Questoin about filters used in 1st and 2nd pass...

[FreierGeist]

ive heard that we have have to use in the second pass exactly the same filters we used in the 1st pass
and i wanna be enlightened about that, pliz
is that true?
why it has to be that way?

thank you all!
;)

[MxxCon]

becuase if you use differnt filters information you generated in 1st pass will not be valid for 2nd pass.

for example if you use some noise removing filter for only 1st pass your compression ratio info will be alot higher then for 2nd pass and you will not get desired filesize.

perhaps you should read doom9's SBC and/or gknot guides to understand how 2pass encoding works.

[manono]

Hi-Not to muddy the waters, but this interests me as well. Of course, if you could just do the first pass with out smoothing-sharpening filters, and then throw them on just for the second pass, a lot of time would be saved overall. If the first pass is done to determine how to distribute bits for the second pass based on motion, luminance, etc., then that aspect shouldn't change if the filters are thrown on just for the second pass (someone correct me if I'm wrong).

So I did some experiments-first and second pass with filters, and then first pass with no filters, second pass with filters. Just comparing the 2 different results, I can't tell any difference. The DivX Quality is pretty much the same as well. As MxxCon stated, there can be a difference in file size (I've found differences of up to 1-2%), but if you put on the same combination of filters at the same settings (as, for example, an anime series I'm working on now), then the file size-bitrate can be adjusted before running the second pass.

I've done this using 2D Cleaner and Xsharpen-perhaps other combination of filters would lead to different results. Anyone else experimented with doing it this way, and have you found any drawbacks besides smaller file sizes?

[FreierGeist]

thanks for the replys!
@ manono: thatas what i have in mind...
putting some settings on the first pass as parameters for the second and increasing some other main setting at the second (even with the limitations of this operation)

[]s