There are many ways to do colour 2d barcodes, like HCCB...
But how viable is it to merge 3 QR or Datamatrix codes into separate Red/Green/Blue colour channels?
The benefit of this, is you can use existing barcode algorithms (not easy rolling your own!). Plus you can embed 3x as much information on the same area.
Backward compatibility with black and white QR cameras is probably not possible. But the ease of programmer implementations might help speed up adoption.
If combining the data from 3 channels into one file, then combine in this sequence of RedData+GreenData+BlueData. aka R/G/B . Good mnemonic.
----
Edits:
There is two additional concept added based on this discussion:
1. Tracking reference colours (BigSleep). To deal with various lighting conditions, perhaps add reference colours to the to the tracking boxes of each QR. Need to adjust vision algo perhaps, so its a bit more complex.
2. Mono Fallback (BigSleep): Embed a mono channel QR by adjusting luminosity per pixel. This allows for embedding a mono channel QR code that can survive black and white photocopy, and also is compatible with most QR code readers. Needs more complex reader tho.-- mofosyne, Dec 18 2014 CYMK QR Code concept http://briankhuu.co...CYMK_QR/CYMK_QR.pngCan't seem to blend RGB in inkscape D: ! But this is not too bad. [mofosyne, Dec 20 2014] CYMK QR Code concept (with integrated reference colour) http://briankhuu.co...CYMK_QR_compact.pngThis one doesn't need reference colour bar, since it already has colour reference integrated in the tracking pixels. [mofosyne, Dec 20 2014] 8 colour level vs 4 colour level 2d HCCB barcode http://spong.com/ne...-Just-Selling-_.jpgHCCB is cool, except its owned by microsoft. And they tried pushing... microsoft tag.... (8 level is Cyan, Yellow, & Magenta) (4 level is Cyan & Yellow only) [mofosyne, Dec 20 2014] .net paint layer filtering http://briankhuu.co...QR/CYMK_QR_Test.pdnJust activate and deactivate layers to expose each QR colour layers. [mofosyne, Dec 22 2014] Mono Fallback Concept for CYMK QR http://briankhuu.co...QR_monofallback.pngokay bigsleep, is this what you mean? [mofosyne, Dec 22 2014] Real Character, could be done in CMYK if you really wanted to ... http://en.wikipedia...Lord%27s_Prayer.jpg [not_morrison_rm, Dec 23 2014] Color barcodes - stanford http://web.stanford...asins/research.html [mofosyne, Dec 23 2014] Color Classifiers for 2D Color Barcodes https://fedcsis.org...s/2013/pliks/67.pdfMarco Querini and Giuseppe F. Italiano University of Rome “Tor Vergata”, 00133 Rome, Italy [mofosyne, Aug 22 2015] Reliability and Data Density in High Capacity Color Barcodes ? http://www.doiserbi...20-02141400054Q.pdfMarco Querini and Giuseppe F. Italiano [mofosyne, Aug 22 2015] High Capacity Colored 2-Dimensional (HCC2D) Code https://en.wikipedi...sional_(HCC2D)_CodeIt's coming? [mofosyne, Aug 22 2015] Wonder when HCCB (High Capacity Colour Barcode) be as free to use as QR...-- mofosyne, Dec 18 2014 Relatively simple for use with phones or similar readers. Most dedicated readers tend to use monochromatic readers (most commonly monochrome cameras). This would raise the cost of your reader.
Also, this raises the risk of a misread. Unless your color discrimination is perfect, you might get crosstalk.-- MechE, Dec 18 2014 Yea, hence I would think this is more for consumer purpose, rather than industry. E.g. Encryption key distribution is more realistic with higher density barcodes.
Industry specific barcodes would just stick to the lowest common denominator barcode for speed, and reliability. And plus they don't need to store much information on it anyway.-- mofosyne, Dec 18 2014 The only way you could truly store 3x the data is if some of the colored spots overlapped. Which then means you need to be able to discriminate not just Red Green and Blue, but the various combinations as well (since when they overlap you get other colors).-- Vernon, Dec 19 2014 --bigsleep
Kind of like the test colour mark you see on cereal boxes? (Used to check alignment, and colour quality)
Maybe set of coloured box of R/G/B and RG/GB/RB pixels next to the QR code as a calibration mark for the smartphone? So in any lighting condition, just check for the (R/G/B/RG/GB/RB) mark and use it as a known reference colour point.-- mofosyne, Dec 20 2014 Added a few [link] of a few CYMK QR mockups images.
Don't want to modify these "must be present pixels", they are for tracking, so need them to be present in all layers. So keeping reference pixel seperate...
Hmm... on second thought. Could work, if the smartphone is smart enough. So added that example in as well.
Anyhow, the examples uses CYMK, since I am not leet enough with inkscape to do additive blending of RGB layers. So am using CYMK colour blending scheme. This might actually be a better idea, since that's what most print shop actually prefer I'm sure.
Anyhow, now we got some concept CYMK QR codes. So if anybody knows how to do visual processing, I'll be curious if anybody is able to decode what I posted here.
Double bun if this works on printed paper from a smartphone. ( Maybe you got some Cyan, Yellow or Magenta colour filters lying around?)
---
since we are embedded C,Y,M layers, and these pixels may merge in any combination. That means there is 2^3 = 8 colour levels you need to distinguish. (It's possible to do so as well, since if you remember... HCCB also has 8 level colour mode. They also have 4 level colour mode as well, which judging by the colour scheme, is restricted to just Cyan and Yellow )
This is confirmed to be working at least via .net paint colour filtering. To extract the Cyan layer, I used the Yellow and Magenta filters (via 'mixing' blend option in .net paint).
Give it a shot in .net paint, via the .pdn file attached in [link]-- mofosyne, Dec 20 2014 Wasn't too sure what you mean bigsleep, so I made a mockup in the [link] as "mono fallback".
It's scannable by my phone, however my biggest concern would be the difficulty of decoding the colour information embedded in it. Especially in an uncontrolled lighting environment on paper via smartphone. How easy is it to equalize luminosity anyway, and how much colour information can be preserved?
But if anybody is able to make this work, and make it popular. Then they will by my hero :D
4x data storage anybody?
Also another thought: This fallback is really great when dealing with paper. Because even if it is scanned black and white, not all information is lost.
So in the context of... say a business card. If people photocopy the card, you want to at least keep the mono channel active, and storing the most important details like your phone number. The colour channel can store the non critical but helpful data instead.-- mofosyne, Dec 22 2014 [-] Due to colour fading on posters/t-shirts, and different "warmth" settings on display devices, the information resolution would be unreliable. You might be able to use a pure red, green or blue, just not orange or purple or brown.-- FlyingToaster, Dec 22 2014 --FlyingToaster
How do you explain the existence of HCCB bar- codes then? They use colour, and was used for the smartphone app Microsoft Tag.
Plus this is not going to replace mono QR codes. It's meant to be complimentary to it. So you'll see this more often in computer screens or printed paper.
On the issue of colour variations. One potential approach is to provide a reference colour of RGB, or CYM, colours in the barcode for calibration.-- mofosyne, Dec 22 2014 Get it over and done with, just use the Real Character..
NB why was all this old stuff written by an Igor...I mean "trefpafs"-- not_morrison_rm, Dec 23 2014 //just use the Real Character// Just use the ad itself: have recognition software look for the matching image in a generic web storehouse, and get the URL from that. No blotches, no codes, nuthin'.
[mofo] deboned, I initially read the post as suggesting there'd be a high bit depth.-- FlyingToaster, Dec 23 2014 Yea, which is why I really like bigsleep's idea of mono QR fallback mode for the CYMK barcode.
Btw noticed a stanford research on this. Looks very similar to the CYMK QR concept, but they coloured the tracking boxes with the 6 reference colours (excluding black and white) levels.
So really, this all seems quite possible. But the real question is if mono fallback mode is practical
BigSleep :~ Credits for the tracking box and mono fallback is corrected.
On observation, it does seem that the change in luminosity might make the colour harder to discern.-- mofosyne, Dec 24 2014 random, halfbakery