Log in or register to post comments

Can QCAR pick up dots and lines?

March 4, 2011 - 9:22am #1

I've ran some samples and am getting ready to put some of my ideas to work this weekend. To help me narrow down my list, can QCAR pick up lines or dots?

I'm referring specifically to dots or lines drawn on a piece of paper. I was thinking of trying out some basic tech demos with some algorithms I've learned in class. Stuff like finding a route through a drawn maze, or doing nearest neighbor with a series of dots, etc.

My ideas may not even be possible with AR, and be more suited to analyzing an image, but was just curious as to what all could be a trackable.

Re: Can QCAR pick up dots and lines?

March 6, 2011 - 7:07pm #4

So 5 trackables at a time isn't a hard limit, it's just a good starting point for performance and usability.

Do keep in mind that this means 5 simultaneously tracked targets, the application can "know" about a much larger set of targets (e.g. that are present in the config file). You're going to have a pretty hard time fitting much more than 5 targets in the camera view anyways, while getting a high-quality view of each.

Also, it sounds like the targets you are going for here might be too simplistic, remember you need a good amount of detail in each image to track off it.

- Kim

Re: Can QCAR pick up dots and lines?

March 6, 2011 - 2:21pm #3

So from what I'm reading, QCAR can only process up to five trackables at a time. If this is the case I'm finding it hard to imagine a scenario I could use it. What I would want to achieve would require more than simply 5 points, and if it couldn't track more than 5, there's no way it could draw the lines to each point. Also, with Markers I don't understand how a complex game could be played with only 5 trackables in play at a time. I think I may understand why Eye of Judgment only had 9 cards in play at once however.

Thanks for your help. :)

Re: Can QCAR pick up dots and lines?

March 4, 2011 - 9:38am #2

The tracker won't return any information about the features it finds, other than the target's orientation. I'm not sure it can help much with the sort of image processing you want to do here... You can get the camera image for each frame though, if you'd like to do your own processing. You could use the tracker feedback to know when the entire image is in view (will require some math), which could be a helpful preprocessing step for your own algorithms.

Hand-drawn mazes and dot sets might make decent trackables, if the features are random and dense enough. You will still need to preprocess each target before tracking it, however.

- Kim

Log in or register to post comments