- This topic has 35 replies, 3 voices, and was last updated 2 years, 1 month ago by
.
-
Posts
-
As explained before i1d3 colorimeter does not outut XYZ coordinates, it outputs RGB RAW counter values than need a RGB -> XYZ matrix. This matrix is computed with CCSS and sensivity curves. ALWAYS. It needs this.
So I looked at the CMX-creation functionality in Displaycal. This allows you to measure the same patches using a spectro and the i1d3. As you said, the i1d3 creates a ti3-file with XYZ-values. But I wonder, there is no option to specify what CCSS is used! If a CCSS is always used, then which one is used for this function?
Probably it doesn’t matter, since it will cancel out anyway, but if this is arbitrary, then the CCMX matrix will not be close to identity. And you need to be sure that when applying the CCMX, exactly the same CCSS will be used again…
Calibrite Display Pro HL on Amazon
Disclosure: As an Amazon Associate I earn from qualifying purchases.As you said, the i1d3 creates a ti3-file with XYZ-values. But I wonder, there is no option to specify what CCSS is used! If a CCSS is always used, then which one is used for this function?
Sorry, did you really read me? I’ve repeated at least 2 or 3 times in this thread. No correction => ArgyllCMS or other software with Xrite SDK uses a default sample display with same spectral distribution (CCSS) as RGB sensivity curves stored in firmware. That’s default CCSS that allows to convert from RAW RGB counters to uncorrected XYZ.
I did read it, and also found it in the source code now. It just didn’t make a lot of sense to me. Why would you assume that the display has primaries that match exactly the sensitivity functions of the colorimeter? The only sense this could make to me is that the sensitivity curves of the colorimeter are the only curves readily available in the code, and you are using these only in situations where they will cancel out anyway afterwards. These “uncorrected” XYZ values might not have a lot to do with actual XYZ values then…
What is weird for me is that you cannot compute a ccmx matrix based on a CCSS file. If you put together a custom CCSS, this could be more accurate than measuring everything with a low-res spectro. The point is that in the current method, you need either a reference-quality spectro (if you use a ccmx matrix), or a reference-quality colorimeter (i.e., one where the firmware correction curves are a perfect fit to the actual instrument). It seems like an obvious omission that you cannot correct your colorimeter readings based on a reference ccss you downloaded somewhere…
I did read it, and also found it in the source code now. It just didn’t make a lot of sense to me. Why would you assume that the display has primaries that match exactly the sensitivity functions of the colorimeter? The only sense this could make to me is that the sensitivity curves of the colorimeter are the only curves readily available in the code, and you are using these only in situations where they will cancel out anyway afterwards. These “uncorrected” XYZ values might not have a lot to do with actual XYZ values then…
They are stored in the device, not in code. Also they include a scalling factor for units.
What is weird for me is that you cannot compute a ccmx matrix based on a CCSS file.
You can, but manually. It has been disccused before. Just run Argyll in high verbose mode -d3 or soemthing like that. Plug colorimeter and take some reading. It will output RGB to XYZ. Do it with and without CCSS. Then this matrix equation holds:
CCMX x NO_CCSS_RGB_to_XYZ = CCSS_RGB_to_XYZ
Solve for CCMX matrix. If you can look at Argyl code and understand it you can code it yourself if yiu want more decimal places on RGB to XYZ matrices.
If you put together a custom CCSS, this could be more accurate than measuring everything with a low-res spectro. The point is that in the current method, you need either a reference-quality spectro (if you use a ccmx matrix), or a reference-quality colorimeter (i.e., one where the firmware correction curves are a perfect fit to the actual instrument). It seems like an obvious omission that you cannot correct your colorimeter readings based on a reference ccss you downloaded somewhere…
As explained above that CCMX will have the SAME accuracy as an CCSS. 100% equivalent, accuracy based on CCSS matching display and firmweate matching device.
What you may mean is to re measure colorimetr sensivity curves and then back to firmware (or computer file). To do that you’ll need a monochromator & a lab grade spectro to take ref readings of monochromator in each of its bands.
That raises so much cost that it will be easier to rent a JETI with other people and batch measure all displays you can (both CCMX for each colorimeter and then CCSS generic for sharing, for each display)Ok, I’ll think about this some more. Just wanted to thank you for your patient explanations so far!
Why would you assume that the display has primaries that match exactly the sensitivity functions of the colorimeter?
Doing that minimizes the error while making no assumptions about the actual spectral power distribution of the display. This fact can be found in scientific literature (e.g. here, sections 4 and 4.1), but I’m afraid I don’t have an explanation for why exactly this works, as I don’t understand the math behind that myself.
