Colorimeter spreadsheet for SpyderTV - Page 2

Rader, have you considered including the dE(1994) in addition to the dE(1976) calculations? Reason I ask is becasue krasmuzik makes a good case in that the 1994 approach yeilds more useful information insofar as actual implementation of the data.

http://www.avsforum.com/avs-vb/showt...1&page=2&pp=30

I don't care for new 1994dE that was Ursa said that - anything beyond 1976 is not incorporated in any video standard. I prefer using the dC*, dH* and dL* measures based on 1976 CIE L'u'v' charts as they match your perceptions on video. I have tested this on customers and calibrators - and they do work to increase your understanding of what is wrong with the display. If you want to read about it more - see the sticky on calibration charts.

The enhancements to dE had to do with getting more accuracy at very precise surface color matches (paint/color) - they did that at the cost of larger inaccuracy for larger errors (you can even see this in Ursa data). Video has large errors and you do not worry about the tiny errors - the sensors cannot measure them well and you can only see it in side by side. Paint and textile is something you see side by side all the time - and in addition they are affected by the incoming light balance. Video IS the light - and there is a difference. Different strokes for different folks.

Paint and textile do not have to worry about their greens' being blue - they worry about it being off 0.00001xy blue. Video absolutely has to worry about the green's being blue.

Rader

Are the "correction factors" for Avia built into the spreadsheet?

You're right. The presentation seemed off to me when I posted the above, but when I went back through your math, it's correct. I am entertained to note that you do RGB balance differently than either Jeff or I do it. AS we have covered before, there really is not a standard way to do this.

Later,
Bill

To be clear, I am mostly agnostic on these things. I take a "use the hardest grading scheme available" approach to most things. If you use the 1994 method, then try to keep dE under 4 if you can. If you use the 1976 method, try to keep it under 4. If you use the CMC method, the same applies. If you use all three, then use the same rule. What is interesting is that they each have different tolerance ranges for what they deem to be error.

Later,
Bill

Ursa

For using dE for calibrating it does not matter as your goal is no error. But if one is using dE for review purposes of factory presets - it does matter.

Kevin - The problem is that, in practice, no error is infeasible. Poynton indicates that a dE of 4 is largely imperceptible under normal viewing conditions, and I don't dispute that. I do think that "normal viewing conditions" can change slightly, and so the appropriate dE method is debatable. If you trace out some of the math, under some circumstances, the 1976 formula indicates higher error, and under others, the 1994 formula indicates larger error. I merely take the max of either when I'm messing around with my H77.

Later,
Bill

I think I am now comfortable with the Spyders measurements and can live within it's margin of error. Found the Avia disk and took more measurements. The direction the spreadsheet took me was to increase RCUT on my Sony 34HS420 almost to it's SM limit. From a CRT perpective this means that the G2 aperture is set too high but I am concerned that increasing the CUT values may clip low level response. One thing I can say is that shadow detail is much better now.

My relative RGB error is now within a few percent from 20-100 (stimulus, IRE?) but the green error is fairly large between 10-20 , it looks exacly like your original spreadsheet when toggling the RGB Corrections. I realize that calculating these values have many different approaches but should they really vary that much?

Thanks again for your efforts.

Rader:

I thought there was a problem in version 6 with the dE calcs, am I wrong? Don't remember the exact dE values, datas at home, but the primaries were really high, way over 4. I still do not understand how to lower this, my white point is fairly close to D65 so I have no clue as to what to do next.

For low light measurement, you need to increase the read time on the Spyder. At the default 4 seconds, you will get lots of bad results. If you increase it to 9 seconds, then you get fairly good precision. Anyone using the stock read times should not be measuring much below 1 cd/m^2.

Later,
Bill

I don't recall seeing an adjustment for measurement time with the Spyder2PRO Studio software. How do you make that change?

Thanks

The PRO Studio software does not have any adjustments for timing, it takes about 5 seconds for it's readings

Try as I might I could not flatten out the relative RGB% from 0 to 20% IRE using xCUT settings. From 30 to 100 IRE I am within 2%. dE is under 4 from 30 to 100. The only thing I can surmise is that my low IRE meaurements are wrong using the Spyder. The values for Y from black, 10 and 20 IRE are 0.36, 0.76, 3.712 respectively. I am using a 9 sec read time per Ursa's recomendation. Any thoughts?

If you are under 4 dE from 30 - 100, then stop. You are done. Go watch a movie. Anything beyond this gets to tweak-itis, and you will find yourself chasing the bulb aging process with constant calibration.

Thanks Ursa, you made my wife quite happy. Rader, thank you for providing an essential tool, your hard work is much appreaciated. On to geometry!

New software update from colorvision, add temperature to "/support" window

>>>> SpyderTV 1.1.120 - international Windows (31,032.50KB, 4 downloads)
This is the installer for SpyderTV for Windows. (Nov. 2005) <<<<

I have Problems to understand about my Gamma.

I have Calibrate my Toshiba MT700 with SpyderTV and Ursas Spreadsheet- for STV.
I put the Values in the HD-D Sheet and before I have a Gamma about 2,74. After I get a Gamma from 2,38 but when I look a Film the Picture looks very bright and overall not good.
Now I have take the Values in Raders Spreadsheet- Model and there a very different Gammas at all.
Ursa-HD-D= 2,38
Rader Avia= 2,09 (confirm my impression from above)
Rader GetGray= 2,44

I think anything I haven't understood ore missing for me to see why there can so much differences between the Spreadsheet- Models. Have anyone a Answer for me, why this is so?

kindly Rgds Damdy

I use Ursas Test-pattern.
Ursa (I Hope he read this): Are they similar to one of the above?

Kindly Rgds Damdy

damdy

Just like we find there is different method of RGB% calculation, there is different method of gamma calc.

Hey guys can someone please correct me if i'm wrong. i've been using Rader's spreadsheet along with my Spyder2Pro colorimeter to try & get my LCD RPTV as close to D65 greyscale as possible. i try first to get my gamma curve close to the target (dashed pattern in the spreadsheet) by adjusting my tv's service menu RGAM, BGAM & GGAM. is this correct, because by just adjusting the Cuts & Drives, it seems to just change my color temperature & not my gamma.

thnaks guys.

Carmine.

Hi Carmine,

While I am certainly not an expert in this area and can say only what I know to be true based on my experience with calibrating an HD CRT. I used CUT and DRIVE to adjust my color temperature though a gray scale ramp of 7.5 to 100 IRE. The color was off by moving my slider to zero. If your using composite you could always only leave Y plugged in to remove the color signal. CUT adjustmentd for lower IRE (7.5-50) while Drive (50+) for higher IRE.. After that I altered my GAMx to what the spreadsheet recommended and by what my eye deemed to look good, they are often not in total agreement.

A word of caution. I found myself going round and round at times with this calibrating thing, it is easy to become obsessive about it. So if your at a point where you like what you see and the data is close enough then your'e done. This is far from a exact science though many like to think it is.

My method is, of course, correct! I will say that there are definitely different ways of calculating gamma. I included two, and they should agree within about 0.05 of each other if the gamma is smooth. If it is not, then the point measurements will be off considerably versus the regression. Also, since my stuff is not going to be released, then maybe we should take this off-line.

On a more general note, you should check whether set-up is present. If you are using AVIA test patterns, then the IRE levels should match using the NTSC disc (I do not have the PAL disc to verify whether set-up is there or not) with set-up included. If you are using the PC patterns, then ignore IRE altogether and compre the digital values versus what you are measuring against. My guess is that this is the culprit. Rader's model needs quite a bit more work to be a generalized model (e.g., HD vs. SD, PAL vs. NTSC). It can be done manually, but you need to know what to change where.

Later,
Bill

I tried to use Derek Smith's S2xyY utility with Sydertv, but it didn't work for me.
At first I start spydertv in support mode. When the colorimeter window has opened I start the s2xyy.exe with one attached ini. The name of the colorimeter tool changed and the s2xyY programm beeps and the s2xyy textfile is generated. I tried a couple of times to read the numbers from the support windows into the clipboard, but without success. I tried also some modification in the inifiles but textfile and clipboard are always empty.

I'm not shure if a make something wrong.

Thank's for your help

Peter