Continous capture to AVI -- not enough memory

Question

I'm running LabView 2009 right now with two Camera Link  cameras (JAI CM-200MCL), each capable of running at roughly 25 fps with an NI PCIe 1430 frame grabber (They're rated for 30 fps, but I haven't been able to get them to run at this speed in LabView). I'm trying to get it so that I can capture a 30 second movie at least in which the images in both videos can be roughly coordinated with eachother in the time domain. 
Currently, I'm running into two major problems:
1) When imaging with both cameras and writing them to their own AVIs simultaneously, the respective camera fps values drop to about 17 each, screwing up the time fidelity of the resulting video and making analysis virtually impossible.  
2) Whenever I try to implement the code with newer AVI sub VIs in labview (requiring a compression codec), or when I try to take the raw VI and compress it by reading it and writing it to a new AVI file, my system runs out of memory and I get a message like this:
Error -1074396159 occurred at IMAQ AVI2 Write Frame
Possible reason(s):
IMAQ Vision: Not enough memory for requested operation.
Does anyone know what I can do to solve this? Currently I'm at a loss as to next steps and can't seem to find a solution in the various discussion threads I've read on similar topics. 
Attachments:
Vision Acquisition_2cam.vi ‏200 KB
Grab and Save to AVI.vi ‏52 KB
Convert Avis.vi ‏18 KB

Answer 1

Hi floomich86,
When do you get the error?  How many times the acquisition loop runs before you get the error?
Looking at it now, I think you're right that it's not a throughput issue, but has to do with the memory being used up.  I have a couple recommendations:
1) Try using a codec after to process your images so they're smaller.  Use the IMAQ Get Codec Names node to find the various codecs available on your computer (it returns an array of all the various codecs), then use an index array function to choose one of these (try an NI one to start).  Then wire this codec into your IMAQ AVI Create node in your Vision_Acquisition_2cam.vi.  Adding this compression should help your images take up less memory.
2) This seems like a good application for the Producer-Consumer architecture.  This would require revamping your code in a major way, but I think using this architecture would improve performance.  This architecture is best for programs that need to gather data in one loop and process it in another simultaneously.
Here are some examples of the Producer-Consumer architecture.  I tried to find some specifically with Vision functions in them but I couldn't.  Yet, you can still see how the architecture and data gathering/analysis works with these.
Here is a white paper to introduce the architecture: 
http://www.ni.com/white-paper/3023/en/
Here are some examples:
https://decibel.ni.com/content/docs/DOC-26558
https://decibel.ni.com/content/docs/DOC-2431
https://decibel.ni.com/content/docs/DOC-2145
Please let me know if these are helpful.
Julian R.
Applications Engineer
National Instruments