Re: Monitor White Point Confusion
Hi Lou, What you would like to do is IMHO possible in theory, but you have to take into account the capabilities of the math model and the devices you use. If you try to mimic the actual printed piece on your screen (meaning you switch to soft-proof mode in Photoshop) you’re working in the 8-bit realm of the current graphics software and device drivers. If you’re far away from the target (eg. on a screen calibrated to a white point of 6500K) the color management engine has to perform a large correction to achieve the desired result. Measurement and profiling errors add up, and the color you see on screen will not be perfect. On the other hand, your original approach (calibrating to 5700-5800) is almost perfect, as that’s the ‘perceived’ white point of today’s OBA-stuffed coated papers under 5000K lighting. In this case, the color management engine performs a relatively small conversion on the white point, so the errors will be smaller or negligible. Until we don’t have 10 or more bit ‘deep’ device drivers and monitors (and work with 16 bit images in Photoshop), the best you can do is to calibrate to the closest perceived white. Another consideration might be to employ displays which are more conservative on the contrast; we don’t need 1:10000 for soft proofing. Kind regards, Peter Nagy
Nagy Péter wrote: Hi,
What you would like to do is IMHO possible in theory, but you have to take into account the capabilities of the math model and the devices you use. If you try to mimic the actual printed piece on your screen (meaning you switch to soft-proof mode in Photoshop) you’re working in the 8-bit realm of the current graphics software and device drivers.
It's worse than that though. AFAIK, Photoshop, ICC and the desktop conspire against using display profiles in Absolute Colorimetric intent, AKA soft proofing side by side mode. 1) As I understand it, Photoshop's proofing mode only uses a pseudo-absolute mode. The print profile is set to Absolute colorimetric while the display profile is left in Relative Colorimetric. So this gives you an impression of the white shift due to the paper color, but adapted to the white point of the display. It isn't actually attempting an absolute colorimetric match. 2) ICC V4 disables display profile Absolute Colorimetric intent by mandating the white point tag be set to D50. So a standard CMM won't render the display output with the absolute intent, ruining any attempt at an absolute colorimetric match. 3) Even ignoring 1) and 2), if your desktop and application have their own GUI elements showing, they will be rendered relative to the native display white point, and so will upset your adaptation state, and make the proofing output appear too yellow. So given all these problems, the only practical approach is to calibrate the display so that it has a white point that matches the paper white. That way a Relative Colorimetric rendering will work for a soft proof side by side comparison. Cheers, Graeme Gill.
Hi, I can recommend to read the Fogra Softproof Handbook https://www.fogra.org/index.php?menuid=626&reporeid=200&getlang=en A lot of theory with a lot of practical advise.
Am 06.03.2019 um 02:53 schrieb Graeme Gill <graeme2@argyllcms.com>:
Nagy Péter wrote:
On the other hand, your original approach (calibrating to 5700-5800) is almost perfect, as that’s the ‘perceived’ white point of today’s OBA-stuffed coated papers under 5000K lighting.
The reason for this is that most instruments which are usually used are simply inaccurate in measuring CCT. Using an accurate instrument an adjustment of the monitor WP to the same CCT as your viewing cabinet gives a visual match. But these instruments are out of budget of a normal end user.
So given all these problems, the only practical approach is to calibrate the display so that it has a white point that matches the paper white. That way a Relative Colorimetric rendering will work for a soft proof side by side comparison.
Using Photoshop I agree. There are dedicated softproofing applications around where the absolute approach works though. Best regards Claas
participants (3)
-
Claas Bickeböller
-
Graeme Gill
-
Nagy Péter