[MEncoder-users] Tuning hqdn3d

Nicolas George nicolas.george at normalesup.org
Wed May 20 12:08:21 CEST 2009


Le decadi 30 floréal, an CCXVII, James Hastings-Trew a écrit :
> So, you want what anyone would reasonably want - decent quality at a decent 
> size.

Yes, but you forgot the part where I emphasize that the primary concern is
the quality. Not in the sens "I want excellent quality", but in the sense "I
want this minimum quality and no less".

> (and yes, depending on resolution, the size can be well over 2Gb in size - 
> most feature films encoded at 1280 x 720, for example can be well over 6Gb 
> in size if you are going for quality)

1280×720 is better quality that what I want for most of my videos.

> However, you have to know that this isn't optimal

You still have to prove that. Unfortunately, the few blatant mistakes in
your message do not seem very convincing.

>						    - you certainly will be 
> spending bits you don't need to on scenes that don't need them, and not 
> spending bits on scenes that do.

If that is so, that means that the CRF evaluation function is poor. Do you
have any elements that indicate that?

>				   The only way to have the bits distributed 
> optimally, all other things being equal, is a 2-pass encode because the 
> encoder needs knowledge of the whole movie to be able to make those 
> decisions.

Unfortunately, that is not what 2-pass encoding do. 2-pass encoding is only
good to distribute bits to achieve a target bitrate. No global optimizations
are implemented.

> other hand quality and file size really do matter, then you need to pick a 
> bitrate that you can live with

Once again no: I want to chose the quality, not the bitrate.

> Not really. The encoder is not going to spend bits it doesn't need to. I 
> can ask for a bitrate of 12000 but if the movie only needs 6000 to get the 
> job done with the encoder features I've requested, then that's all that is 
> going to get used.

That is blatantly wrong. If the requested bitrate is 12000 and the required
quality requires 6000, then the 6000 will be spent to get still higher
quality, including faithfully encoding the artifacts in the source material.

It may happen that the codec can really not use more bits, but with x264,
that is only when doing lossless, and the result is really huge.

The "encoder features" govern the balance between encoding and decoding time
and encoding efficiency, they do not limit quality or bitrate.

> Very realistic and reasonable (rolls eyes).

You are only proving that you are not able to read the messages: you cite
the joke, not the serious experiment.

>						Try the encodes with crf=18 and 
> then 2-pass with bitrate=6000. That would be a more even comparison.

Where does that 6000 come from? Nowhere. This experiment is exactly as
stupid as the joke I made.

>> 1. Constant-quality encoding at an arbitrary quality to get file A.
>> 2. 2-pass target bitrate encoding to get file B with the same size as file A.
>> -> Which is better, file A or file B?
> File B will be better. That's what we've unsuccessfully been trying to tell 
> you.

I have experimental results _and_ mathematical arguments that say that
file A is better.

You have bogus arguments that say that file B is better.

I know which one I will trust.

Regards,

-- 
  Nicolas George


More information about the MEncoder-users mailing list