Today at part of my ISMAR poster demonstration I am releasing the new SSTT visualizer. It looks quite tame but I can tell you there is more to it than just the looks. It will take a while to document all features.
Just as I mentioned before leaving PPC behind and cutting a few corners here and there I would like to confirm SSTT Visualizer is fully capable of running on Snow Leopard
Unfortunately, due to unresolvable issues with a new and very important feature I had to remove support for PPC. Thus, SSTT Visualizer 1.x will be first and the only Universal Binary build. I think this is more than reasonable with Snow Leopard shipping tomorrow and almost everybody has upgraded to an Intel Mac in one way or the other.
Update - and while I am at it, there is another feature that also makes it impossible to run SSTT Visualizer on OS X 10.4 aka Tiger. To sum up: 2.x will be Intel only and run on Leopard 10.5 and higher - no Tiger and no PPC.
As part of the SSTT framework there was always the question what can you do once the object is on the marker. Now SSTT Touch actually enables multitouch with AR markers, which opens up a number of application areas. Read more about multitouch AR and watch the video
LDraw enthusiast Jim Devona has been quite busy experimenting with the SSTT visualizer. He put together a tutorial on how to use LDraw models, the newest LDview with 3ds export (as of this writing you need to compile LDview yourself to get this functionality) and SSTT visualizer for viewing.
Looking at the new Android NDK (Native Development Kit) the SSTT Core library should be easily portable to Android as well. This is great as the iPhone OS 3.0 SDK was a huge disappointment.
Yes, there is no video capture and 3D rendering support in the NDK but for something like SSTT I only need libc - that's it. Rendering and video capture can be done in Java thus only leaving the tracking to an JNI wrapped version of the SSTT Core. Very smart choice from Google, lets hope Palm follows suit.
Ok, I might sound like a nitpicker - if you don't like research based argumentation and like the loud noise of the marketeer, please move on. My take: Layar has as much in common with AR as Wikitude and whatever LBS you can augment. And the Wikipedia article on Augmented Reality (no I am not linking it here because its just wrong) is written by a bunch of PR monkeys.
Lets look at the definition for Augmented Reality by Ron Atsuma one of the pioneers in this field and he provided this long standing three main characteristics:
A quick scan through the iPhone SDK 3.0 - very disappointing - no regular live video camera access, except for the known hack. So, we need to go back to the hack and hope Apple "ignores" it when submitted to the AppStore.
From an article on the WSJ:
Marketing executives are trying to make sure AR isn't just a flash in the pan. "Just because you can make a logo spin doesn't mean you should make a logo spin," says Mike Sabatino, senior vice president and partner at Fleishman-Hillard, the Omnicom Group public-relations firm that created the Papa John's pitch.
Sure, we just waited for all you marketing experts to tell us AR researchers; let me rephrase it: "Just because you can make a car drive on a marker doesn't mean you should make a car driving on a marker".
Hunting down some problems with the performance in the pose filtering some quite essential change needed to be introduced: fully pipelined tracking.