DaveEL
19th May 2002, 21:47
i dont know if this is practical (due to file sizes) or worthwhile (i dont know how much time doing the DCT takes up).
My idea was to store the output of DCT in a file like hinted ME data when doing a first pass and then using this file instead of the file that is sent to the codec. This would have the advantage of removing the need to do several operations twice when doing two pass encodes.
1) decode mpeg2
2) deinterlace/ivtc
3) crop/resize
4) DCT
But shouldnt lead to any loss in quality like using the output of a first pass does as both passes need to go through the DCT anyway. One thing is you still need to do is pass something of the correct length to the codec i thought this might be possible if empty first pass could be decoded to say a 1 pixel video of the correct length.
Like i said im not sure if this is worth while but i thought it might be worth mentioning. Also im not sure what order the DCT is done in on p-frames as in do they DCT first and then take the differences between blocks or do they take the difference between blocks and then DCT that if the latter is the case i guess it could only be done when using hinted ME.
DaveEL
My idea was to store the output of DCT in a file like hinted ME data when doing a first pass and then using this file instead of the file that is sent to the codec. This would have the advantage of removing the need to do several operations twice when doing two pass encodes.
1) decode mpeg2
2) deinterlace/ivtc
3) crop/resize
4) DCT
But shouldnt lead to any loss in quality like using the output of a first pass does as both passes need to go through the DCT anyway. One thing is you still need to do is pass something of the correct length to the codec i thought this might be possible if empty first pass could be decoded to say a 1 pixel video of the correct length.
Like i said im not sure if this is worth while but i thought it might be worth mentioning. Also im not sure what order the DCT is done in on p-frames as in do they DCT first and then take the differences between blocks or do they take the difference between blocks and then DCT that if the latter is the case i guess it could only be done when using hinted ME.
DaveEL