1 pass, 2 pass, ect.

[edu]

In encoding, what does "pass" mean or do? Will that be analogous to paint coat, i.e., two coats = double the density and thus the quality? Thanks.

[jimmalenko]

Sort of. But just because you use 2-Pass, doesn't mean your video will necessarily be better than a 1-pass method - it's all in the settings you use. The most common encoding algorithms are 1-Pass (CBR), quality-based 1-Pass (CQ), and then n-pass VBR.


1-Pass (CBR) attempts to assign a set bitrate to each and every second of video, irregardless of if it actually needs that particular bitrate to suitably represent the video. Pros are that you can almost guarantee quality (provided you use a high enough bitrate), and you can easily calculate the final size. Time taken to do the encode will be reasonably fast. Cons are that in order to suitably represent the whole video, you need to set the bitrate high enough to cope with your most dramatic, fast-action scene, which means that a lot of scenes will receive more bitrate than they require - this isn't an issue with short recordings ( ~ 1 hour or so), but does make the file size quite large and so can affect longer videos in terms of fitting material to one disc. Probably the best to use if quality is your major concern, but filesize is of no consequence.

1-Pass (CQ) attempts to use a variable bitrate between given boundaries, and depending on the quality setting, will attempt to assign just enough bitrate to suitably represent the second of video in question. Pros are that you can guarantee quality (provided your bounds and your Q value are appropriately set), and time taken will be reasonably fast. Cons are that filesize can vary greatly, depending on the content of your video. This can make it hard to judge / calculate beforehand what settings to use with longer videos. One pass means the encoder is not aware of the bitrate required for scenes following any given scene also. Probably the best to use if quality is your major concern, and filesize is of lesser concern.

n-Pass VBR uses n-1 passes through the video to determine the best bitrate to give each second of video between given boundaries and a suggested average value. It then does the actual encode on the nth pass. Pros are that you can guarantee quality (provided your bounds and your average are appropriately set), and file size can be easily calculated. The encoder can use n-1 passes to look at the content of the entire video and refine it's estimations, knowing what scenes are to come. Cons are that time taken will be n times longer than either of the 1-Pass methods. Probably the best method to use if quality AND hitting a given file size are priorities.

[edu]

Thanks. Now I have an idea of what to choose in my next encoding.

[whitejremiah]

im probably speaking for most of the people HERE at least...most of us GENERALLY use nth pass, i do and i normally do at least 3 passes...quality turns out great...but the encoding time is a pain in the neck....using nth pass does increase the quality over time, but like i said, its a long process compared to using a 1 pass method...its all dependant if speed is your priority or if quality is...i'm personally of the mind that i dont care if i have to leave my pc running for a week straight to encode a video, if the output looks nice enough, i'd rather have it take a while for the encode than to have it look bad and have it done in ten min. Also, if I'm using divx for my encodes, i crank all the settings up for quality...it takes a while more, but the output ends up looking nearly indistinguishable from the original source. (and a LOT smaller, too)

[Soopafresh]

Good explanation, jimmalenko.