"We offer new support options and therefor the forums are now in read-only mode! Please check out our Support Center for more information." - Vuforia Engine Team

Big real model and the size of the multi target

Hi, for a demo, I'm using iPad2, Vuforia and Unity3D to show a real size model of ultrasound machine. These are approx 1.5m x 1m x 1m so the viewing distance is relatively big. The models should be changed at runtime, and the user should be able to walk around, zoom in by coming closer etc. Also, the height of the model should be appropriate, so the user needs to duck to see e.g. the wheels in detail. For this, I've been using multitarget (cuboid), physically a box approx 1.2m x 0.4 x 0.3 (I couldn't find a bigger box). The tracking images are bit difficult to find. I am able to print A0, so I just used some complex images with 4-5 star rating in TMS. However, the tracking performance is not very good. Once the tracking is lost, the recovery is quite lengthy. Also, the tracking seems to perform better from distance (~2-2.5m) than close up. I've been trying to experiment with different images, also modifying the XML by adding more image target in my multi target (e.g creating cuboid with more "layers"), but it did not seem to help. Any help will be appreciated! Thanks matej

DavidBeard

Tue, 06/05/2012 - 22:08

Can you post your MultiTarget configuration file? The detection shouldn't lag w/ 4 and 5 star targets. Also a picture of the physical box will help. You might try scaling your model down so that it is contained in the box for testing purposes - to determine the effect on tracking and detection.

Hi The thing is that the idea is to have approx. 1:1 scale, so if the real object is tall, the user will have to duck. This somehow prevents scaling the model down, otherwise it may disappear quite quickly from the view. You can also see the photos of the box.

DavidBeard

Thu, 06/07/2012 - 20:27

You might try composing each wall of that box as a MultiTarget, so that you can get closer to w/o losing the Tracking. That is, section the single ImageTarget into multiple regions - you're losing Tracking close-up because only a small proportion of the image is within the camera's FOV.