Yay finally!!!

How does it look now? The charts look excellent. Do people have a nice natural look and everything else look right?

Grayscale and gamma look great, maybe the 10% gray step could have lower dEUV (currently at dEUV 5.0).

I don't see any gamut work done though, CMS nor color/tint.

dEUV isn't a great tool for measuring dark areas. It only factors chromaticity. Shades that are nearly black all look almost black regardless of their color.

I'm sorry but this is flat out wrong. Red tinted 1-10% stimuli looks completely different compared to blue tinted 1-10% stimuli and can picked out by an untrained eye.

I'm going to have to agree with D-Nice here, large grayscale errors can go unnoticed when relying on other dE formulas like dE76 and dE94 in CalMAN. Certainly a dEUV of 5.0 can been seen by the naked eye, especially when looking at very dark grays specifically.

I interpreted his comment to mean one shouldn't use dEL(uv) as an accurate measure for dark scenes (I don't know whether that is true or not) and by inference that one should use either a different dE measure or your eyes.

it is accurate as it errs on the side of caution, dEUV values are always higher than dE76 or dE94 values in the low end and midrange and match the top end

I'm not saying they are indistinguishable from black. What I was saying is that a 1% fully saturated Red doesn't have near the perceptual saturation as 100% luminance Red.

The phenomenon makes dEUV unsuited to compare errors at different luminances. Any of the year based dEs don't have that issue.

Perhaps dark gray errors are not as visible as bright gray errors but in my opinion and that of some others (including certain professional calibrators) the LUV and LAB based formulas don't do justice to visible grayscale errors in the dark end. I'd rather err on the side of caution and overestimate low end grayscale errors than underestimate them and be left with some easily visible errors.

Just my 2 cents...

Nice job kanti123

Now that your thread has been hijacked (sorry)...

I am always curious to hear opinions on the defacto standard dE that should be used. I have heard varied opinions on this over the years with most feeling dE76 is the proper one to use.

Jason

trying to find a common ground on dE is almost as hard as trying to find a common ground on whether to clip or pass WTW content when setting contrast and if clipping, at what digital level specifically

regarding hijacked threads, this happens all the time in discussion forums since discussions often go on tangent(s)... such is the very nature of discussion

I think it may be more likely the visible errors have more to do with the meter being unable to accurately measure near black values than it is about the formula.

The formula is only as good as the data.

regardless, I wanted to share that I've always had best results with dEUV for grayscale and that's because it reports the highest numbers of any dE formula, making the calibrator dial the grayscale in to a much tighter tolerance than dE76 or dE94... meter errors are what they are and so there's nothing that can be done about that aspect of calibration aside from upgrading or profling the meter

well, it started with my comment about the RGB balance at the 10% grayscale step being a bit off

my advice regarding that is get the dEUV lower at that step if the 10-pt controls on the Samsung are fine enough to allow it

if not, then don't worry about it

If you're looking at the dE numbers while you're working you are doing it wrong anyway.

these are various formulas for evaluating color error; regardless of which formula you select you want to get the dE values as low as possible noting that errors of 3 or less are generally invisible in regular program material

then what should I be doing? I use both the actual RGB % values and dE as I'm working

the Samsung 10-pt controls work best with contrast set in the low to mid 90s

Yeah I figure the dE that should be used will always ruffle some feathers but at the end of the day logically there has to be one that is the standard (otherwise we have no standards for our standards ).

Regarding the hijack, my comment was simply an apology to the OP and did not require an explanation.

Jason

Just forget the dE numbers, try to get the RGB percentages with in a tick one way or the other from the target.

The dE numbers don't give you any actionable data, IMHO the only value they have is in reporting and testing conformance.


The exception is the dE94 LCH break out is extremely useful because it does actually give you information in how to correct the error.

And that's true for where you actually have standards for pass/fail on calibration.

Rec.709 the standard for TV only defines the targets, not the tolerances for what is considered a successful calibration.

There are standards like DICOM (medical imaging) that do have very specific pass/fail criteria for calibration.

since you have a LED-LCD, you can turn down backlight instead of contrast to lower overall light output and improve black levels as a side benefit

yes, the D2 may not work well below 20% gray

that last bit is great for gamut, but I was referring to grayscale (color) error specifically

also, with dEUV the better the RGB percentages the lower the dE and all grayscale points are treated equally since the L portion of the formula is removed... in other words, what you suggest is not that different from what I do, I just look at dEUV in addition to RGB %

gamma is checked separately so if my target is 2.2, anything from 2.15 to 2.25 is acceptable