Two pass xvid question - converting a TV capture and saving time

[lchiu7]

I have captured a large file using a WinTV card in MJPEG format. The picture is interlaced of course.

I want to convert this to xvid to send somebody. I am using VirtualDub to edit out ads and saving the file with compression using the xvid codec. I also want to de-interlace the picture since it will be viewed on a PC.

With two pass xvid do I need to load the deinterlace filter for the first pass or only the second pass? On my modest Athlon XP 2400+ it is taking 7 hours(!) to convert the 90 minute video fro mjpeg to xvid and with two passes that's 14 hours! If I remove the deinterlace filter from pass 1, that drops to about 3 hours so there is a considerable saving in time. My understanding ism pass 1 just looks at the entire file, max and min bitrates etc and creates a stats file that optimizes pass 2. Would the application of a de-interlacing filter make any difference to the stats in pass 1?

Just wondering and trying to save time

Thanks

Larry

[jagabo]

If you don't deinterlace on the first pass the information recorded by Xvid during that pass will not be valid during the second pass (where you do deinterlace). The frames need to be identical for both passes.

If you don't need a file of a certain size you can use single pass, constant quantizer mode (constant quality). Pick the quality level you want (the quantizer value) and encode in a single pass. In this mode you know the quality of the output but not the size. In 2-pass VBR mode you know the size but not the quality. When the two make files of the same size the quality is almost identical.

With a quantizer value of 2 the output will be nearly identical to the input. At 3 there is a little macroblocking if you look at enlarged still frames but they are not visible at normal playback speeds. At 4 you might start noticing macroblocking if you look very closely during playback. Larger values will result in smaller files and lesser quality.

Another thing you can do to increase speed is use lower Motion Precision settings. Setting Motion Search Precision higher than "5 Very High" and VHQ mode to higher than "1 Mode Decision" will not increase quality or reduce file size by much. But they will take much longer to encode.