bitrate.log and one pass quantizer

andy

is there some way to use the bitrate.log file produced by the divx codec to determine a quantizer number that would yeild a one-pass quality-based encode of approx equivilent bitrate.

why?!?

Possibly, but why not just run a 2-Pass encoding?

ok just for example: you have a series of episodes or a movie trilogy or something. you want to encode all episodes with the same constant quality for uniformity. but you dont want any one episode to exceed xxx mb. you figure out the quantizer that will fit the longest episode into xxx mb and encode all of them (one pass each) with that quant. it saves a lot of time compared to doing two passes each and they all look the exact same quality.

anyways. i have been using educated guesses and trial and error to determine this value, but it would be easier to process the bitrate.log file.

i have always wondered about this: if a two pass encode is truly constant quality then why produce complex bitrate data for the second pass? why not use the first pass to simply determine the appropriate quant number (instead of an 8 mb file), and then feed that number into the second pass. it just seems so simple.

for instance with X-Men 2 i have determined that a quantizer of 3.18 will produce an appropriate bitrate to fit it on one cd at a resolution of 448 x 336 x 23.976

/divx.

thank you jggimi. i think i understand what you are saying. i did now read the parts of the divx manual that pertain to this and i understand that quatizer is an integer value for each frame and the divx encoder allows you to choose an AVERAGE quant for encoding 1-pass quality-based

this being said, is it not true that this average quant value does directly correlate to quality, and that if i encode a whole series of videos at 3.18 average quant they will all use whatever bitrate necessary to mantain a fixed overall quality, in turn making them all look uniform in appearance from start to finish?

and if i choose my average quant correctly, i can fit whatever im encoding into xxx mb effectively?

and if this is the case then it would only take an algorithm one pass to look thru a video and determine the maximum average quat would allow the video to stay below a limit of xxx mb. and this average value could be used in the second pass to encode the video to the desired size.

Don't confuse bitrate with quantizer.  They're different metrics.  From the Guide (page 44):...when you fix the quantizer it is not possible to set or predict the bitrate and hence file size because the encoder will always spend as many bits as required to encode each frame at the given quantizer.

And according to to guide, regarding your situation and using 1-pass quality based mode (page 36):When working from static sources (for example re-processing existing stored video) you will achieve more consistent quality by using Multipass mode.

Yes, you can use a floating point quantizer value (from 1.0 to 31.0), and I expect that when non-integer, the quantizers must in some way be quot;averagedquot; using modulation.  I've never chosen a non-integer when doing fixed quantizer based encoding, so have never looked into it (and the Guide doesn't say).