PTKen
2009-07-02 12:35:35 UTC
Hello. I have a question regarding matching digital captures of
artwork to the original art. Before I ask the question, I’d like to
give some background information.
When we color correct our digital captures of artwork to the original
artwork, our viewing environment is setup as described in ISO 3664 and
ISO 12646. We are using viewing condition P2 (500 lux) instead of
viewing condition P1 (2000 lux). P2 is meant to emulate typical room
viewing light levels and P1 is meant for making critical color matches
between two copies. The ISO doc states that P2 is not appropriate for
comparison of two copies (i.e. original and reproduction), but due to
conservation concerns, we are not permitted to use the illuminance
level of P1 since this is considered too much light, and is
potentially harmful to the artwork.
When we feel we have achieved a good color match (on screen and/or
print), this is what I consider an original-referred rendering (though
it’s not strictly original-referred since we are not making any
absolute measurements). This makes for a good color match when
comparing directly to the original, but it does not make a good
reproduction because it tends to be perceived as too dark and too flat
when viewed on it’s own (i.e. not compared directly to the original).
So, we next adjust the tone curves to make a more pleasing
reproduction for print and web (output-referred rendering). Lately we
have been using the “Exposure…” adjustment layer in Photoshop to alter
the white point, black point, and gamma of the luminance channel.
Finally my question!
The problem is how to achieve the output-referred rendering in a more
systematic way. Right now our correction is fairly random and based on
individual preferences and whims. One option that my boss has
suggested is to dim the light box that we view the printed
reproduction under to signficantly “less than” 500 lux while making
color corrections and comparisons to the original. His reasoning is
that when we view one of our “output-referred prints” in a dimmed
light box, we get a visual match to the original artwork (which is in
the P2 viewing condition). Therefore, he concludes that if we do our
color corrections in a dimmed light box (say 200 lux), then when we
raise the illuminance back to 500 lux we should have an output-
referred reproduction that is not dark and flat (because we will make
it lighter and more contrasty to compensate for the dim illumination).
This reproduction will naturally not be considered a visual match to
the original—which is okay since we are trying to make an output-
referred reproduction.
I don’t agree that this is a good method. What are your thoughts?
(I’ll post why I don’t think this is a good method after I hear some
responses—I want to see if others come up with the same reasons I did
independently, or if others think this might actually work.)
Are there any other suggestions for how to re-render the images for
output? I’m having difficulty coming up with a good solution in part
due to the restrictions on the light level. I don’t believe this is an
issue of ICC rendering intent because the image is being color
corrected while comparing directly to the artwork. Therefore the tone
range that is encoded by this method is just not appropriate for
reproduction.
I hope this posting has been clear, but if it is not, please ask me to
clarify. Thank you in advance for your input!
--
Ken Fleisher
Photographer
Imaging & Visual Services
National Gallery of Art
Washington, D.C.
k-***@removethis.nga.gov
artwork to the original art. Before I ask the question, I’d like to
give some background information.
When we color correct our digital captures of artwork to the original
artwork, our viewing environment is setup as described in ISO 3664 and
ISO 12646. We are using viewing condition P2 (500 lux) instead of
viewing condition P1 (2000 lux). P2 is meant to emulate typical room
viewing light levels and P1 is meant for making critical color matches
between two copies. The ISO doc states that P2 is not appropriate for
comparison of two copies (i.e. original and reproduction), but due to
conservation concerns, we are not permitted to use the illuminance
level of P1 since this is considered too much light, and is
potentially harmful to the artwork.
When we feel we have achieved a good color match (on screen and/or
print), this is what I consider an original-referred rendering (though
it’s not strictly original-referred since we are not making any
absolute measurements). This makes for a good color match when
comparing directly to the original, but it does not make a good
reproduction because it tends to be perceived as too dark and too flat
when viewed on it’s own (i.e. not compared directly to the original).
So, we next adjust the tone curves to make a more pleasing
reproduction for print and web (output-referred rendering). Lately we
have been using the “Exposure…” adjustment layer in Photoshop to alter
the white point, black point, and gamma of the luminance channel.
Finally my question!
The problem is how to achieve the output-referred rendering in a more
systematic way. Right now our correction is fairly random and based on
individual preferences and whims. One option that my boss has
suggested is to dim the light box that we view the printed
reproduction under to signficantly “less than” 500 lux while making
color corrections and comparisons to the original. His reasoning is
that when we view one of our “output-referred prints” in a dimmed
light box, we get a visual match to the original artwork (which is in
the P2 viewing condition). Therefore, he concludes that if we do our
color corrections in a dimmed light box (say 200 lux), then when we
raise the illuminance back to 500 lux we should have an output-
referred reproduction that is not dark and flat (because we will make
it lighter and more contrasty to compensate for the dim illumination).
This reproduction will naturally not be considered a visual match to
the original—which is okay since we are trying to make an output-
referred reproduction.
I don’t agree that this is a good method. What are your thoughts?
(I’ll post why I don’t think this is a good method after I hear some
responses—I want to see if others come up with the same reasons I did
independently, or if others think this might actually work.)
Are there any other suggestions for how to re-render the images for
output? I’m having difficulty coming up with a good solution in part
due to the restrictions on the light level. I don’t believe this is an
issue of ICC rendering intent because the image is being color
corrected while comparing directly to the artwork. Therefore the tone
range that is encoded by this method is just not appropriate for
reproduction.
I hope this posting has been clear, but if it is not, please ask me to
clarify. Thank you in advance for your input!
--
Ken Fleisher
Photographer
Imaging & Visual Services
National Gallery of Art
Washington, D.C.
k-***@removethis.nga.gov
