Bit depth required for colorspace
Posted: 2014-04-10T11:02:49-07:00
Dear reader,
I've often read something like "use 16 bit depth with ProPhotoRGB to avoid color banding".
Well, my question was "how more worse than sRGB is ProPhotoRGB at 8 bit"?
here the short answer:
AdobeRGB1998.icc: max CIEΔE2000 1.644989 at <29.882812 29.882812 31.875000>, direction <0 1 0>
BetaRGB.icc: max CIEΔE2000 1.903843 at <29.882812 29.882812 31.875000>, direction <0 1 0>
ProPhotoRGB.icc: max CIEΔE2000 2.866776 at <17.929688 17.929688 19.921875>, direction <0 1 0>
sRGB.icc: max CIEΔE2000 1.168979 at <23.906250 23.906250 25.898438>, direction <0 1 0>
The values in <r g b> are of range 0..255
ProPhotoRGB's largest color granularity is about 2.5 times larger than sRGB, which would require 1.3 bit more Information than sRGB.
(I did not do a full exhaustive search, just 1/8 of all possible <r g b> values)
Here the C source ( cat ... | bzip2 -d -c | base64 -d ):
I've often read something like "use 16 bit depth with ProPhotoRGB to avoid color banding".
Well, my question was "how more worse than sRGB is ProPhotoRGB at 8 bit"?
here the short answer:
AdobeRGB1998.icc: max CIEΔE2000 1.644989 at <29.882812 29.882812 31.875000>, direction <0 1 0>
BetaRGB.icc: max CIEΔE2000 1.903843 at <29.882812 29.882812 31.875000>, direction <0 1 0>
ProPhotoRGB.icc: max CIEΔE2000 2.866776 at <17.929688 17.929688 19.921875>, direction <0 1 0>
sRGB.icc: max CIEΔE2000 1.168979 at <23.906250 23.906250 25.898438>, direction <0 1 0>
The values in <r g b> are of range 0..255
ProPhotoRGB's largest color granularity is about 2.5 times larger than sRGB, which would require 1.3 bit more Information than sRGB.
(I did not do a full exhaustive search, just 1/8 of all possible <r g b> values)
Here the C source ( cat ... | bzip2 -d -c | base64 -d ):
Code: Select all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