Color HCFR Calibration Discussion (Post your calibration files here) - Page 114

AFAIK couldnt gamma be calculated while realtime since it needs the Luminance values for calculating. I've never had an eye on gamma while live, so I couldn't tell what it reads meanwhile. My wild guess is that it takes the values from the measurement took before?

well, i think i've figured out why HCFR can't currently show live gamma, but also how to do it myself. this is probably not news to the experts here but this is what i think works.

gamma is defined as the exponent of output to input luminance, referenced to
100% IRE. since HCFR doesn't measure 100 IRE until last, there is no way to calculate gamma until the end of the gamma measurement sequence.

if i understand this correctly, if HCFR measured in reverse order starting with 100 IRE, it could display gamma live with a simple calculation.

for gamma target of 2.2, the luminance reference curve in HCFR gives the % of 100 IRE luminance for each input level from 10 to 90 (put mouse pointer over each point to see the % of 100 IRE required to hit target).

so my procedure to adjust gamma in live mode is
1) display 100 IRE pattern and write down Y measurement.
2) continue in live mode and change to any other IRE pattern.
the target Y for 2.2 gamma is the value at 100 IRE times the following
constants

10 (0.0060)
20 (0.0281)
30 (0.0619)
40 (0.1308)
50 (0.2146)
60 (0.3217)
70 (0.4530)
80 (0.6093)
90 (0.7914)

edit: it appears there may be a way to display live gamma in HCFR in the gamma reference
part of preferences, but it is not yet obvious to me how it works.........can someone explain how
to use this mode to do live calibration of gamma (and save me from having to do the math
myself during the calibration process)?

attached is the starting point for bringing the gamma into alignment. i used the RGB in the DVDO Duo
to get the color balance pretty close as my starting point.......next i adjust RGB up/down as a group
to move the gamma

attached file from above post

 

jvc-duo-and-sage-test-patterns-standard-low-best-rgb.zip 0.8359375k . file

Can someone answer this?

The problem is that at the high-end (80 IRE) the greyscale is 100-100-80%, and the blue gain is maxed out. Any help?

Without actually looking at your data, I'd say lower your gain controls in unison so you then can raise blue relative to the other two - this will have a similar affect to lowering contrast so you'll need to compensate by raising your contrast control.

cheers,


--tom

You also need a measure at 0% stimulus if you're using Display Gamma with Black Level compensation which is the default and what I'd recommend using.


I'm not sure what you're referring to but if you do a grayscale measurement run and then look at the grayscale measurement data in the main view window, you'll see a column called 'Gamma Y' - this is the target Y given your gamma reference, gamma calculation preference and your measures at 0% and 100% stimulus. Assuming those end point measures don't change as you change your calibration controls, you can move your controls to target the Y value listed.

hope this helps,


--tom

thanks tom!

for a while i couldn't figure out what you meant as i could not see a 'Gamma Y' anywhere.........eventually i found that my computer screen only shows me
(x,y,Y,delta E, delta xy) with no sign of Gamma Y............the scroll bar on the right had only a tiny gap, so it didn't jump out at me that there was one more row of data to be seen...........after much head scratching, i accidentally moved the scroll bar and Gamma Y came into view.

i also understand now that both 0 IRE and 100 IRE are required before gamma can be calculated since the gamma is 'corrected for black level' which requires 0 IRE.

thanks muchly! more optimization fun planned for today

I have some questions about how HCFR calculates Gamma.
I have the following measuments.

100 IRE Y= 180,565

IRE Gamma Y
10 1,23
20 5,21
30 12,6
40 23,94
50 38,9
60 58,1
70 81,9
80 110,1
90 142,94

IRE Target gamma 2,2 (calculated)
0
10 1,08336 (Y 100IRE(180,565)*0.0060)
20 5,073736 (0.0281)
30 11,176664 (0.0619)
40 23,617248 (0.1308)
50 38,748176 (0.2146)
60 58,086152 (0.3217)
70 81,79368 (0.4530)
80 110,015208 (0.6093)
90 142,895184 (0.7914)

The Gamma Y measurements are pretty close to the calculated target for 2,2. So far so good. But if i look to the corresponding gamma curve it doesn''t look that good, how is that possible, anyone?

GAMMA curce HCFR



Robin

You were right, thanks.

Now, I have a problem with the 10-20 IRE. I tried but didn't manage to lower the deltaE.

Cheers

 

calibration2.zip 0.7431640625k . file

What meter are you using

it depends on your settings in Preferences->References

My preferences are:

Colorspace: SD/TV REC601 (NTSC)
Gamma calculation: Display Gamma with black compensation
Reference Gamma: 2,22

dfgsgt<<bgergxxxxxx DAMN, COLORSPACE IS WRONG, IT HAS TO BE PAL/SECAM FOR EUROPE!!!

Robin

Eye-One LT

Here are my final calibration files for the Sony KDL-32W4210 (W4000), using the Eye-one display2 and HCFR. Looks good i think for an amateur The before files are the factory setting for cinema. Only the CIE diagram isn't better after the calibration.

Setting before:
Pic mode: Cinema
Backlight: 2
Contrast: 80
Brightness: 50
Colorsaturation: 50
Colortemp: warm2
Sharpness: 3
All advanced settings: off

Setting after:
Pic mode: Cinema
Backlight: 0
Contrast: 65
Brightness: 45
Colorsaturation: 50
Colortemp: warm2
Sharpness: 0
All advanced settings: off

RGB before

RGB after

Luminance before

Luminance after

Colortemp before

Colortemp after

CIE before

CIE after


Comments please

Robin

 

Cinema factory settings before.zip 1.166015625k . file

 

Cinema after.zip 4.259765625k . file

 

Sony adjust AFTER RGB levels.zip 6.630859375k . file

looks pretty good. maybe an opportunity to touch tint? move all secondaries a little counterclockwise (make yellow a little greener).

Hi,

My sony doesn't have a tint control or CMS. I cannot make the primaires and secundaires better, but i think it's good enough

Robin

Which one of the primairy and secundairy color files are best do you think?

Usersetting: Color 49


Usersetting: Color 50


Robin

Your screen shots are tantalizingly close to including the information that would help answer the question.

Expose delta luma, as shown by the upper red arrow in the attached graphic, and you will see which one gives better values.

Note if your primaries are way off, this method is bogus. But yours are close enough that I would go ahead and use the delta luma numbers in HCFR as a helpful aid to setting the color control. If it isn't obvious, you want delta luma to be small.

It is also best if DeltaE are te be small. In my case the lowest colorsetting in the usermenu gives the lowest delta luma but the highest deltaE?
So which one to choose???

color 41


color 49


color 50


Robin

Well, in general you are trying to minimize delta E.

But delta E is comprised of different components, and the color control (usually, mostly) affects the luma component, so if you are only changing color, improving delta luma will probably improve overall delta E. If that's not the case on your display, then yes, the larger goal is to reduce delta E, so don't get carried away chasing low delta luma values.

To review, the Delta E row shows total error. The two rows below that attempt to show the two sub-components of that error. Delta xy is a measure of how far off the hue and/or saturation are. Delta luma isolates just the "luma" (luminance, intensity, whatever- not sure Color HCFR terminology is Poynton-compliant here) error.

Stepping back even further, the CIE chart shown in HCFR is actually a view looking "down" on a 3-dimensional volume of valid colors in xyY space. This top-down view shows x and y, but not Y. The Y dimension comes up out of the diagram towards your eye.

There is a good visualization of this space here:
http://www.brucelindbloom.com/ChromaticityGamuts.html

So to get back to delta xy, that has to do with where you are in a given xy "plane" sliced through the xyY gamut. It's easy to see the dot move in the xy plane on the CIE chart in HCFR. The further it is away from where you want to be, the greater the "delta xy".

But it doesn't tell you whether that reading is too intense or not. That's what the color control affects, and that's what is being addressed by the delta luma row in HCFR.

When you're relatively close in xy-space then the delta luma readings can be relevant.

(The way HCFR calculates delta luma assumes you are very close to the standard primaries. If you want to know how far off you are from an ideal luma when you are forced to use primaries that are in non-standard locations in the xy plane, you'll need to use a spreadsheet. Your primaries are close enough in my opinion for the default delta luma calculations to have meaning.)

On most displays, the "Color" control mostly changes the "luma" of the primaries all at once. There may be changes in saturation (how much you are displaced inward or outward along a radius extending out from the white point) or hue (how much you are displaced clockwise or counterclockwise along a line *perpendicular* to a radius extending out from the white point) as well, but mostly this control affects how far above or below the ideal height above the xy plane you are. This is not easily visualized in the CIE chart in HCFR, so the delta luma row can be a useful aid in setting color correctly.

(It's also a good way to expose differences in luminance between the primaries. Sometimes displays are deliberately designed to "push" red to be more intense than green or blue for instance)

So to bring it all back to minimizing delta E, you would like to be able to control all three dimensions in the xyY space. Color and Tint are crude controls that affect all primaries and secondaries at once. Color mostly changes Y position, tint mostly changes xy position.

You say you don't have a tint control, so all I'm focusing on is the color control.

If two or three of your primaries are too intense, turning down the color should lower *overall* delta E. (How you decide what is the best way to determine "overall" delta E is, especially when a change moves the delta E for different colors in opposite directions, is a little more complex, but usually you can tell if you're improving things or not.)

If adjusting the color control raises delta E, then your luma is either close enough already or your display's color control is affecting hue and saturation as well, which would show up in the delta xy row.

I couldn't actually read the last jpegs your posted - the resolution was too low. As annoying as it is when people post massive jpeg's, there is such a thing as being too conservative. Of course you can always just post a .zip file containing all the relevant .chc files output from HCFR (the title of the thread even encourages it after all).

hi, thanx for the explanation. Here is the zip file of the relevant .chc files.
Which one is the best you think?

 

Calibratie Sony KDLW4210.zip 428.974609375k . file

Without seeing the display, I imagine 49 or 50 are looking best and 41 looks not colorful enough. I don't see a lot of difference between 49 and 50. If the red and magenta become objectionable, you may have to back it down a little. Your settings seem to be good for this particular display. Enjoy.

strangely the 41 setting is set by the instructions mabe by "calibration for dummies". That means measure a white 100IRE window, write down the Y reading, multiply it by 0,21 and that is the y reading a 100% saturated red window should measure. So this is not the best method for my panel 50 is the factory setting, so that's at least one thing Sony did good

During all the measurements i havel done the last weeks, one thing was very obvious. If i increase contrast above 70 than the blue level doesn't track so well! The R en G levels were pretty tight.

Robin

I've just installed HCFR and the driver for i1LT and first thing is I want to calibrate the display the PC is connected (then my TV, then ultimately my projector). There was a step by step guide how to do this on the forum here somewhere and for the life of me I can't find it now. Can someone please direct me to it? I've got no idea how to proceed past installing the software. Thanks.

http://www.curtpalme.com/forum/viewt...asc&highlight=

My question is about the RGB levels portion of my .chc file: Subjectively, will I be able to see any of the errors from 20% to 100%? There is an annoying zigzag between 30% and 40%, which made getting blue close challenging. All the dE's are below 2.5 and my understanding is that dE's within the 2.0-3.0 range are at the threshold between visible and invisible errors to the human eye. Keeping in mind the readings are not stable at 20%, which only puts out 0.648 fL, is it safe to assume I've done the best I can from 30% to 100%? Does there appear to be any room for improvement? The first file in the post above is the .chc.

What is the Bruce Lindbloom method in lay terms? I did a search and couldn't understand much. Very technical and mathematical .


bob

I find your contrast setting to 100 very high. Is that it's maximum? That perhaps explain the lower Red levels from 90-100 IRE. Decrease the contrast to 35 ftl and you probably notice an increase of the red level. I also find your Gamma too high at avg 2,5, but it's pretty smooth from 10-90IRE. DeltaE for grayscaling is low enough to be concerned about as for color temperature. But if i was you i set the gamma to 2,2. Your picture is now too dark i guess. Perhaps if you increase brightness and decrease contrast the bluelevel may also tracks a bit better from 40-70 IRE.

Robin

The easiest way to think of it in my case was to set color so that the R value for red matches or is slightly lower than the Y value for white (in cd/m2 not ftL). Also, the R and G values for yellow must be equal or as close to equal as possible using the tint control.

I'm actually happy with the gamma and contrast settings and I've tried 2.2, 2.3, and 2.4 gamma before as well as 95 and 90 contrast. I was primarily concerned with the zigzag around 30% to 40%. The lower red levels from 90% to 100% are quite minor and I clip white above 235, so I don't see any color shift.