Hi everyone, I am a new to this whole augmented reality thing and am
very inspired by the flartoolkit. As I started to play around with the
demos, I started to realize that everyone was using a different
marker, even if it was the same marker, it was of a different size.
Hence, I ended up with a table full of different markers I have
printed out (since I am kinda messy I have no way to figure out which
marker was for what app).

So I was hoping if people could chime in if they have any ideas for an
open standard augment reality fiduciary marker. I believe that this
would generally help the adoption rate, if people could reuse the
marker for different applications.

it is interesting idea, but solution lies on the other end.

instead of making universal marker, one should make universal matcher.
this is actually perfectly possible at the moment for single marker
detector - you take single marker detector and replace standard
matcher with code that detects direction from marker contents in some
deterministic way. you obviously lose the ability to recognize things
(i.e. to return meaningful confidence level), but the rest of
flartoolkit operation will be unaffected.

anyone up for a job? eric? don't make me do it myself again :P

well, this is basically what mikael is working on, with the 16-bit  
(4x4) ID codes, no?
and yes, i agree with makc -- better to make a universal matcher than  
standards for codes.
sleepytime now, so don't have brainpower to think more about this  
right now but i think mikael's on the right track.

-e

On Nov 6, 2009, at 12:34 AM, Makc wrote:

btw, wouldn't it work already if you simply ignore confidence?

the problem arises if the application need to determine the size,
since the marker is printed by the user, there maybe scaling issues
(the 2 most popular paper sizes letter and A4, are slightly different
sizes). If I remember correctly, the flartoolkit, allows us to input
the size of the marker, and I am guessing it uses that to scale the
object and understand the distance from the camera. if the size it
off, then from an app standpoint we would have to offset or calibrate
the size (trying to avoid this route).

In general , I am thinking of what are the best way of reusing the
markers. And from a application standpoint and be able to confidently
determine the size and orientation. For example a dollar bill, is
always the same size, many people would place it in their photos to
help give context of the scale.

Another idea, is maybe we have have the marker displayed on a mobile
device, e.g. iphone, for people that don't want to bother with
printing out markers to easily start playing with AR (and it saves
trees too).

ok just tried it, tested a marker on my iphone, works pretty well, but
then again the size of the marker is significantly smaller then that
the app specified :(

the problem is non-existant in the case of 1 marker, flartoolkit
cannot derive physical dimensions from bitmap and therefor does not
care if your marker is 2mm and photographed through microscope or 200
km and photographed from satellite.

the size of marker you enter into flartoolkit is logical size in
virtual 3D space, and it only matters for correct scaling your 3D
models.

what about power wasted to charge your phone? any way, this was
already discussed here, in fact I tried that myself and my webcam was
not working with my digital camera lcd in specific application that I
did not want to print a marker for.

so what?