Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: MultiPass Lame (Read 4449 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

MultiPass Lame

The Lame scene seems to be pretty slow.  Have the developers just run out of ideas?
Well, something I would like to see that could generate new interest in the scene would be multiple pass encoding, like is done with Divx.  It seems that you should be able to more effectivly use the availiable bits to get better quality at smaller sizes.

What do you think?

MultiPass Lame

Reply #1 –
Quote
The Lame scene seems to be pretty slow.  Have the developers just run out of ideas?

I don't think it's the lack of interest.  It's the lack of time, structure, and manpower.

Quote
Well, something I would like to see that could generate new interest in the scene would be multiple pass encoding, like is done with Divx.  It seems that you should be able to more effectivly use the availiable bits to get better quality at smaller sizes.

What do you think?


This has been proposed before.  Quite frankly, I doubt it will ever happen.  If there was a renewal of interest in LAME development, there are certainly other areas which could be worked on which would provide much more fundamental improvements than 2-pass encoding might possibly provide.

MultiPass Lame

Reply #2 –
Quote
than 2-pass encoding might possibly provide.

You have to admit, though; two-passing video makes it look considerably better at the same bitrate than a one-pass encode.

EDIT: tpyo

MultiPass Lame

Reply #3 –
I can imagine that the improvements in quality wouldn't be great for songs because of the length of the files in question. But if we consider the length of an audio track for a film, I’m almost certain that a 2-pass mode would improve quality considerably.... How’s the codec going to know that a complex piece of audio will occur in 10 minutes time so it must conserve bits for it if it hasn't been analysed before in a previous pass?

Maybe this feature should be added primarily to improve the quality of ABR/VBR sound tracks in movies? Maybe we should run a poll?

Edit: Typos

MultiPass Lame

Reply #4 –
Quote
Maybe this feature should be added primarily to improve the quality of ABR/VBR sound tracks in movies? Maybe we should run a poll?

What's the worth of a poll if there's noone to implement this feature?

A poll would only be of use if some developer came here and said "OK, I've got some free time, I will implement either IS or twopass. What do you guys prefer"

Unfortunately, that's not the case...

MultiPass Lame

Reply #5 –
BTW, it's worth noting that there's not much sense in comparing audio coding with video coding here.

What we call VBR, they call CQ (Constant Quality)
What they call VBR is our VBR but with upper and lower bitrate constraints, so as to fit in some kind of media or to stick to some standard (9.8Mbps for DVD, 2.7Mbps for SVCD, etc.)
And their ABR is a two pass (at least) encoding. The user sets the desired bitrate or final file size, then first the encoder analyzes the complexity of each frame, and on the second pass it distributes the bitrates accordingly so that the user settings are matched as perfectly as possible.

That is similar to Microsoft's "bitrate-managed VBR" in WMA. First pass analyzes the complexity, second one distributes bitrates accordingly.

MultiPass Lame

Reply #6 –
Another words, 2-pass VBR is needed only when floating bitrate wants but with predictable filesize. And we already have such mode, yes? In the LAME world this is ABR - isn't it?
And - will 2-pass VBR be more effective than existent ABR? In terms of quality/size ratio?
So - is there any sence to develop 2-pass VBR for LAME?

MultiPass Lame

Reply #7 –
isnt 2 pass or how many passes only really for if you have a set filesize that the encoded file needs to be under and the passes then determine where the bits are needed most?
Chaintech AV-710

MultiPass Lame

Reply #8 –
Yes, it is effectivly called ABR. When we use ABR, we are encoding the file for a target average bitrate, and hence a target filesize. This can be important especially when encoding movie soundtracks. But if you think about how ABR works, it isn't going to know how the file is structured in terms of complexitivity over time so it could be wasting bits on easily encoded passages and thus not enough bits on hard passages which could occur later on in the film. The only way for a codec to know this is to create a profile of the complexitivity using multiple passes.

rjamorim: You are right in saying theres no point in comparing audio with video, but the very very basic idea is the same. And i only suggested in a poll to see peoples opinions as to wether they think a 2 pass mode is a good idea or not, weighing up the advantages and disadvantages (a disadvantage would be developers availible and their time) and it was only a suggestion      - But i do believe that a 2-Pass mp3 codec would be a first and a great feature for film encoding and would promote lame to a wider user base... well, its only a thought    B)

Edit: Typo

MultiPass Lame

Reply #9 –
If they were optimal (and of course they're not, hence we have --alt-preset standard etc) you could just try the ten different VBR settings (i.e. V0, V1, V2, V3, etc...), and pick the one which most closely matched the bitrate you wanted for the content you were encoding!

Not the quickest job in the world, but quicker than waiting for a two-pass mode in lame! 


btw - isn't it true that ABR seems to work better than VBR at lower bitrates? 128kbps ABR really is surprisingly good. If it's not good enough for you, and you want ~128kbps, then you probably shouldn't be using mp3.

Cheers,
David.