On Sat, Oct 30, 2004 at 06:42:18PM +0200, Dik Takken wrote: > I have done a bit of testing, comparing these two encoding > pipelines: > > png images -> yuv4mpeg -> mjpeg -> mpeg2 > > png images -> yuv4mpeg -> mpeg2 > > The quality produced by the second pipeline is clearly a lot > better. The mjpeg compression step itself should not lead to any > *visual* image degradation, because I used the -q 100 flag.
But, *visual* to a human and *visual* to an mpeg encoder are two extremely different things. The incredibly complex signal processing and filtering that occur in your retina and brain would leave even the most complex mpeg encoder looking like junk. Or in other words, simply because your eyes/brain can filter it out for you does not mean that the mpeg encoder is capable of the same feat. > My only explanation is that mjpeg2enc amplifies the mjpeg > compression artifacts far enough to become visible. This does make > sense, because compression artifacts are hard to compress, which > leads to even more artifacts. This is most likely the cause. There are always artifacts from jpeg compression (even at -q 100). Their lack of visibility to a human observer does not mean they do not effect an mpeg encoder. And since the artifacts would appear as edge boundaries, they would require a fair number of extra bits to encode. ------------------------------------------------------- This SF.Net email is sponsored by: Sybase ASE Linux Express Edition - download now for FREE LinuxWorld Reader's Choice Award Winner for best database on Linux. http://ads.osdn.com/?ad_id=5588&alloc_id=12065&op=click _______________________________________________ Mjpeg-users mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/mjpeg-users
