SharpCap Forums

Robin:

Installed newest version on my windows 10 pro machine, and attempted to rerun the sensor analysis on my ASI294 pro camera. The analysis had previously run well on this machine. I encountered some problems this time around:

The 8 bit analysis ran fine. But when I tried to run the analysis in 16 bit RAW space, it encountered several problems. First, when running at a small capture area (320by 240)) I was continually having frame drop problems. Once, when i changed the ROI to the full area, it started to pick up the frames again. But, usually when in 16 bit mode, there are always frame drops, regardless of how I play with frame rates, turbo settings, etc. I'm running on USB 3.0 BTW.

After I shut down the computer and restarted it seemed to be working (at first) but I had the lighting set too high, so I cancelled the analysis and tried to restart the tool. But it kept bringing me back to the position where i stopped the previous analysis, and I could not get it to keep going. After a restart, I ran into the dropped frame problem again.

Attached are two log files - the "dropped frames" was associated with the first issue mentioned above, the one titled hang on cancel was with the second one.

I've also included a comparison of the two different 8 bit analyses I've done - one about two weeks ago, and one from this morning using .4949

I should also note that Avast quarantined the install file for some time, but eventually let me load it. Not sure what that means, or if Avast was somehow contributing to the issues.

Best

john

Hi John,

thanks for the report - I think the likely cause is selecting too large an area for the measurement - that makes the calculations needed for each frame slower and that means frames get dropped because the analysis code isn't ready by the time the next frame comes around. I also made a change to the way I work out the median value of the frames between 4944 and 4949, but this doesn't seem to be slowing things down in my testing compared to the old way.

I have decided that the best approach to this problem for now is to limit the maximum (and minimum) area that you are allowed to select for sensor analysis (it will have to be between 10000 and 300000 pixels or you will not be allowed to proceed).

cheers,

Robin