Monday, December 8, 2014

Technical Mishaps and Adventures....

So, I went & checked my motion recording setup and shiz had broken. Le sighz. Again, I am having problems with 32 vs 64bit. Upon checking my iphone as webcam setup for motion capture recording, I find I get this strange bug:

*** QTCaptureSession warning: Session received the following error while decompressing video: Error Domain=NSOSStatusErrorDomain Code=-12905 "The operation couldn’t be completed. (OSStatus error -12905.)". Make sure that the formats of all video outputs are properly configured.

After much googling, I couldn't really find a good answer. I do know that it is something in my build or project settings, but what, I don't know. I'd be interested in an answer of course...  It is definitely related to something I did to make the IT++ library work with the 64-bit dynamic libraries since it was working before then. (or perhaps something when I was integrating matlab/mex in C++ but now decided didn't need it) Yes I could check BUT: 

Also, the Mobiola Webcam does not work in 64-bit (the OS X desktop half hasn't been updated since 2012). Other iPhone as webcam software I have tried either doesn't work on OS X or has around a 1sec. delay... which is pretty horrible. Mobiola so far has had waaaay better performance, as a real-time webcam.

After some mucking around... (too much), I finally decided to simply create a 32-bit video capture recording application that received sync'ing messages from OSC so I could capture the information. Even if I got the video capture in my current main application, it would be 64-bit & I'd have to use an iPhone app with a lot of latency. I figure its a temporary fix until Mobiola updates or I find a better iPhone app**. & then I will go through the pain of trying to fix that weird error.

Secondly, I found a strange Android bug in the Shimmer API. At least, I'm pretty sure it is from their code since its basically a memory leak in the JNI & I am not touching that part of the code. What happens is gradually, memory leaks until it runs out of space, causes a JNI pinned array ref problem and Android restarts the app. This happens after 30-45 min. so its not a disaster... I can simply restart, I guess, but I will need to hunt that down and fix it at some point...it is kind of a pain to have to force a restart every 1/2 hour. First, I'll report it to the Shimmer pplz, tho. I suppose its possible I need to explicitly free something, but I would think the garbage collector would be doing that, since I'm not dealing with the JNI directly... Its like all unsafety of C++ with the annoyances of JAVA. Joyous.

 **Hahaha I am not going write one I have better things to do

No comments:

Post a Comment