if you need your code to be reliably readable by older devices, probably stay away from the fancy tricks.
The fancy tricks are cool though, worth all the caveats if they match your stylistic needs.
I love seeing companies blindly encoding massive URLs into QR codes that end up being ridiculously detailed images. Like they would be impossible to scan unless you were right up close with a good camera because the individual dots are so small.
If you zoom out they are very recognizably QR codes. A commenter made the point that what you're seeing in that case is the low frequency component of the image. Kind of like the Marilyn Monroe / Albert Einstein illusion.
There is, surprisingly, no correct labels exist for this class of images. Anime is incorrect because it’s not an animated film or directly derived from one, manga is kind of correct but modern definition is just panelists comics, nijigen is another term but it’s slightly deprecative, hentai is only for porn and also won’t be understood in Japanese. There just don’t seem to be one.
"Anime style" seems to be a decent compromise that is easily understood by most people, even if technically incorrect. AFAIK a similar term is often used in Japan: "アニメ風イラスト" (Anime style Illustration)
To the vast majority of people though, anime just means anime style. See for example the common term, anime profile picture. It's like the word meme, or irony, or begging the question.
I'm guessing bc "hentai" is only "pervet/weirdo/weird person", "ero" or "ero anime" is used for that in Japan, although I'm almost certain if you walk into a shop that sells it, and refer to it as "hentai" they'll know what you mean.
Assuming you mean a sex shop, 'hentai' would refer to basically the entire store's stock, so I think you'd need to be a bit more specific! It doesn't automatically refer to manga or anime like it does in the West.
Well, they're "the style that image models called anime produce".
These mostly start with real-world images and then are fine-tuned with Chinese/Japanese commercial illustrator art, which isn't anime (flat-colored TV episode screenshots) or manga (monochrome halftoned comic book pages).
The result is like 15% photorealistic shaded, so it's not a lot like anything else out there.
I just wanted to make it clear that its a long “e”, but pretty much the same word as “animation”, not the correct english spelling of an english-imported word :p
What I've seen most is the line put above the vowel, resulting in animē (but difficult to type). As there is no additional (fourth) syllable, a-ni-mee would be more logical to me, rather than a-ni-me-ee or a-ni-me-e.
That being said, I like how the triple vowel makes it clear it's not the same sound as in say meet. (This gives me the idea to change my name to tuuukkkah...)
"Manga style drawing" would be generally understood to mean looking like 188/190 of the pages of a manga rather than looking like 2/190 of the pages of a manga.
Your pedantry is neither appropriate nor correct. Manga is most often black and white purely because of time and budget constraints, not because manga's style is defined as black and white. There are tons of full-color manga, and many that become popular have color editions released.
It's not pedantry. I'm trying to convey some actual understanding here. These drawings aren't manga-style, you can tell because they don't look like (typical) drawings from manga; most obviously in that they're not black and white.
> Manga is most often black and white purely because of time and budget constraints
True but irrelevant. The style of the medium coevolved with those constraints and is suited towards them (as with most media). A manga-style drawing is stylised in particular ways that (among other things) allow it to look better in black and white than a typical drawing would.
> There are tons of full-color manga, and many that become popular have color editions released.
They exist (indeed I own some) but they are the exception rather than the rule. Again, "manga-style drawing" would be generally understood to mean a drawing that looked like a typical drawing from a manga.
> "manga-style drawing" would be generally understood to mean a drawing that looked like a typical drawing from a manga.
Generally understood by who though?
I see your point about what you said about the colours
> manga-style drawing is stylised in particular ways that (among other things) allow it to look better in black and white than a typical drawing would.
In particular I agree that the black and white manga drawing uses special techniques to look good in black and white. Such as notably the half-tone patterns that they use.
But I think you are placing more weight on this particular thing than perhaps most people do.
There are many defining features of manga aside from the lack of colour.
I guess to determine whether or not it is generally agreed that it must be black and white, we’d have to interview a bunch of people about what style some images were, and see if an overwhelming majority thought that colour mattered or not :p
Or, less thoroughly we might just ask the denizens of the 4chan /a/ board. They probably have some strong opinions about what counts as manga or not.
I think it's less about black/white and more about the outlined art style. A lot of these QR codes are painted. There are painted manga (Lost in Abyss comes to mind) but they are a rarity.
> From the title I was expecting the QR codes to move, but they're actually "anime drawing style", not "anime"
Given the description of the apparent technique, its likely trivial to extend it to animated QR codes using the same ControlNet models and ControlNet inputs and existing techniques of animating travel through prompt and/or seed space for StableDiffusion.
That is impressive. I know the images will have been selected for the fact that they are particularly good examples, but I've seen a number of implementations of trying to mix a photo into a “standard blocky code” (the girl and Trump examples on that page) and none previously produced results quite that good. I shall have to have a play.
> Next, they fed existing QR codes into the Stable Diffusion AI image generator and used ControlNet to maintain the QR code's data positioning despite synthesizing an image around it, likely using a written prompt.
A few months ago, after seeing that ChatGPT could produce reasonably accurate MD5 hashes, I tried generating QR codes with a Midjourney prompt. It failed in a really interesting way - it generated _symmetric_ QR codes, even though QR codes are explicitly designed to be asymmetric (think of the timing pattern, the three large blocks). So while I could claim ChatGPT "understood" MD5 hashes, I can't claim that Midjourney "understands" QR codes.
How long until generative models can produce working QR codes on their own?
Can they now? I had the impression that stabled diffusion is not moving forward that fast because the original team that built it is not the ones that are in stability.ai
For the love of sanity, please don’t. In real life, I struggle enough to scan normal QR codes when the redundancy of error correction is here to… you know… correct errors and help the scanning process.
Hah, even without the hacking angle, having URLs to promote shit popup when you point your camera somewhere could be the next obnoxious advertising.
For a few years now phone cameras have been able to automatically detect QR codes, maybe their next feature will be an option to turn off "smart features"...
I wonder if this would work by simply using out-painting to preserve the black dots, to make a background, and then use in-painting on the dots with the condition that they stay dark
Which app are you using? For most of them I needed to get a few feet away, as mentioned in the article. For some it seemed I needed to be more exactly facing it, not angled. I got all of them to scan pretty easily.
I'm using the default camera on my phone. It's called "Camera". The phone is a Samsung Galaxy. I've tried it close, far, angled, straight, big, small. I haven't gotten any of them to work a single time.
