Implementing a way to load several model formats, such as obj or 3ds. This is the best AR-SDK I have used so far, but since you are basically required to use a 3D-Engine to display models, there is a lot of overhead. Just looking at the GLES forum here, I see a whole lot of threads about using models, that aren't in a native GL format (which doesn't surprise, since as far as I know, modellers such as Blender or Maya don't support exporting in this format).
It would also be cool to capsule the required initialization logic into something much simpler as it is now. Yes, the sample projects provide the necessary code, but if you had it to develop by yourself it would probably take very much time to do it as efficiently as it is there. Its like 400 lines of code in the main activity, plus the custom GLSurfaceView, plus the required native methods. I could never figure that out by myself.
shovancj, render camera to texture is supported in the latest beta. Have a look, it is pretty cool.
(they aren't doing AR, just feature tracking supported with IMU). These are academic guys - so far as I know it isn't available commercially... yet. Seems to me that whoever gets it out first in an SDK will be way ahead of the pack. This thread is at http://ar.qualcomm.at/node/2001332
Render the Camera to Texture. :D
here is another one.
- the ability to convert manually in unity arranged image targets to one multi target and generate the according .xml code.
i guess this could be done with a editor script but it would be a nice thing to have it already implemented in the QCAR package.
- a "debug" modus where the feature points, that are actually found, can be visualized.
- a feedback on how good the actual performance of the tracking is (on the running app). could be a ranking system like in the TMS.
2 more wishes on the wishlist itself:
- a system where users can rank wishes to give it higher priority similar to this.
- feedback from Qualcomm on what wishes are probably going to be considered. i know you won't say anything for sure but some feedback would be nice.
and a completely different one:
- what do you think about a webinar or something similar once in a while. this would be a much more direct way to communicate and learn some new features.
[U" />Larger tracking distance for Frame/pattern-markers[/U" />
I'd just like to add that for various applications, we at twnkls would also REALLY like (frame) markers to be detectable/trackable at larger distances (as with ARToolkit, for example).
This is one of the things we just had to ask for when we met the Qualcomm AR team at ISMAR this year: we are developing several novel fusion algorithms that combine the vision tracking with the device's sensors, and the ability to track markers at a greater distance would really add great value. In fact so much so that with the current max distance, we would have to use an alternative tracker, which is certainly not what we want.
For these scenarios, we have no problem whatsoever with losing the picture in the marker, so an alternative marker type would be a good solution.
[U" />More ID's for pattern markers[/U" />
Tracking at larger distances would also benefit our SocializAR application, for which we would also appreciate a larger number of ID's (2000 or more, so that we can also handle larger conventions with our application).
i have a few features at the moment that i would really like to see in a future version.
1. a unified version of the unity plugin for both, android and ios, or at least a clear step-by-step guide on how to implement both plugins in one single project. multi platform publishing form one single project is one of the best features of unity and you should definitely take advantage of that!
2. it would be great to have a feature that autoadjusts the positions of the parts of a multi target in runtime. as it is sometimes hard to match the real position of the single targets exactly with the multi target definition a solution similar to what elecman did here could be a great improvement and used for fine tuning the position. i don't know if you actually have something like that already implemented in your code but if not maybe you want to consider it.
3. non rectangular trackables. i'm mainly interested in triangular trackables that can be used to build up more complex multi targets out of them based on a triangulated mesh.
4. is there any possibility to allow us to manually set/delete feature points in the TMS? i'm working with pictures and in many cases it would be really helpful to be able to do something like this directly in the TMS not just via image editing.
i hope i made myself clear, if not please get in contact with me.
Target Management System API
Some sort of API to generate new image targets on the fly.
This could be either built in or via the Target Management web service.
Maybe you could outline reason why it may not be add it to the SDK?
I know it may slow down the app etc. but it'd be good to have the choice and who knows what kind of use cases people would com up with.
I like the idea of a user being able to take a picture of something and then use that picture for tracking.
The image could be thrown up to the Target Management web service using an API of some sort and the .xml and .dat files would then be available to download for that particular image. Current API's could then be used to swap in those files and start tracking using the custom images.
Are you sure you want to delete this message?
Are you sure you want to delete this conversation?
To reset your password please enter the email address associated with your account. An email will be sent to you with instructions on how to complete changing your password.