Re: There is no place for individual taste in screen appearance.
So if you used your calibrated screen to create a specific color numbers, then everyone viewing your numbers would read like your screen?
Halfway there. Both sides need to be calibrated. Let's say I finally notice my display doesn't output as much green as it should, and I calibrate the display. The calibration process outputs a display profile, and this display profile boosts the greens to make up for my flawed display. That profile is only relevant to my display, it only describes the variation between my display and some theoretical ideal. So I recreate the logo in Photoshop, with my new display profile tweaking my flawed display, and the eyedropper now says the numbers are 255 128 128. Perfect. Until I send that tiff to you. Because unfortunately, although you haven't noticed it, the blue channel in your display has been gradually increasing its midrange output over time. To you the eyedropper reads 255 128 128, but it doesn't look like a salmon at all, more like pink bubblegum. So you have to calibrate your display too, creating a profile that reduces the midrange reds to make up for the idiosyncracies of your display.
it only describes the variation between my display and some theoretical ideal.
Well, more accurately, it describes and adjusts. Description is referred to as "characterization", adjustment is referred to as "calibration". The process usually accomplishes both, and the profile contains elements which do each.
On Jun 6, 2014, at 6:27 PM, John Gnaegy <gnaegy@apple.com> wrote:
That profile is only relevant to my display, it only describes the variation between my display and some theoretical ideal.
Just a nitpick: there isn't any "theoretical ideal" in the process, any more than there's a "theoretically ideal" throttle position in your car for 55 mph on a flat surface with no wind. So long as there aren't any discontinuities (i.e., instances where a larger number causes a dimmer output) and so long as the desired white point lies within the display's gamut and so long as there're enough bits for the resolution you want, anything else is equally valid. As a practical matter, the closer R=G=B=255 is to your desired white point, and the closer equal RGB values are to the neutral axis for that white point and spaced for the desired encoding gamma, the more likely the profiling software will be able to create an highly accurate characterization of the display and thus create an high quality profile. That's what calibration is designed to do: adjust voltages (perhaps manually with knobs according to on-screen guidance) or DAC lookup tables or that sort of thing so that a neutral evenly-spaced gray ramp sent down the display cable would actually look like a neutral evenly-spaced gray ramp. But even if you skip the calibration and go straight to profile construction, the profile will take care of mapping the actual neutral axis to whatever unequal RGB values are needed and so on. The profile will be more prone to banding or other quality-degrading artifacts, but that's an implementation problem and not a theoretical one. Cheers, b&
Are the retina pads, older pads, LCD screens in general all viewing whatever I place in sRGB profiling anyway, so if I work in sRGB on an LCD screen and color measured as red appears red, blue appears blue, white is white w no cast, shadows have details, there is still no guarantee anyone viewing my work will see it the same as I profile, even using calibration devices. It's like there is no guarantee a printer will use my crafted profile, so I plan the image in Photoshop to print as well as possible w it's profile being totally ignored for his press. Sent from JRs iPad Air
On Jun 6, 2014, at 8:47 PM, Ben Goren <ben@trumpetpower.com> wrote:
On Jun 6, 2014, at 6:27 PM, John Gnaegy <gnaegy@apple.com> wrote:
That profile is only relevant to my display, it only describes the variation between my display and some theoretical ideal.
Just a nitpick: there isn't any "theoretical ideal" in the process, any more than there's a "theoretically ideal" throttle position in your car for 55 mph on a flat surface with no wind. So long as there aren't any discontinuities (i.e., instances where a larger number causes a dimmer output) and so long as the desired white point lies within the display's gamut and so long as there're enough bits for the resolution you want, anything else is equally valid.
As a practical matter, the closer R=G=B=255 is to your desired white point, and the closer equal RGB values are to the neutral axis for that white point and spaced for the desired encoding gamma, the more likely the profiling software will be able to create an highly accurate characterization of the display and thus create an high quality profile. That's what calibration is designed to do: adjust voltages (perhaps manually with knobs according to on-screen guidance) or DAC lookup tables or that sort of thing so that a neutral evenly-spaced gray ramp sent down the display cable would actually look like a neutral evenly-spaced gray ramp.
But even if you skip the calibration and go straight to profile construction, the profile will take care of mapping the actual neutral axis to whatever unequal RGB values are needed and so on. The profile will be more prone to banding or other quality-degrading artifacts, but that's an implementation problem and not a theoretical one.
Cheers,
b&
participants (3)
-
Ben Goren
-
John Gnaegy
-
John Robinson