[MEncoder-users] Tuning hqdn3d

[James Hastings-Trew]

Nicolas George wrote:
> Le nonidi 29 floréal, an CCXVII, James Hastings-Trew a écrit :
>   
>> vmrsss wrote:
>>     
>>> On 15 May 2009, at 16:15, Phil Ehrens wrote:
>>>       
>>>> Look, people... 2-pass just plain RULEZ when it
>>>> comes to quality. If you don't believe it, do the
>>>> friggin' experiment for yourself...
>>>>         
>>> it's not that I don't believe that 2-pass is better
>>> than 1-pass. I don't believe that 2-pass is better
>>> than crf if the encodes end up with the same
>>> avg bitrate.
>>>       
>> I believe that you are incorrect. What you are saying is that 2-pass can't 
>> possibly give you a better quality encode than doing it in one pass with a  
>> chosen crf value that you choose.
>>     
>
> 2-pass for constant-quality could allow encoding improvements, but it is not
> implemented.
>
>   
>> 				    My experience with doing encodes both 
>> ways shows that the 2-pass encode is visually better in quality, and, as a 
>>     
>
> As I said, I did a very quick experiment a few days ago, and the result was
> rather in the opposite direction.
>
>   
>> bonus, the file is not arbitrarily large.
>>     
>
> I think you missed the point where some people (including me) explained that
> what they want is a file that is large enough to achieve the visual quality
> they want, and no more.
>
> If the file is smaller, then it does not achieve the requested visual
> quality: this is not a bonus.
>   
If file size is of no concern, then why compress the video at all? 
Maximal quality will be acheived with a lossless codec such as ffvhuff. 
Your experiment of a couple of days ago was unrealistic, and you know 
it. Additionally, I was not talking about doing 2-pass at constant rate 
factor - that would be silly. But I suspect you knew that too. I will 
not feed this particular troll anymore.


__________ Information from ESET NOD32 Antivirus, version of virus signature database 4083 (20090518) __________

The message was checked by ESET NOD32 Antivirus.

http://www.eset.com