Sunil's Blog

I’ve been experimenting with how Flash and FlarToolKit handles multiple patterns using FLARMultiMarkerDetector .  I’ve found that a lot of light is needed for the detector to work reliably with mutiple markers. Here are a couple screens of my progress so far using animated MD2 models of Boba Fett and Starscream. Thanks to my cat Jasmine for the moral support.

With Jamario Moon being dealt away to Miami just last week, it seems fitting that I revisit an old project Paper-Robo and resurrect Paper Jamario Moon.

I finally got a chance to sit down and play with FlarToolkit. For those unfamiliar, FlarToolkit is a open source library that allows you to manipulate 3D objects using a pattern (like QR codes), a webcam and Flash.

Thanks to Mikko Haapoja and Saqoosha for some awesome start up guides:

http://www.mikkoh.com/blog/?p=182

http://saqoosha.net/en/flartoolkit/start-up-guide/

I had to jump through some Flex hoops to get through some of the demo examples to work. Apparently you can only use the following syntax in Flex to embed binary files:

“[Embed(source="../lib/mypattern.pat", mimeType="application/octet-stream")]”

Since I haven’t used Flex before (I use FDT), I got around embedding these camera calibration and pattern binary files by using loaders. If you are using loaders, be sure to add MIME types on your server for .dat and .pat files. Other than that I didn’t change the original source code too much, other than replacing the basic cube primitive with a collada model. I also threw in some animated MD2 quake models (Boba Fett and Starscream) for good measure.

Briefly to use this library you need two binary files, a calibration file “.dat” for a webcam and a pattern file “.pat” representing your printed pattern. You specify a desired accuracy, and using a detector class you can determine if the marker is FOV of the camera. If the marker is detected you add whatever 3D object you want to a Flarbasenode, which is a Papervision3D Display Object. The Flarbasenode can then be translated/rotated using information from the detector class.

Try it yourself:

Download and print pattern

Test the application here (Webcam required)

Here is some sample code and other resources to jump start your Augmented Reality project:

Here’s a good repo of free animated MD2 models Project:

http://polycount.com/models/quake2/

Sample Code:

http://www.mikkoh.com/blog/wp-content/uploads/2008/12/learningflartoolkit.zip

http://www.boffswana.com/news/wp-content/uploads/augmentedreality/AR_dev_source.zip

Air application to make binary marker files:

http://saqoosha.net/lab/FLARToolKit/MarkerGenerator/MakerGenerator.air

Welcome to my first blog post, in the post-IMM era! Right after the IMM Open house and volunteering at FITC 2008 Toronto, I parachuted out of Toronto with my girlfriend for a much needed vacation to Paris (France not Ontario). I had a relaxing vacation away from my computer for 8 days.

While I enjoyed the wine, crepes with nutella and croissants, I also stumbled across something neat at the Adidas Store on the Champs-Elysées while I was there. They had what looked like a Gesture Tek installation right in the middle of the store, where you could design your own shoe. These kids ran up to the screen thinking it was a touch screen, a la iphone. They were clearly lacking the appropriate training required to operate such a display! Having come across Gesture Tek displays before in our multimedia pioneering class, I decided to put on a clinic. It was funny seeing people stop in their tracks while I created my own personalized Toronto Raptors sneaker, using a user interface that screamed Minority Report. After you create your shoe, you could save it in a gallery and walk over to a “virtual mirror” and see what it looks like on your foot. Unfortunately, the “virtual mirror” was broken, so I only got to play with the interface. Thank you to my wonderful girlfriend Kerrie who filmed me looking like a dork.

Adidas Paris – Shoe Designer with Gesture Tek Interface from insultcomicdog on Vimeo.

I’m already thinking of ways to port this over to a Papervision 3D project using Dan Zen’s motion cursor classes as a user interface. You have already seen examples of the motion cursor classes used in Rock Revolution a project I built with my classmate Nick Poison. I don’t think motion cursors have been implemented with Papervision 3D yet, so hopefully I’ll come up with something neat.

So here’s my first attempt at creating 3D panoramas in QTVR and Flash using the Interactive Multimedia Lab as a demo. I used a “point and shoot” Canon PowerShot and a tripod.

Things to note:
- Flash file isn’t optimized for the web
- The top and bottom panels are blank
- Limited to a cube, which forced me to squish the photos
- The QTVR file is a lot lighter in file size

QTVR: (Needs Quicktime Plugin)
http://www.suniljohn.com/projects/panotests/qtvrs.html

Flash w Papervision3D and AS2:

http://www.suniljohn.com/projects/panotests/simplePanorama.html

I would like to figure out how to shoot the ceiling and floor to make the cube complete.