Is discarded and not full quality first pass average bitrate accurate?

[LightningFire]

If the first pass encode-xvid encoding dialog box-shows a certain average bitrate, with settings for first pass:
- discard first pass checked
- full quality first pass unchecked
can I then use this bitrate value as a measure for a good quality target bitrate, on the condition that min I and P-frame quant is 2 (no zones set in either pass.)*, and all other settings are left as they were during first pass.

I could of course just use target quant=2 in second pass instead of target bitrate, but I am not likely to do that, for no particular reason.

* I used to set min I-frame and P-frame quant to 1, but those are as stated
by Didée and Manono. (http://forum.doom9.org/showthread.php?p=550379#post550379), not used in first pass, unless zones
are set.

Also Sharktooth wondered about this before (http://forum.doom9.org/showthread.php?p=566643#post566643).


Related to this, I wonder wether setting zones with a certain target quant will override the quantizer restrictions, go beyond the restrictions set?

Greetings.

[Teegedeck]

can I then use this bitrate value as a measure for a good quality target bitrate, on the condition that min I and P-frame quant is 2 (no zones set in either pass.)*, and all other settings are left as they were during first pass.

I could of course just use target quant=2 in second pass instead of target bitrate, but I am not likely to do that, for no particular reason.The bitrate achieved in first pass is considerably higher than what you get in a normal 1-pass at quant=2 because the first pass is tuned for speed, not for efficiency (provided you leave 'full quality first pass' unchecked).

So there is no reason to do a 2-pass if you want maximum quality.

[LightningFire]

Thanks.

So whith having done a not full quality first pass the codec should/will
be able to reach the same avg quant 2 quality with less bits in a full quality
second pass, but that would be a waste of time as you could have just encoded full quality one pass with avg quant 2. I guess that is just a rewording of what you stated, without mention of what efficiency measures are disabled in "not full quality" first pass, but I believe to have read that in another post already.

If I want to use a target bitrate and have a measure of the bitrate the codec
uses on a certain quality, with the only possible plus of 2nd pass being knowledge of (complexity of/bits requirement of) frames to come at that given quality,should I use zones (in first pass) and full quality first pass?

Or am I missing something about the reasons of using 2 passes?

Greetings and thanks again for the previous clarification.

[Teegedeck]

To put it another way: You will NOT be able to reach the bitrate of first pass in second pass if you have set minimun quantizer = 2.

The codec will just saturate . At quant=2 the bitrate wil be lower than first-pass-bitrate.

You will NOT have any advantages from two-pass if you set the bitrate to the average bitrate of first pass. No redistribution of bits in 2nd pass will take place (unless there is a bug) because it will not be necessary in order to reach the target.

If you use a full-quality first-pass and set your bitrate for 2nd pass to the average bitrate of that first-pass - then your second pass should be exactly the same as your first pass, with every p-frame compressed at quantizer=2.

Again: You cannot reach a better quality than a constant quantizer=2 encode. The rule is 'no size restrictions --> constant quantizer encode'.

If you are not content with that, you can try using a high-bitrate matrix. This would take much higher bitrates and make 2-pass a senseful measure again - if you have any size-restrictions at all.

[LightningFire]

To put it another way: You will NOT be able to reach the bitrate of first pass in second pass if you have set minimun quantizer = 2.

The codec will just saturate . At quant=2 the bitrate wil be lower than first-pass-bitrate.

You will NOT have any advantages from two-pass if you set the bitrate to the average bitrate of first pass. No redistribution of bits in 2nd pass will take place (unless there is a bug) because it will not be necessary in order to reach the target.

I noticed that, now I understand it as well.

If you are not content with that, you can try using a high-bitrate matrix. This would take much higher bitrates and make 2-pass a senseful measure again - if you have any size-restrictions at all.

Thanks, though I was really content with the smoothing of H.263, and with xvid that manages to cramp material encoded with that into such a small size at quant=2.

Greetings.