Last night I did the firmware hack on my 840K Toucam to effectively make it a 900nc. The hardware is exactly the same, but the 900nc has the option to use the lower compression YUY2 codec. The hack went without a hitch, so I decided to do an experiment to see exactly how much less compression there was on my video captures. The following 4 captures were each for exactly 60 seconds:
5fps IYUV codec - 133MB
5fps YUY2 codec - 180MB
10fps IYUV codec - 270MB
10fps YUY2 codec - 234MB (dropped ~200 of the 600 frames, about 34%)
Based on this data it, you can clearly see how much less compression there is between the YUY2 codec compared to the IYUV codec. However, it appears to me as though there is no difference in compression rate when moving from 5fps to 10fps. When reading Mikes planetary imaging guide I was under the impression that there was higher compression when moving from 5fps to 10fps, but this data doesn't seem to show that. Am I missing something here?
Based on this data, I would say my best option would be to capture at 10fps YUY2 codec in good seeing, accepting the frame loss as I still capture 33% more frames when compared to 5fps. In poor seeing I should capture at 10fps IYUV codec for no frame dropping, to capture more frames with higher compression.
Maybe I am mississing the point here somewhere. Is the render size of the file on the HDD independant of the transfer of data from the camera to the pc with regards to compression? In other words, the file size on the HDD will be the same even if the data transfer is of a higher compression?
Thinking about it more, the USB 1.1 specification only allows a maximum data transmission of 1.5MB/s which is well below the 4.5MB/s that I seem to be capped at with my AVI files, so there must be some underlying compression that I am not taking into account.
push it to 15fps and see the compression kick in, then again on 20 ~ 30 fps.
widen parameters to get broader results.
I found when using the default phillips software with my 900nc, @ 25fps, a 2 min video would be around 30mb. using K3CCDTools, @ 10 fps a 2 min video would be closer to 1gb.
I did some more testing and couldn't repeat the 'dropped frames' I was having last time. The size on HDD went up proportional to the fps captured at. At 25fps I could visually see the differences compared to 5 or 10fps though, so compression was definately effecting the quality, but not the filesize.
The comparison between YUY2 and IYUV still stands though, since for the same fps rate the filesize was always bigger.