New Calibration Tool, Excel Download - Page 2

Well I tried it with excel 2007 and got the same result, it only pastes the visible digits? Maybe it's an XP thing.

Dan,

Here is my contribution. I put together a spreadsheet that calculates the new targets based on primaries that are off. So far I am getting the same results as the Accupel display calibration calculator. I was hoping others could check it and test it out. And maybe it could be incorporated into your tool. The formulas came from the article in Widescreen Review Magazine, December 2008. Thanks very much to Tom Huffman for making the complex math simple to understand (and all your other contributions). Let me know what you think.

I have updated to version 2 which cleans it up a bit and makes it a little more simple to use.

 

angryht color calculator ver 2.zip 6.90625k . file

So, I'm trying out the spreadsheet, and the results I see don't make sense.

Here's the raw measurements I started from for R/G/B/W:

Red Green Blue White
x 0.62700 0.29600 0.15200 0.31100
y 0.33700 0.58900 0.06400 0.16200
Y 16.58200 56.14500 5.76900 79.23100

Using your spreadsheet, under NEW TARGETS it comes up with a weird recommendation for Red Y of 54.42373, Green of 4.31013, Blue 20.49714. Possibly a rotation of variables somewhere? The corresponding calculated values from Accupel are: 16.582, 56.145, 5.769.

Another discrepancy with Accupel, the recommedations for the secondaries: the spreadsheet derives for the secondaries:
Yellow Cyan Magenta
0.61265 0.15522 0.31123
0.34792 0.07573 0.15551
58.73386 24.80727 74.92087

The yellow xy is about where Red is on the CIE chart, the Magenta xy is about where White is, and Cyan is about where Blue is. The corresponding recommendation from Accupel gives:

Yellow Cyan Magenta
0.415 0.224 0.323
0.498 0.327 0.163
73.2254 61.3788 23.8578

I don't have the article, yet. My usual book/magazine store had the Nov 2008, but not the Dec 2008, so I've not been able to verify the formulas. I probably need to make the tour of several more bookstores this weekend to see where I can find the Dec issue.

That y value for white seems off by quite a bit. Typo?

Yes, typo, I accidentally copied the y value from the Magenta instead of the White column. The correct value should have been 0.328.

Bill, you can also go to their web site and sign up for an on-line subscription - it's $15 - and gets you access to all issues with the articles in PDF format. Both of Tom H's articles are available via PDF viewing.

cheers,

--tom

OK, I get a perfect match between the accupel calculator and the angryht calculator. Isn't that what you got? See files.

 

Angryht calcd and accupel.zip 98.34375k . file

Absolutely worth it!

Sheesh tom, where the heck have you been? Glad to hear from you again!

You're right. Double checking again with the corrected y value for white (now why didn't I see that right away myself?) the results are the same. My bad.

No problem. Thanks for checking.

Thanks - yeah, I took a bit of a break from the audio/video hobby to concentrate on a few other hobbies but with winter firmly entrenched, I've gotten back into it. It's good to see all the familiar names still here in the forum.

cheers,


--tom

Might want to delete the bad attachment off your post at the top of page 2

I'm not sure what you mean.

It has been proven that it works. See the 2 files in this

post.

Edit to get rid of the file attachment in post #31....as you have a corrected file in post #38, so someone does not download file in post #31 now.

The file in post #31 contains the spreadsheet and is named angryht color calculator ver 1.zip. The zip file in post #38 is called Angryht calcd and accupel.zip and contains 2 .bmp (pictures only) files showing that the results from the spreadsheet match those from the accupel calculator. I just zipped them together. That is why they have different names.

After working through myself the math behind the discussion in the other thread on the AVS HD disk, especially the gamma adjustment to even out the measurements, and with much assistance from Poynton's information and the equations on Lindbloom's site, I've worked up a spreadsheet that recalculates the saturation targets based on the measured primary points and the measured average gamma. Using these points does isolate the effects of saturation error a little better from the other CMS adjustments on the color gamut.

Dan, I did not get to the point, yet, of figuring out your spreadsheet well enough to adjust the comparison points for the delta E (Luv) calculation. And, in my calculation, I just used the raw RGB percentages provided by Tom Huffman, neglecting any quantizisation of the values when represented as integer values in YCbCr or RGB.

Enjoy,
Bill

 

SaturationReferences.zip 14.5576171875k . file

Thanks, Dan. Any update on being able to incorporate primaries that are not in line, or adjustable, with the standards? I gave it a shot, but I don't want to post a modified version of yours and Bill's.

You bet! Actually, I have updated it:

Feel free to use it, that was my intention. I notice you have posted an update; hopefully I'll have a chance to try it out this weekend.

Bill

I will look at it again. Since Tom did not post the 0% numbers, I didn't bother recalculating these, and I've been thinking those values should be easy to figure out.

Looking at your spreadsheet, it is curious that the Y values are not identical across the red/green/blue reference values. I picked up on all this in the other thread, which discussed how the values HCFR sent were designed to give a uniform luminance. So I would have expected the Y values to be identical in the reference data. I assumed the difference came from rounding errors in the calculation, but it does leave me wondering. Similarly I attributed small differences I saw between your reference values and my calculations to rounding differences in Excel.

I'll let you know what I learn.

Cheers,
Bill

Just as there are several different formulas for Delta E, there are several different formulas for this.

Again, use Lindbloom.

I re-tested the spreadsheet today, plugging in the Rec709 color primaries instead of my measured primaries, and I see the correct values for all the 100% saturation points. In the spreadsheet I had posted, I had left my measured primaries and modified gamma, to illustrate how to use it to predict the modified saturation points. My guess is that you were looking at results without modifying the colors to match 709, so of course the 100% saturation points looked different.

I replaced the file with a new one, that starts with the Rec709 colors and assumes a decode gamma of 2.22. I seems to me that this predicts the values you have hardcoded into the Gamut/Saturation/Brightness spreadsheet as the 709 saturation targets. The updated version has the benefit of incorporating 0% saturation targets that I have calculated. Tom Huffman did not include these in his table, but I have another more detailed spreadsheet that reproduces the calculation behind his numbers.

As regards the formulas in the spreadsheet, these of course are copied from Lindbloom's website, so the calculations should be the same as his online calculator. Obviously, Excel may display more or fewer digits depending on the formatting.

The model in the spreadsheet still uses the RGB percentages given by Tom directly. It does not yet model the process of 709 encoding to and decoding from the disk in YCbCr, which would introduce another couple rounding steps to integers, but might make the predictions slightly more accurate. I've already worked through those as part of another exercise I did to understand color encode/decode errors, so it would not be hard to incorporate those here. I might go ahead and do that, just for fun, to see what effect it has on the values.

By the way, I can explain the small dips in the Y values at the intermediate saturation points. As mentioned in the other thread, the HCFR provided values were designed to give the same brightness at all saturations with a gamma of 2.2. The percentage values you used in your spreadsheet assume a display gamma of 2.22. This small difference in gamma creates the dip in Y at intermediate points.

Cheers,
Bill

As I suggested above, I went ahead and calculated what happens to the intermediate saturation RGB points when they undergo the Rec709 conversion into the YCbCr representation on the disk. It's really pretty interesting, because with the different coefficients and thus ranges for each of the values in PbPr, where Y would have equal Red and Green components, 50% Y in the YCbCr representation after rounding ends up with slightly different values for Red and Green.

So, I've replaced the spreadsheet again. At the top, you can see the HCFR percentages posted by Tom Huffman, and what I believe end up being the percentages after they are encoded to and then decoded from the YCbCr representation on the AVS HD disk. If you modify the spreadsheet to use Tom's values instead of my calculated ones, it matches very closely the values currently coded into the Gamut/Saturation/Brightness spreadsheet. It could be that my calculated values would represent exactly what a set perfectly calibrated to Rec709 would display from the AVSHD disk. As my set is not nearly so perfect, I can't confirm this through measurement.

Dan,

This is my first post and my first attempt at calibrating. I originaly input Dusters settings and saw an improvement but I knew that I was still off. I am a photgrapher and already owned an I12 so it was no biggy to give it a shot. I followed the dummies guide and nailed the white balance.

My question is on the primary and secondary measurements. I am not clear on which patterns I should be measuring. The color or the saturation on the AVSHD disc I downloaded for my PS3.

Also, Once I start changinf the settings in RGBCYM, Will I then need to address the white balance again?