Hi everybody! My question is the following: is a pre-pass with vqscale=2 a good indicator of the compressibility of a file? I have a collection of episodes 13 that I want to backup to a DVD+R. Obviously, I could simply calculate 4.38GB / 13, and have each episode use the same bitrate. However, I expect that some will require a higher bitrate than others. Hence the need for some sort of "compressibility measure": I could just do a pre-pass with vqscale=2, and use the resulting file size to calculate the proportion of the available space which should be given to the episode. Thanks in advance for yor time! Cheers, Jean __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com
Hi, On 4/27/05, Jean Hoderd <jhoderd@yahoo.com> wrote:
Hi everybody!
My question is the following: is a pre-pass with vqscale=2 a good indicator of the compressibility of a file?
I have a collection of episodes 13 that I want to backup to a DVD+R. Obviously, I could simply calculate 4.38GB / 13, and have each episode use the same bitrate. However, I expect that some will require a higher bitrate than others. Hence the need for some sort of "compressibility measure": I could just do a pre-pass with vqscale=2, and use the resulting file size to calculate the proportion of the available space which should be given to the episode.
My advice would be to encode 10 equally long (e.g. 0.5% of the movie) sequences distributed at 10 different moments of your movie, all at quant 2. You want those sequences to add up to around 5 % of the movie duration. Then, you smash those together and look at their size, which would be a good enough indicator of its compressibility. In short: you don't need a full pre-pass, just snippets, to get an idea of how much each episode needs. Guillaume
Guillaume POIRIER wrote:
My advice would be to encode 10 equally long (e.g. 0.5% of the movie) sequences distributed at 10 different moments of your movie, all at quant 2. You want those sequences to add up to around 5 % of the movie duration.
Guillaume
it seems that vqscale=2 doesn't indicate anymore "best quality", maybe since Michael introduced Lagrange multipliers. Using vbitrate=9999 or vqscale=0.01 always raises the bitrate by a large amount. Michael can you clarify this, please?
On Wed, Apr 27, 2005 at 04:33:46PM +0200, Nico Sabbi wrote:
Guillaume POIRIER wrote:
My advice would be to encode 10 equally long (e.g. 0.5% of the movie) sequences distributed at 10 different moments of your movie, all at quant 2. You want those sequences to add up to around 5 % of the movie duration.
Guillaume
it seems that vqscale=2 doesn't indicate anymore "best quality", maybe since Michael introduced Lagrange multipliers. Using vbitrate=9999 or vqscale=0.01 always raises the bitrate by a large amount. Michael can you clarify this, please?
vqscale=2 has NEVER meant best quality. qp ranges from 1-31 for mpeg formats. Rich
participants (4)
-
Guillaume POIRIER -
Jean Hoderd -
Nico Sabbi -
Rich Felker