Thanks for your very informative tutorial "Encoding screen recordings for Silverlight in VC-1 with Expression Encoder". I have experimented with various settings and settled on VC-1 Quality based VBR mode. I had 2 interesting findings:
1. I thought B frame can reduce file size because it can be more efficiently encoded than P frame. However my experiment seems to indicate the opposite. Changing B frame from 0 to 2 caused 18% file size increase in quality based VBR mode, so I would guess in peak bitrate VBR mode it will cause quality decrease. Is there an explanation for this?
2. play back video with long key frame interval. I have encoded a 2 hour long screen capture video with a key frame interval of 20 seconds. when I play the encoded video file in WMP 11, I found it difficult to randomly jump to the middle of the video. WMP 11 always reverts back to the previous key frame which could be up to 20 seconds away. However, using exactly the same source and same setting, I encoded the first 10 minutes to a video file, WMP 11 can smoothly jump to the middle of the video and start playing at the exact location I specified, without reverting back to the previous key frame. I can consistenly reproduce this phenomenon with various sources and various key frame intervals up to 100 seconds(Expression limit). It is puzzuling. do you have any insight into it?