kilg0r3
24th March 2004, 14:47
I recently did some SVCD encodes using the DVD2SVCD - RoBa Method (http://forum.doom9.org/showthread.php?s=&threadid=64886), which considerably shortens the encoding time. I think that a similar functionality would be very useful inside producer. It speeds up encoding, with a minimal impact on the accuracy of file-size prediction and resulting quality (only in the case of an undersized file).
The RoBa plugin essentially substitutes the analysis pass(es) of cce with a sequence of compressibility tests at various fixed quality settings. With compressibility test I mean the procedure as it is done in Gordian Knot where the program analyzes several sets of frames that together make up about some 2-5% of the movie. Both GordianKnot and RoBa do this by means of an avisynth script containing the SelectRangeEvery() function.
As I said, RobA does this several times. Yet, up to now I've never counted more than five of these partial first passes. After such a pass it calculates the approximate size that a full pass at the selected quality setting would have had. This value serves as a basis for the selection of the settings for next analysis pass etc. This process is repeated until a quality setting delivers a predicted file size that is very close to the filesize the user aims for.
Please note that, the description above comprises only that part of the functionality of RoBa , which is the most important for me.
Thanks for Your patience, kilg0r3
BTW. I'd like to 'advertise' this functionality method in other sections of the forum (e.g. Vdubmod, GordianKnot, Xvid), yetI don't want to be striked for cross posting. Any suggestion from moderator's side?
The RoBa plugin essentially substitutes the analysis pass(es) of cce with a sequence of compressibility tests at various fixed quality settings. With compressibility test I mean the procedure as it is done in Gordian Knot where the program analyzes several sets of frames that together make up about some 2-5% of the movie. Both GordianKnot and RoBa do this by means of an avisynth script containing the SelectRangeEvery() function.
As I said, RobA does this several times. Yet, up to now I've never counted more than five of these partial first passes. After such a pass it calculates the approximate size that a full pass at the selected quality setting would have had. This value serves as a basis for the selection of the settings for next analysis pass etc. This process is repeated until a quality setting delivers a predicted file size that is very close to the filesize the user aims for.
Please note that, the description above comprises only that part of the functionality of RoBa , which is the most important for me.
Thanks for Your patience, kilg0r3
BTW. I'd like to 'advertise' this functionality method in other sections of the forum (e.g. Vdubmod, GordianKnot, Xvid), yetI don't want to be striked for cross posting. Any suggestion from moderator's side?