Matthew
29th April 2004, 03:41
Am I correct in saying this bitrate calculator uses a fixed value for overhead (well, at least video overhead) and simply subtracts the size of audio streams and menus etc?
Taking the case of a simple M2V+AC3, the following formula (which I have use personally) produces average bitrate figures which are only a few kbit/sec off DoItFast4U's, regardless of length (30 mins or 240 mins = under 2 meg difference in video stream size).
(4575000000-SUBS/MENUS)/((FRAMES)/FPS)/1000*8)-AUDIOBITRATE
The figure of 4575000000 was designed with a 180 min movie in mind so of course the result of this is undersizing for sub-180 min movies (while a 4 hour movie would be oversized).
mmmm...DTS seems to yield a different result though, which indicates audio overhead is taken into account at least for DTS.
Anyway my point is merely, if this is the case, it might be an idea to introduce differential fixed video overhead figures, e.g. if movie>90 mins use XYZ fixed figure, if movie>100 mins use ABC fixed figure, etc. The practical difference is a bit slim I suppose, but meh...
Taking the case of a simple M2V+AC3, the following formula (which I have use personally) produces average bitrate figures which are only a few kbit/sec off DoItFast4U's, regardless of length (30 mins or 240 mins = under 2 meg difference in video stream size).
(4575000000-SUBS/MENUS)/((FRAMES)/FPS)/1000*8)-AUDIOBITRATE
The figure of 4575000000 was designed with a 180 min movie in mind so of course the result of this is undersizing for sub-180 min movies (while a 4 hour movie would be oversized).
mmmm...DTS seems to yield a different result though, which indicates audio overhead is taken into account at least for DTS.
Anyway my point is merely, if this is the case, it might be an idea to introduce differential fixed video overhead figures, e.g. if movie>90 mins use XYZ fixed figure, if movie>100 mins use ABC fixed figure, etc. The practical difference is a bit slim I suppose, but meh...