Alt cc and short videos - any point?

[zettai]

This is a general question which I've always been curious about but still can't determine with testing, so maybe someone else could shed light on this.

How significant will the effects of an Alt CC be with something that is only, say, 3 minutes long? I understand the theory behind the alt cc but in terms of agression exactly how fast is fast? Is it the kind of thing that only really benefits full movies interspersed with long scenes.

What I encode the most is music videos, which often have lots of fast scenes but ocassionally have a mix of fast and slow cuts. However, a "slow" cut in a music video isn't anything like a long scene in a movie - a slow cut in a music video lasts about 5 to 10 seconds, a fast cut can be a couple of frames.

I've been encoding with different methods for a while but it's often difficult to tell just by seeing the results. So, does anyone familiar with the code know whether alt cc settings are beneficial *in theory* with short videos or are the videos too short to show any effect?

[duartix]

Man, do I know what you are feeling!

Unless you are ripping a whole DVD of music videos, I have a very good advice I wish someone would have given me when I was encoding music videos.

If you are encoding isolated clips to fill up CDs, there is only one encoding that makes any sense: encode at 1 pass Constant Quality like Quantizer 3 for example.
This way is the only way to make sure all your videos have good quality and the same across them all. You aren't limited by filesize since it's not 1 or 2 videos that are going to fill up a CD.

Why even bother with 2 passes? Any 2 pass results would be worse in terms of quality! The worse that could happen is for you to have a few xx MB remaining on a CD until there comes a video that compresses to under that size, and there, your problem is solved...

Anyway that is the principle of 2 pass encoding: to allow all the frames to have the same quality with the constraint of filesize.

[zettai]

That's great if you are doing what you are doing.... but I'm not.

All my videos are designed for online distribution so I need to squeeze as much out of a codec as I possibly can in order to keep my filesizes small enough so they can be hosted but still maintain excellent quality. I try to aim for 70% first pass but I dont think anyone would download a 110mb 4 min video because thats how big some of them can be with cq 2.

Lots of effects = lots of bits.

So I have to use 2 pass, which is why I'm curious about the two bitrate compression philosophies.

[rui]

Well, i mostly test by encoding a trailer (little over 2 minute video).

If you read the thread about koepi's stats reader, the latest experiments are that you should encode using default cc at high/low 0%, and not use HME. Let the codec decide the curve compression for its own.
With this method your encodes will have lower quantizers all around.
By using alt cc and messing with it, you will probably be able to lower average quantizer, but end up with some high absolute quantizers (at least in my testings is what happens), and according to koepi, who knows about this things much more than we, this is bad (see that thread to understand why).

[zettai]

yeah, thanks rui - I've been keeping a close eye on all the curve compression threads.

[duartix]

I still get the feeling no one gets my point:( :( :( ...
If you are not constrained by filesize why use 2 passes?
I just don't understand.
Of course you are not going to encode them at cq2.
Ever tried cq3 or cq4?
It's quick, give it a try!!!
Or are you thinking "Now I'm going to encode a Depeche Mode video, I'll use 20 MB, next I'm going to encode a Rammstein video I'll use 20 MB."
It doesn't make any sense!
You could end up with the first at average QT 2.5 while the second at average QT 4.5 and thus completely different qualities.

You want small? Use bigger quantizers like 3 or 4. Use 1 pass quality 40% or 50% or 60%. Isn't that the whole idea?

[zettai]

Originally posted by duartix
I still get the feeling no one gets my point:( :( :( ...
If you are not constrained by filesize why use 2 passes?


No, you don't get my point - I AM CONSTRAINED BY FILESIZE!

I've already said that my files are made for online distribution so I need to have them fit within certain limits in order to get them hosted. So, obviously, what I want to do is to get the maximum quality out of the filesize limit that is available to me.

This is why I want to use two pass, see?

[MoonWalker]

Or you can use linear scaling by putting in the default cc high/low : 0/0

MoonWalker

[duartix]

@zettai:No, you don't get my point - I AM CONSTRAINED BY FILESIZE! Well, I deserved that.:o :o :o
I'm sorry, it's just that when you sad you had to maintain the filesizes small I though about bandwidth limitations, not filesize limitations.

[Stabmaster-Arson]

2 pass is always better than 1 pass, always. better and smaller. And compared to a file the same size as a cq 1 pass, the 2 pass will be better.

1 pass is for the lazy man. When I encode, I want the best possible use of all my bits no matter how large want the end video.

You could make a decent 100 MB video with 1 pass, but goddamn it could be that much better @ 100 MB and 2 passes.


you wont see any hollywoor DVD's using CBR........

hehe