Can the "q. level" figure in bitrate viewer be used as a reliable guide of quality?

[Matthew]

I'm not sure this question belongs in this forum, but as the old dvd encoding forum is no longer around I thought I'd try here ;)

After re-encoding a DVD with CCE (4 pass, bitrate fairly high, e.g. 3800 in current case) is there any point in loading the mpv in bitrate viewer and having a look at the min/max/average q values? And if this is worthwhile, how should the figures be interpreted?

Thanks :)

[auenf]

the dvd encoding forum was moved and is now called "other mpeg1/2 encoders"

basically under 9Q means there shouldnt be any macroblocking, but there will be some jpeg style artifacts on edges closer to 9Q, not sure at which Q they generally dissapear.

Enf...

[Matthew]

Thanks for the reply :)

Sounds like an average of 5.05 and peak of 8.62 is okay then.

The figures on the original (complete) video stream are 3.38 and 9.56.

[RB]

Alternatively, click the "Advanced..." button in the CCE encode settings and switch to the Quantizer scale display (top right). It will tell you the min./avg./max. quantization factors at the bottom.

[Matthew]

I did try that before, unfortunately the figures were not accurate - and different from that reported by bitrate viewer. e.g. the max and average were the same number, 4.61

Scrolling through, the values on a GOP-by-GOP basis seemed accurate, but it's nice to know the overall figures.

[RB]

Originally posted by Matthew
I did try that before, unfortunately the figures were not accurate - and different from that reported by bitrate viewer. e.g. the max and average were the same number, 4.61

Well, one can certainly question the accuracy of BV :) Anyway, I think CCE knows better about it's own work than any other 3rd party tool...

[Matthew]

Yeah in the abstract I certainly would trust CCE more than BV because it is CCE's own work, but for that encode I tried it with, the fact the max and average Q were the same (which is not possible) led me to question the aggregate stats.

However, with another encode I just did, the max Q is correct. The max bitrate isn't though, but when I go to the GOP where CCE says the max is located, it is indeed the bitrate highpoint.

So there is a bit of buginess there, but it probably doesn't affect the validity of the "average" figures though.

[mb1]

CCE shows its values on a per GOP basis whereas Bitrateviewer does it per second. That's the difference.

It is only possible to compare quantization values within the same encoder and within the same filtering.

Example:
Normally lower values mean better quantization which should mean better quality.
If I set a strong blur values become much lower because a lot of details get lost. Thus the video can be compressed much stronger (stronger quantization). In fact the video looks very unsharp and therefore bad (but there are no longer any sort of blockings :D )