yaz
7th August 2003, 09:43
hi all !
i've seen that there's a new release availabe but i've not seen any note on these topics so i put my problems.
i use these functs extensively. (imho, most of the 'blocking problems' can be prevented this way.) so as to get closer to ColorYUV i tried to convert some of my older scripts using the equations given here&there(faq, manual, ...) surprisingly i found significant diffreneces in some instances. (i always checked the transformations with 'analyze') so i went on a more thorough investigation. here are some of my (re)quests occured
- tweak has a kinda 'auto-clamping' feature. i mean, if the transformation would lead outta ccir(tv) standard range, the values are clamped. a 0-255 clamping is essential with all filters affecting yuv values, but here i found a definite 'tv-scaling'. is it intentional, or does my system maulfunction?
- in the manual tweak is told to accept yuy2 only but it works with my yv12 clips. what's the case here?
- levels exhibits some more strange behav. levels(0,1,255,0,255) would do anything (according to the manual) but in fact, it does a 'pc->tv' scaling if the yuv values are outta tv scale. but levels(0,1,255,255,0) makes the 'photo-negative' wout clamping. can anyone explain it?
- afais, the manual entry of levels is that of ben's (ver1.01?), no word about yv12. does this function work properly with yv12 data?
- i can't check coloryuv transformations, cus i don't know how it calculates the new yuv values. (i've thought i knew, but i always get more or less different values than i expected) would anyone (sh0dan???) give me the exact form of the transformations? say, i'd be very greatful for a kinda y'=y(off,cont,gain,gamma) equation. if s'one would add the derivation of this params i'd be pleased for a lifetime:-) (pls, don't bother about complexity! i luv complex equations:-)
thx for any answ
y
i've seen that there's a new release availabe but i've not seen any note on these topics so i put my problems.
i use these functs extensively. (imho, most of the 'blocking problems' can be prevented this way.) so as to get closer to ColorYUV i tried to convert some of my older scripts using the equations given here&there(faq, manual, ...) surprisingly i found significant diffreneces in some instances. (i always checked the transformations with 'analyze') so i went on a more thorough investigation. here are some of my (re)quests occured
- tweak has a kinda 'auto-clamping' feature. i mean, if the transformation would lead outta ccir(tv) standard range, the values are clamped. a 0-255 clamping is essential with all filters affecting yuv values, but here i found a definite 'tv-scaling'. is it intentional, or does my system maulfunction?
- in the manual tweak is told to accept yuy2 only but it works with my yv12 clips. what's the case here?
- levels exhibits some more strange behav. levels(0,1,255,0,255) would do anything (according to the manual) but in fact, it does a 'pc->tv' scaling if the yuv values are outta tv scale. but levels(0,1,255,255,0) makes the 'photo-negative' wout clamping. can anyone explain it?
- afais, the manual entry of levels is that of ben's (ver1.01?), no word about yv12. does this function work properly with yv12 data?
- i can't check coloryuv transformations, cus i don't know how it calculates the new yuv values. (i've thought i knew, but i always get more or less different values than i expected) would anyone (sh0dan???) give me the exact form of the transformations? say, i'd be very greatful for a kinda y'=y(off,cont,gain,gamma) equation. if s'one would add the derivation of this params i'd be pleased for a lifetime:-) (pls, don't bother about complexity! i luv complex equations:-)
thx for any answ
y