CCE and 1pass VBR

[wulf109]

In CCE's 1 passVBR mode is there a way to predict file size?? Or is it like Tmpeg's CQ mode,completely unpredictable?
The defaults when you start CCE are 2000 minimum and 9800 maxium.
I encoded the first vob from StarWars II as a test and the .mpv file was about 550MB,indiacating that entire disk would be under 4GB's.
Do you think 1 Pass VBR produces good results for DVD production?

[Truman]

CCE comes with a multipass (1 pass) mode and a one-pass VBR mode. The first one is with *.vaf creation, so actually it is 2 pass. One-pass VBR is real one pass without the information file. That does not mean, that you can not predict the filesize. The guys of kvcd.com did not sleep and found a way to predict the filesize quite exactly.
To do this you have to encode a small test clip. The Avisynth filter SelectRangeEvery collects n frames every m seconds. So the testclip takes frames from start to end, that leads to quite an exact calculation.

[Adder_78]

One-pass VBR is real one pass without the information file.

no... CCE will create *.vaf file if you want... just enable it

read this guide and you see why you should have vaf-file!

http://www.doom9.org/mpg/cce-advanced.htm

Zzzzzzzzzzz

[Truman]

Sure, but there is no need for a *.vaf file if I just want to do a one pass encode.

[Poplar]

You can calculate file size quite accurately if you set Quality Factor=1 in One Pass VBR with CCE. The bitrate you get from any Bitrate Calculator should be the maximum bitrate in One Pass VBR setting.

[adam]

You can calculate file size quite accurately if you set Quality Factor=1 in One Pass VBR with CCE. The bitrate you get from any Bitrate Calculator should be the maximum bitrate in One Pass VBR setting.

That's because you are practically encoding in CBR. You are telling the encoder to almost never lower the bitrate, but capping it at the max you set. So it is essentially encoding in CBR at whatever max setting you use, no wonder the filesize is predictable. You can do the exact same thing in TMPGenc by using a very high CQ setting like %95 or higher. This really defeats the purpose of using VBR though because you are losing most of its benefits.

There are, however, some scripts that create samples from your source and can calculate very accurately what bitrate to use for 1-pass vbr. I know DVD2SVCD has one and there is another one available at kvcd.net.

[snowmoon]

Yes, file size prediction is possible. From my limited testing it appears to be a little less accurate than TMPGenc at this point, only because I still have not run enough samples to get a good multiplier factor.

The trick is to use AVIsynth http://www.avisynth.org/ with the mpeg2dec.dll plugin. It adds the function SelectRangeEvery(SkipStep, numberOfFrames ) so you could very easy do something like this.

AVISource("myvideo.avi")
SelectRangeEvery(1800,36)

And use that avisynth script as your video. Since we are taking 36 out of ever 1800 frames we have to multiply the output filesize by 50 ( * the correction factorsome are using .95 ).

Much more info can be found on the kvcd forums.

[Adder_78]

but why use only one-pass VBR? Easy way is to use 2-3-4 passes and get exact file size and VERY good quality...I know it takes time to encode but when you sleep it dosent really matter... 8)

Zzzzzzzzzz

[snowmoon]

1) I have a pIII700, 1 pass is signifigantly shorter encoding. I also use filters ( tv cap ).

2) For most things I have found one pass vbr to be higer quality than multi-pass since it appears to distribute the bits better. In multi-pass it appears to second guess itself way to often when doing low bandwidth stuff.

[HideOut]

IM confused (nothing new) but why would you encode for DVD-R? I use encoding for SVCD but cant you normally just copy the thing on dvd-r, assuming its small enough?
HideOut

[wulf109]

If the DVD is over 4.37GB's and you want to fit it on one DVDR,transcodeing is required.

[Adder_78]

1) I have a pIII700, 1 pass is signifigantly shorter encoding. I also use filters ( tv cap ).

2) For most things I have found one pass vbr to be higer quality than multi-pass since it appears to distribute the bits better. In multi-pass it appears to second guess itself way to often when doing low bandwidth stuff.

ok then I understand because you are using tv-capturing material to encode -> filters -> takes more encoding-time 8)

I still think that Robshot method with multipass(and some improvements) gives better results than only one pass(DVD -> DVD-r)... and the size issue is much more easy to handle

Zzzzzzzz