Re: [MPlayer-users] Bitrate filesize problem (2)
First off, please CC me in any replies. I recieve the daily-digest, so it will take longer for me to see the message, and be a little more difficult to reply, otherwise. (more below)
The calculation of 526kb/s seems correct. mencoder says it actually used a bitrate of 418kb/s. This usually means it didn't need all the bits you gave it to encode "perfectly". That seems unlikely in this case. What does the output look like?
It's obvious that it isn't as good as the source. It's a very rough picture, and it get's VERY blocky when there is a lot of movement. This is from a source that wasn't all that clean to begin with, so it's a major step-down in quality. In addition, I have successfully encoded it with vbitrate=1200, and the resulting file is 1.1G. Unfortunately, I didn't log that pass, or I would provide you with the output. I can encode something else (and log the output) with any parameters you please, if that would be of any help.
I'm using MPlayer 0.90pre6-2.95.3 on OpenBSD 3.2, but I saw the EXACT same behavior with 0.90pre8 on FreeBSD 4.7.
Yikes! Get something newer, at least rc3.
If someone would like to provide a set of patches to allow MPlayer to compile on OpenBSD 3.2-stable (a port/package would be even better) I'd be happy to upgrade. Otherwise, I plan to wait until there is a non-fractional release available, before I do some work on porting a newer version of MPlayer.
On Fri, 07 Feb 2003 19:14:04 -0800 rcooley <rcooley@spamcop.net> wrote:
It's obvious that it isn't as good as the source. It's a very rough picture, and it get's VERY blocky when there is a lot of movement. This is from a source that wasn't all that clean to begin with, so it's a major step-down in quality.
Yes, I would expect that even if it used all 526 kbits. For a 718x432 movie you will need at least 900kb/s to approach acceptable quality. I would do some denoising (pp options) then scale to 352x224. Even then getting a 2.75 hour movie on one CD is optimistic. And use the -ofps 23.976 thing that Richard suggested.
Otherwise, I plan to wait until there is a non-fractional release available, before I do some work on porting a newer version of MPlayer.
We shouldn't have long to wait for 0.90. Martin
On Sat, Feb 08, 2003 at 05:49:10PM +0000, Martin Collins wrote:
[Automatic answer: RTFM (read DOCS, FAQ), also read DOCS/bugreports.html] On Fri, 07 Feb 2003 19:14:04 -0800 rcooley <rcooley@spamcop.net> wrote:
It's obvious that it isn't as good as the source. It's a very rough picture, and it get's VERY blocky when there is a lot of movement. This is from a source that wasn't all that clean to begin with, so it's a major step-down in quality.
Yes, I would expect that even if it used all 526 kbits. For a 718x432 movie you will need at least 900kb/s to approach acceptable quality. I would do some denoising (pp options) then scale to 352x224. Even then getting a 2.75 hour movie on one CD is optimistic.
Scaling down to that size will make the movie look much much worse. Unless you can encode with constant qscale 1 or 2, it'll be really ugly at such low res. A much better approach would be applying strong denoise3d and/or gaussian blur, then using very strong adaptive quantization (*_mask and naq options) and single coefficient elimination (maybe vcelim=10, vlelim=-7). Also keep in mind that it won't look very good until after the second pass (vpass=2)!
And use the -ofps 23.976 thing that Richard suggested.
This should help a LOT. Rich
participants (3)
-
D Richard Felker III -
Martin Collins -
rcooley