GOP / Rate / Size Question for pros.....

[taliyev]

I am trying to figure out the optimum balance between compression rate, GOP and file size. If I am not mistaken the size of the file determined by the compression rate. So change in GOP (even with I frames only) should not result in the file size change. I tried different GOP with the same constant rate and ended up with the same file size.

On another hand I read that GOP does affects compression/file size. At the moment I am not sure how exactly it happens since my experiments do not show this effect. What I see is a difference in quality. It feels like when I use ‘I’ frames only I get bad reaction on any movements (according to the literature result should be opposite). Longer GOP seems to produce smoother image (sometimes too smooth/unsharp).

What am I missing here? What it the correlation between GOP/Rate/Size?

[AceYu]

If I am not mistaken the size of the file determined by the compression rate. So change in GOP (even with I frames only) should not result in the file size change. I tried different GOP with the same constant rate and ended up with the same file size.

I'm not sure of the specifics of GOP etc, except to say that as part of the data stream they contribute to the overall bit rate. Therefore by using constant bit rate encoding when testing the only change you will ever see will be in quality. File size is determined by the bitrate, quality is determined by the amount of compression, both dependant variables, ie a higher bitrate allows less compression and therefor better quality. Sorry this doesn't really help with the origional question but might help with your testing procedures.

Later BRETT