05-07-2010 07:50 AM
When I grab images from a single PXI-1411 framegrabber the latency is low.
When I try to grab images from two PXI-1411 framegrabbers (sequentially) the latency is approx 400ms, i.e. if I wave my hand in front of the camera it takes 400ms before I see the action on the screen.
From trial and error it seems that just calling 'IMAQ Grab Setup.vi' for a 2nd framegrabber causes latency in the 1st framegrabber.
The attached vi (LV 2009) is a modification of the example 'high level grab vi' to illustrate the problem. When the boolean constant is false there is no latency. When the boolean constant is true I get 400ms latency.
Framerate is ok, only latency is an issue. Why am I not getting the latest frame? Is there a buffer setting somewhere?
Thanks,
James/.
05-12-2010 08:20 AM
Hi James,
There is also an example that ships with LabVIEW that demonstrates how to acquire from two boards at the same time. This one is titled HL Grab with two boards.vi and I've attached it in case you can't find it.
There is also a community example that shows how to use a RTSI cable between the two boards and synchronise them if you want to do that: http://decibel.ni.com/content/docs/DOC-3924
Regards,
Imtiaz Chowdhury
Project Manager
Green Running / Austin Consultants
05-18-2010 05:14 AM
Imitaz,
The example you gave cannot on run with my frame grabbers unless I remove the trigger configuration vis, but then it is effectively the same as my example and the problem is still present. Is my example valid without triggering? One channel is OK without triggering.
Cheers,
James/.
05-19-2010 06:10 PM
Hey,
We are taking this problem you are seeing very seriously. I'm working on reproducing the problem you are seeing. To help us reproduce this problem, could you help answer a few questions? What OS are you using? (Windows or RT) I have been unable to reproduce this behavior with two PXI-1411's in Windows XP and Windows 7 so far using LabVIEW 2009 and NI-IMAQ 4.4.
Also, what camera file are you using? Is it the same camera file for both framegrabbers? What controller are you using to produce this behavior? We want to reproduce it in-house so that we can get started on a fix as soon as possible.
Thanks for your patience while we figure this out,
05-21-2010 03:47 AM
I've now tried to reproduce this with a PXI RT Controller and a PXI Windows Controller and haven't had any luck with 2 1411's. I've also tried 2 1409's and 2 1428's. Let me know if you're still experiencing this problem.
I hope all is well!
05-23-2010 08:18 PM
Evan,
Thanks for your efforts and sorry I haven't been able to get back to you sooner.
Windows OS. Both 1411's are set to standard PAL camera. LV 2009.
I need to check the PXI controller model and IMAQ version .
I should get the system back in the lab today or tomorrow so it will be easier to debug and try any suggestions.
Cheers,
James/.
05-24-2010 06:48 PM
Thanks for the reply. When you find out what kind of chassis you have, could you also note which slots you are using the 1411s in?
Thanks,
05-25-2010 05:38 AM
Evan,
I now have the system back in the lab so I can experiment.
The details you asked for:
Controller: PXIe 8130
Chassis: PXIe 1065
Slots: img0 = slot 3, img1 = slot 4 (both are non-express, non-hybrid slots).
Version numbers:
NI-IMAQ = 4.3.0
NI-IMAQ I/O = 2.3.1
NI-IMAQdx = 3.4.0
NI Vision = 9.0.0
Some more observations:
- Using my original example, if there is an error from 'Grab Setup.vi' on img1 there will be no latency on img0 (same as if the boolean constant is False). Examples of errors I noticed are if there is no camera input on img1 or if the img1 channel is selected in MAX.
- I have updated my example so that I grab and view video from both cards. See attached vi. I use the same video input for both frame grabbers. When I run this vi I observe the following:
img0 shows 400ms latency all the time.
img1 starts with no latency but the latency slowly increases until it matches img0, after about 10 seconds. Then both remain at 400ms latency.
- If I change the strings 'img0' to 'img1' and 'img1' to 'img0', the results are the same. This suggests no fault on either card.
- If instead I change the data flow so that img1 is intialised and setup before img0 then the results are reversed, i.e. img1 shows the 400ms latency all the time and img0 starts with no latency and gradually slows to 400ms latency.
Hope this helps.
Cheers,
James/.
05-25-2010 06:28 PM
Have you tried running the two acquisitions in two parallel loops? I am wondering if your problem is just due to the two reads being in the same loop. Sometimes you see delays when reading from two cameras in the same loop.
How do you know the delay is 400 msec?
Bruce
05-25-2010 08:25 PM
Hi Bruce,
I have tried parallel loops and the result is the same. Note that the framerate is OK, just latency is poor.
I measure latency by starting my stopwatch and moving my hand at the same time. When I see my hand movement on the screen I stop the stopwatch. I use my analogue wristwatch which only displays the stopwatch timer to the nearest 0.1 sec - anyway my reactions aren't better than that ;). I admit it's not very accurate but I get 0.4sec repeatedly.
Cheers,
James/.