I completely agree with the statements made by Blair MacIntyre in this article. No miracle, he is one of THE experts for augmented reality from the beginning. And, also opposite to me, a developer. He really knows what he is talking about, when he states that object recognition, accessibility or processor power are bigger issues, than they appear from a hype point of view. It reminds me of a lunch talk with one of our mobile AR developers, where he drawed my attention to the problem of the very thin lines of a barcode and the interlacing in a live-video. The real world reference is an established standard, OK, but he doesn´t want to be the one to deliver a robust tracking within a mobile (low resolution camera!) consumer application (very robust, very safe, very “I push the button and it works”) within the next year. So, a video like this for example will perhaps win the “AR Mock-up Fake and Post Production Award” (ARMFPPA) – established right at this moment, so no problem in case you don´t know – but it has got nothing to do with reality. Not even in an augmented sense. The problem is: they get press, they get blog entries and they are creating a hype wave, whereas the really cool developments and small but safe steps get owerflown. I hope that we can nevertheless create islands where the good demos can live a happy life and grow up until they lead to bigger and better applications. There we´ll wait until the deluge is over.
If you see this sign (Razzie Awards Icon Halle Berry) on a demo, beware! It could be on the shortlist for the ARMFPP-Awards…
Given that the Gizmondo, last gen laptops, and many other devices of not to powerfull processors can do Augmented Reality to say processing power is a problem is quite heavy handed.
A problem for -some type- of AR applications maybe.
But if we can do this already;
http://nl.youtube.com/watch?v=Y9HMn6bd-v8
I dont see much limits.
Also, cameras on AR devices really shouldnt be interlacing to start with. We have moved beyond VCR’s and CRT displayed, its not of benifit any more.
Resolutions too small will naturaly make barcodes hard to read still.
What might be of benifit is dynamicaly changing the resolution of the mobil’s resolution to suit the situation. So it recognise’s its a barcode, switchs mode, and then tries to read what the barcode actualy says.
A barcode to be “scaned” by a camera dosnt need realtime tracking. It just needs a simple picture, of say, a few megapixals.
After scanning it can flick back to a resolution that can be updated fast enough for fluid(er) AR.
So a mobile device can fairly cheaply have a high-res camera these days, but it dosnt need the processing power to use the full resolution AR at 25fps.
Not everything so bad. Here is Nokia N95 AR game running at 11-12 fps. With more hardcore optimization it could probably run considerably faster.
Thank you very much, guys. That´s what I wanted to have as a reaction. Although I do not agree with all the points stated above, I apreciate sending in demos worth to mention – what they definetely are! Not everything so bad, well, I agree.