Why VBR ? Quality and/or space ?

[SingSing]

Hi,

I have been doing CBR SVCR for as while.
I would like to know why VBR ? Better space and/or Quality ?
and how much longer does it take to TMPGenc the same movie ?

[bugster]

Any discussion about VBR vs CBR is invalid unless you assume that the AVERAGE bitrate of the VBR encode is the same as the bitrate used for the CBR encode (and all other things being equal!)

In theory 2 (or more) pass VBR should at least equal and possibly surpass the quality of the equivalent CBR encode. The reason being is that a good multi-pass encoder will allocate more bits of data where required and less when not needed, thus improving overall image quality. Of course if your CBR encode is at the same data rate as the MAX in a VBR encode, this argument is not valid and then the only advantage of VBR is smaller filesize.

As for how low TmpGenc takes to do VBR, it depends on what VBR method you choose. CQ takes a little longer than CBR (perhaps 50% more or so, havn't actually tested it myself) whereas 2-pass VBR will take at least twice as long, probably a little more.

Hope this helps.

BTW, there are already several threads, some quite long, discussing the various merits of VBR, CBR, CQ etc on these forums. If you really want a lot of opinions, some quite strong, try the search button!

[solarfox]

Either, or both, actually -- it all depends on your source material and your settings. By shifting to lower bit rates during frames that don't have much change between them and don't need the full bandwidth to encode them, the encoder can ultimately produce a smaller file. (A good example of this kind of scene would be a few seconds of a camera pointed at a building -- known as an "establishing shot" in cinematography -- to clue the viewer in to where the following scene is taking place. Establishing shots typically have little motion in them -- in fact, some low-budget productions don't even point the camera at the actual location; they just use a still photograph and zoom in or pan across v-e-r-y s-l-o-w-l-y to make it look like a live shot. )

On the other hand, since the encoder can typically produce a smaller file using VBR than it would be able to with CBR, you can choose a higher VBR bit rate and wind up with a file roughly the same size as a lower-rate CBR, but with higher quality since the encoder now has more bandwidth to work in when the scene demands it.

[jkl]

most people who use cbr are more worried about the time it takes to encode instead of worring about the final quality.

[DJRumpy]

The two aren't mutually exclusive. When encoding to DVD, I always use 3-Pass VBR, as it allows me to raise the bitrate, since the file produced is smaller, while still fitting the project on a single DVD.

[SingSing]

How much longer does a "3 pass VBR" vs
a CBR for encoding ?

I assume VBR quality produce a result when it's
max bitrate vs CBR with that same bitrate.

[bugster]

SingSing wrote:

would like to know why VBR ? Better space and/or Quality ?
and how much longer does it take to TMPGenc the same movie ?

and he also wrote:

How much longer does a "3 pass VBR" vs
a CBR for encoding ?

"3 pass VBR" is not supported under Tmpgenc, DJRumpy is probably using CCE, so a direct comparison of encode times here is meaningless. TmpGenc does support 2-pass VBR and it takes a little over twice the time as it does for a CBR encode. Double, for the two passes and a little extra for the VBR.

[DJRumpy]

Bugster is correct, I'm using CCE. There is no compaing of times between CCE, and TMPGenc. TMPGenc takes about 4 to 5 times longer to encode anything. A single SVCD disk in CCE, takes about 20 minutes to encode (CBR) (PIV 1.5 Mhz). I've never timed each pass, but I suspect that they take about the same time. I've seen alot of disk activity, especially on the last pass in CCE. The first pass doesn't even write an MPG, or M2V.