PDA

View Full Version : Does 2 pass deliver smoother video or not?


Commander XJL
21st April 2003, 21:03
Is the only advantage of 2 pass in tmpg the fact that you can more acurately figure file sizes? Lets say file size isn't a concern, would video look as smooth with constant quality as it does with 2 pass or does 2 pass deliver smoother video?

Arky
22nd April 2003, 00:40
Well, in theory, the second pass of 2Pass VBR has a marginal potential advantage in that bitrate can be allocated that little bit more effectively on the second pass, allowing the limited bitrate to be used most effectively and thus, yes, I would say it is quite possible that you would get a slightly smoother quality to your MPEG. However, this really depends quite a lot on a number of things:

# what average bitrate you are encoding at (higher bitrates would yield smaller subjective differences between the two techniques)

# how noisy the source footage is (the noisier it is, the more demand it places on the encoding process, so bitrate needs to be more carefully allocated to maintain decent quality - 2pass VBR would win out over constant quality in this situation).

# how much fast-action there is in the source material -same reasons as for noisy material, above.

# If the source footage has rather subdued lighting (dark). Again, for similar reasons to noisiness, above.

I would suggest that you find the most demanding few minutes of your source footage, on the basis of the above criteria, or any combination of these criteria, encode using constant quality, and see if YOU are satisfied with the result. If you are, then fine - don't bother to do the whole job with 2-passes. If it isn't, then run the same few minutes through 2-pass VBR and see if there is a worthwhile improvement to justify the whole film running through 2-passes, particularly in light of the slowness of TMPGEnc per-pass.

I don't use TMPGEnc very often these days (purely on the basis of speed - the quality itself remains superb, unless extremely high speed action is being encoded, which can occassionally trip TMPGEnc up), but I regularly used to use Constant Quality and got great results.

Trust your own judgement, on the basis of the short-test technique I outlined.


Arky ;o)

Commander XJL
22nd April 2003, 01:36
Thanks Arky!