IMAQCamera.setup_acquisition() costs very long execuation time

[jianrong2]

Hi,

Thanks for your contributions to the awsome project. I find it very useful to acquire high throughoutput video via the IMAQ wrapper.

The library works well though I found a minor issue when initilizing the camera interface. It take much longer time than the native application such as the NI-MAX.

With the help of breakpoints, I located the line self._max_nbuff=self._find_max_nbuff() of the method def setup_acquisition(self, mode="sequence", nframes=100) in IMAQ.py, which might cause the issue. With hard-coded self._max_nbuff=5000 instead, the script runs much more smooth.

Notes: My setup is NI PCIe-1433 frame grabber combined with a GL2048R InGaAs line camera. And I use cam0.setup_acquisition(mode="sequence", nframes=100) to initilize the camera.

[AlexShkarin]

Hi,

Thank you for the kind words, and for pointing out this issue!

Yes, it is a known problem with IMAQ that allocating buffers takes super-linear (almost exponentail at some point) time, and it can take a many seconds for more than ~30-50k frames. Since the code has to do it to find the maximal available bumber of buffers, that causes the slowdown.

I agree, this is not handled in the best possible way right now, and it should, probably, be a constructor parameters. If you want to avoid changing the pylablib source files (since it makes the software not transferable between different PCs), you can also call cam0._max_nbuff=5000 right after opening the camera and before the first call to setup_acquisition or start_acquisition.

[jianrong2]

Thanks for your reply. I tried cam0._max_nbuff=5000 instead of modifying the library source file, and found it worked perfectly 😄.