Some time back we added some information to the i1D3 User Guide regarding the accuracy of Correlation Files, and the underlying i1D3 factory calibration.

We tested 3 i1D3 probes at random, and plotted their measurement readings without a Correlation File active.
Unfortunately, some people could not get their head around the concept of what using a Correlation file would/would not do.

So, we have re-performed the same test, this time with 5 different i1D3s, all WITH the same Correlation File active.

The results are the same as the previous tests, but should now be easier for all to understand, with a maximum probe-to-probe deviation of over 2.5 dE00, which is obviously beyond acceptable tolerance levels for accurate calibration.

If you have been to the i1D3 User Guide page before, use Ctrl F5 to refresh your browser, and look for the ? CORRELATION FILE ACCURACY modal button.

I have deliberately randomised the 5 tested probes, but interestingly there is some similarity between some of the tested probes, although as they were all different models, with wildly different manufacturing dates, it is impossible to derive anything from the similarities.

Steve

So a Spectro for probe matching is really the best way? and correlation files wont really give you any benefit due to different hardware responses of each individual device?
Am I reading that right.

The only way to be 'accurate' with any calibration is to use a good colourimeter matched to a good spectro, on the actual displays you are calibrating.
Anything else is a variable, due to the issues outlined.

But, ALL calibration is based on human perception, as that is how the various CIE standards, etc., were defined.
And that itself is compounded with issues such as metameric failure.

There is a point where an acceptance of 'achievable accuracy' has to be adopted.
All you can do is aim for the best possible calibration with the tools you have available.

Steve

Hi Steve, this is really good info to have. Do probes 'drift' over time?
i.e., could I rent a spectro to create correlation files for my i1D3, and be able to trust those corr files for years to come?

As the I1D3 probes have glass filters, they should remain stable for a long time, if looked after correctly.
No large temperature variations, no moisture, not left exposed to sunlight, etc.

Steve

Steve
Perfect. Excellent information.

Steve

Would the correlation file then be accurate for the forseeable future for a particular panel type?
The specific question is - if the SPD drifts on a display, I assume that would make the correlation file inaccurate.
However - is SPD what drifts on a display, or does that stay constant? Or is it just the relative balance of the pixels?

The issue with any single 'probe match', be that actual probe matching, or correlation file, is that even the same panels can have different characteristics on different individual displays... as well as drifting with age/use.
That is why for true accuracy, a probe match should always be performed on any display immediately before calibration.

But, it would certainly be better than nothing...

Steve

To help understand the underlying inaccuracy of any i1D3 the below is the data from the recent Bravia8 QDOLED correlation file generated by Enrico using his CR-300.
As can be seen, the direct Probe Match generates near perfect results, while the Correlation File (EDR) is less than perfect, by some way.
Again, this can be seen on the i1D3 User Guide page within the ? CORRELATION FILE ACCURACY modal.

Steve

EDR

If talking about a panel tech like White LED that a good i1D3 is suited for, how would you differentiate a good unit of i1D3 from a bad one? Is there a point of quality where even a probe matching to a high-end spectro isn't gonna do much to make a bad unit useful?

No probe is suited to any specific display technology, especially colourimeters.

And there is no way to define a better i1D3 from any other i1D3, other than to verify against a high-end spectro, as per the above example which basically shows an average i1D3 when used with an EDR/Correlation File.

When probe matching, any non-damaged i1D3 would calibrate to basically match the spectro, as shown above. You are then left with normal i1D3 probe restrictions, such as low-light limitations, potential volumetric gamut issue, etc.

Steve

Steve
So any random, functional unit of i1d3, even those with high deltaE deviation when compared to a spectro, should work fine for calibration when using probe matching?

Within the limits of the CIE filters for volumetric non-linear accuracy when using FCMM or FCVM, as that just matches RGBW.
Theoretically, MPVM would reduce such issues further.

Steve