- This topic has 1 reply, 2 voices, and was last updated 3 years, 7 months ago by
.
-
Posts
-
I have been banging my head against an EPSON projector display for a few weeks now, trying different settings and running quick profiling tests to figure out its tone response curves so that I can get a nice calibration. The reason this is necessary (I think) is that some of the display settings have degenerate input-output (tone-response) curves. Specifically, they have a flat saturation at the top and bottom, where the different input causes the same output (i.e. RGB values 180~255 all cause the same Y luminance. Even with some of the “best” settings, the slope of the tone response curves “flattens” at the top. In other words, it looks more like an S than a purely exponential line (see attachment).
I have a very general (novice) question.
I feed ramped RGB 24-bit (RGB 8,8,8) inputs into the projector and then measure output XYZ coordinates via a colorimeter (SpyderX in my case).
I had assumed the “goal” would be to achieve a tone response curve very close to a target gamma curve (i.e. 2.2)
Question 1: I have so far been only feeding R==G==B inputs (i.e. achromatic greyscale), and then plotting the input against CIE XYZ Y coordinate. I guess I should feed only one pixel at a time to get separate R/G/B tone response curves though?
Question 2: When I tried calibrating using dispcal/DisplayCal, I was getting “failed to invert Jacobian”, which I assume was happening because it was trying to figure out a regression fit or something to the response of the monitor, but it couldn’t because of the degenercy I described above. So, is the only requirement of a Tone Response Curve that it be monotonic (monotonically increasing I guess). Is it typical to have such “flattening” at the top of the curve? I imagine this will cause errors due to discretization of the input when the software side (graphics card?) tries to send the “correct” input value to get the “target” output luminance (is that what is meant by “banding”? because it will not have sufficient granularity due to the weird shape of the “native” input-output curve of the dispay).
SpyderX Pro on Amazon
Disclosure: As an Amazon Associate I earn from qualifying purchases.Hi,
Question 1: I have so far been only feeding R==G==B inputs (i.e. achromatic greyscale), and then plotting the input against CIE XYZ Y coordinate. I guess I should feed only one pixel at a time to get separate R/G/B tone response curves though?
Depends on what your goal is.
Question 2: […] So, is the only requirement of a Tone Response Curve that it be monotonic (monotonically increasing I guess)
After calibration, that is what it should be to get good and consistent results (the calibration failing to figure out a way to deal with the defective native response may hint at a more severe problem of that particular projector).
Is it typical to have such “flattening” at the top of the curve?
Unless the projector is somewhat defective, no. It may also be caused by a temporal dimming function (usually employed by projector manufacturers to protect the lamp and increase its lifetime).
