Donc demain soir, séance recalibrage

|
Modérateurs: Modération Forum Home-Cinéma, Modération Forum Installations, Modération Forum Univers TV, Le Bureau de l’Association HCFR • Utilisateurs parcourant ce forum: Aucun utilisateur enregistré et 2 invités
-g2.1
apres j'ai toute ma cuisine dans les pilotes ATI/ffdshow/MPC HC, mais aucun triturage forcenet des niveaux......"as good as it gets"
La configuration dans mon profil
stoub2000 a écrit:Bon je viens de refaire un calibrage, en srgb 2.1 donc. J'ai toujours le même problème sur mplayer, le noir est grisatre, sur la mire luminosité du DVD HCFR, on voit bien le rectangle noir du bas se détacher sur le fond grisatre, ce qui est anormal, si j'ai bien compris il doit se confondre avec le fond. Je suis donc obligé via ffdshow de faire un mapping 10-255 -> 0-255 pour rétablir les choses :-S
I think I've mentioned a few times on the list that there is ambiguity in applying the idea of an ideal gamma curve (ie. a power curve) to a real world device that has a non-zero black point.
You can compare an ideal power curve to a real display response, and there will be an infinite (ratio) error at the device zero. How do you modify an ideal gamma curve to accommodate a non-zero response at device zero input ?
There are two obvious ways of doing this. One is to offset the whole curve by the zero value. So the expected response becomes:
out = zout + ((1 - zout) * in ^ gamma)
Another way of doing it is to assume that there is
an offset in the input value that accounts for
the zero value (ie. cut off the lower part of
the ideal curve to match the non-zero output):
out = (zin + in * (1 - zin)) ^ gamma
where zout = (zin + zin * (1 - zin)) ^ gamma
These two approaches give subtly different curves.
If you compare one to the other they disagree.
I would suspect that the HCFR program is either
comparing to the ideal curve or the first equation
above, while Argyll dispcal uses the second equation
above.
Graeme Gill.
leeperry a écrit:arf, je voulais voir comment ARGYLL t'avais arrange ton gamma.
car j'ai l'impression que ARGYLL et Color.HCFR n'utilisent pas la meme formule
pour ton probleme de niveaux, installe ce patch et reboot :
http://exdeus.home.comcast.net/~exdeus/ati-hd2x00/
Maximum neutral error (@ 0.020845) = 1.332490 deltaE
Average neutral error = 0.230249 deltaE
> did you change anything with the gamma curves from beta7 to beta8 ?
The target for the calibration curves didn't change, no. There may have been a bug fix to the black setting level though, that may have improved the black.
Graeme Gill.
leeperry a écrit:sinon il a tjs pas l'air chaud pour offrir un gamma custom auquel on specifierait la valeur de debut et de fin...ce qui serait TOP
leeperry a écrit:arf, je voulais voir comment ARGYLL t'avais arrange ton gamma.
car j'ai l'impression que ARGYLL et Color.HCFR n'utilisent pas la meme formuleI think I've mentioned a few times on the list that there is ambiguity in applying the idea of an ideal gamma curve (ie. a power curve) to a real world device that has a non-zero black point.
You can compare an ideal power curve to a real display response, and there will be an infinite (ratio) error at the device zero. How do you modify an ideal gamma curve to accommodate a non-zero response at device zero input ?
There are two obvious ways of doing this. One is to offset the whole curve by the zero value. So the expected response becomes:
out = zout + ((1 - zout) * in ^ gamma)
Another way of doing it is to assume that there is
an offset in the input value that accounts for
the zero value (ie. cut off the lower part of
the ideal curve to match the non-zero output):
out = (zin + in * (1 - zin)) ^ gamma
where zout = (zin + zin * (1 - zin)) ^ gamma
These two approaches give subtly different curves.
If you compare one to the other they disagree.
I would suspect that the HCFR program is either
comparing to the ideal curve or the first equation
above, while Argyll dispcal uses the second equation
above.
Graeme Gill.
La configuration dans mon profil
|
|