[MEncoder-users] Tuning hqdn3d

[Andrew Berg]

RC wrote:
> Unless there's some limitation with x264 I don't know about, you should
> be able to use the CRF pass as the first of the 2-passes without any
> quality loss.  Of course "turbo" might mess this up just a bit, but
> you'll simply have to try a few to figure out how far CRF needs to be
> adjusted to compensate.
>   
Turbo messes it up a lot. With a regular CRF encode, it says it's around
600kb/s. If I make it a first pass with turbo=2, it's 2x as fast (the
man page says it should be up to 4x as fast) and says it's about 1000 kb/s.
> Personally, I think this is all pedantic...  Just because you want a
> certain level of visual quality doesn't mean you can't get very close to
> it, without wasting many bits, by just picking a bitrate and using it.
>   
I'm not psychic.
> Take a look at all the videos you've encoded with the same CRF settings,
> and I'm willing to bet the vast majority will have a pretty similar
> bitrate.  There will always be a few outliers, but I'm not sure I'd
> trust the CRF equations to be more accurate, more of the time, than a
> bitrate (just MHO).
>   
The problem is that I don't know what the bitrate would be for a certain
encode, and it's not too important to me anyway. Sure, I could do a CRF
encode to find out, but that rate only applies to a limited number of
videos. For example, a good bitrate for the Season 1 DVD set of Married
With Children won't apply to the Season 7 set because even though they
are very similar in content (live action, not much fast motion and so
on), the quality of the Season 7 set is going to be higher. And then
there are other things with different content and different quality.

Trying to achieve something with 2-pass that CRF is meant for is just
plain silly. I would have to see some serious improvements in the
filters with 2-pass because target quality is my goal.