no way to compare when less than two revisions
Differences
This shows you the differences between two versions of the page.
— | transcode [2013/06/28 22:03] (current) – created beandog | ||
---|---|---|---|
Line 1: | Line 1: | ||
+ | ====== transcode ====== | ||
+ | |||
+ | === archives: transcode notes === | ||
+ | |||
+ | howto - http:// | ||
+ | |||
+ | < | ||
+ | |||
+ | < | ||
+ | |||
+ | dvd::rip transcoding | ||
+ | |||
+ | (pass 1) | ||
+ | < | ||
+ | |||
+ | (pass 2) | ||
+ | |||
+ | < | ||
+ | |||
+ | framerate.txt example for variable frame rate (updated) | ||
+ | |||
+ | < | ||
+ | |||
+ | *** Ripping TV Shows *** | ||
+ | |||
+ | These are my notes from all the testing with bend on TV shows. | ||
+ | |||
+ | Some generic conclusions: | ||
+ | |||
+ | - Do a two-pass and force deinterlacing on a live-action show for best | ||
+ | results if the show is not variable framerate. | ||
+ | - Do a one-pass with hard_fps dropping on cartoons | ||
+ | - If it's a variable framerate, do a one pass and force the input and | ||
+ | output framerate. | ||
+ | |||
+ | Everything else documented revolves around those three basics. | ||
+ | |||
+ | * 2pass notes | ||
+ | |||
+ | The two passes on cartoons seems to blur the lines in every case. It | ||
+ | could be because I'm not doing --hard_fps at all, but adding that | ||
+ | usually throws the A/V out of sync. | ||
+ | |||
+ | Without deinterlacing on a 2pass, you will see some horizontal lines as | ||
+ | artifacts on *some* shows, not all. It seems to be the minority, | ||
+ | actually, as most do fine with just 2 passes. | ||
+ | |||
+ | For deinterlacing (on 2 pass), -I 1 and -I 3 seem to work fine. -I 3 | ||
+ | has consistenly delivered great results, and limited testing proves that | ||
+ | -I 1 works fine. The only problem with -I 3 is it doubles the encoding | ||
+ | time, so a one-hour show will take up to 4 hours to encode on two | ||
+ | passes. | ||
+ | |||
+ | * 1pass notes | ||
+ | |||
+ | If you are forcing the framerate, go ahead and use all the filters | ||
+ | mentioned in framerate.txt. | ||
+ | |||
+ | If not, just use 32detect. | ||
+ | sync. | ||
+ | |||
+ | You can use --hard_fps on both, though. | ||
+ | |||
+ | Generally speaking, 1pass usually will result in larger filesize. | ||
+ | difference between one and 2 pass for filesize can be dramatic (I' | ||
+ | seen from 20 megs to 200), but not consistently rational. | ||
+ | is working and looking really nice, go for it. You'll probably end up | ||
+ | with larger files, but save time and get good quality. | ||
+ | |||
+ | Overall, I'd say use the one-pass method if you can afford to be very | ||
+ | liberal with harddrive space. | ||
+ | actually causes jitter, so the two-pass is better because it cleans that | ||
+ | up. | ||
+ | |||
+ | # Building CVS | ||
+ | |||
+ | Checkout the CVS | ||
+ | |||
+ | < | ||