cq vbr vs manual vbr - bitrate ??

[spidey]

Hey Guys,

Question in a nutshell: Manual VBR vs CQ VBR 100 %
- Bitrate prediction accuracy ? How do they differ.
- Quality - SInce both are single pass VBR, how do they differe if at all.




I have been using Manual VBR to do all my SVCD encodes. THis has been very accurate, and the results have been great. I was curious however as to how this encode's quality compares to CQ vbr.

I am very curious as to how it would be, in cq vbr, if you set it to 100 % quality, while using the figured bitrates of a CBR VBR.

Could / Would this be unpredicatable as to the output file size ???

I have been using the CBR bitrates as a guideline for a manual VBR encode.

For example, if the TMPG wizard tells you the clip can be at 2320 as it's CBR bitrate, I then manually set the template to 2320 max with a 2200 minimum range. My goal was to give the VBR a good range to work, while keeping the bitrates high for quality.

In light of this, if I used CQ VBR and made it 100 % quality, could I still use this method to figure the rates ?

Or does the 100 % quality inflate the bitrates PAST the specified maximum bitrate ?

That's where I am confused. If you set both the max and min in the cq vbr template how is it able to go past what is specified ?

Also, how does the manual VBR stack up against the cq vbr ? Is the manual VBR simply a 100 % quality single pass encode ? How does the cq differ ?

I hope this makes sense in what I am trying top compare. Thank you all as always. Any ideas / thoughts are appreciated

[spidey]

or also if anyone has some guidelines as to what % to set the CQ VBR at to match the bitrates specified for it to be reliably accurate as to the file size that'd be great.

Above all, I'd just like to know anyone's perception as to how the two modes compare, since they're both single-mode VBR encodes, I don't know how else they'd differ greatly.

Thank you.

[spidey]

<bump>

[spidey]

<bump> come on guys any info would be great. I feel that I help out on here more than enough to at least deserve some replies..............

[vitualis]

Firstly, please don't bump the thread -- especially if you've only been waiting a few hours.

Hey Guys,

Question in a nutshell: Manual VBR vs CQ VBR 100 %
- Bitrate prediction accuracy ? How do they differ.
- Quality - SInce both are single pass VBR, how do they differe if at all.

With manual VBR, you set the max, min and average bitrate and the encoder works to those parameters. You can accurate predict exactly how big the file will be.

With CQ encoding, it is still VBR but the encoder decides on the bitrate on any particular scene (within the max and min limitations) depending on the "Q" co-efficient which is related to the "quality". The encoder aims to maintain a particular quality. It is probably easier to conceptualise but you don't have control of the size of the file.

I am very curious as to how it would be, in cq vbr, if you set it to 100 % quality, while using the figured bitrates of a CBR VBR.

I'm not entirely sure what you are asking here but CQ VBR at 100% equals CBR at the max bitrate.

For example, if the TMPG wizard tells you the clip can be at 2320 as it's CBR bitrate, I then manually set the template to 2320 max with a 2200 minimum range. My goal was to give the VBR a good range to work, while keeping the bitrates high for quality.

This isn't how you should encode with VBR. VBR works best when there is a wide range for the bitrate. Thus, you max. bitrate should always be the max. bitrate possible for SVCD (approx. 2500 kbit/s). Your average bitrate should be whatever you need it to be for the particular clip. Your minimum bitrate should be theorectically 0 kbit/s. However, some encoders screw up if you set too low a min. bitrate so perhaps a figure of 500 kbit/s is appropriate.

That's where I am confused. If you set both the max and min in the cq vbr template how is it able to go past what is specified ?

CQ tries to maintain the quality of the clip -- but it can only do so within the max and min contraints. You should set the max. to 2500 kbit/s for SVCD (more if you know what you are doing but be aware that you clip is no longer SVCD compliant) and min. to some low figure. As I stated before, VBR encoding needs a wide range of bitrates to work in to be efficient.

Although the philosophy of CQ is "constant quality", realistically, it cannot be attained within the constraints of SVCD. You've hit the nail on the head and have self-realised something that many of the CQ proponents apparently haven't.

Also, how does the manual VBR stack up against the cq vbr ?

If you know what you are doing, manual VBR (1-pass) can probably be made to yield better quality than CQ -- insofar as you can better optimise your disc space. However, I can think of scenarios where CQ can look better than 1-pass VBR so this is not clear.

VBR encoding is hamstrung by 1-pass and multipass encoder will yield better results.

TMPGEnc is optimised more for CQ encoding. CCE, however, has excellent multipass VBR encoding.

I posted a fairly long post here: http://forum.vcdhelp.com/viewtopic.php?p=271526#271526

The argument between CQ and multipass VBR carries on, however, and you should probably read the entire thread.

Regards.