Color HCFR Calibration Discussion (Post your calibration files here) - Page 129

That was my primary concern. I hear they drift. What good is beginning this hobby if I wind up calibrating the set incorrectly, due to the meter drifting.

I wouldn't get the benefit of seeing a correctly dialed in set....

However, This free software + i1d2 = absolute cheapest price to begin calibrating.

True ... OTOH, also be aware that there are some serious flaws in HCFR ... particularly if you're going to try to use it for color/tint/CMS adjustments.

I don't agree with some of the design choices in ColorHCFR. I would say the same for competitive software. For anyone taking the quote at face value, HDTVChallenged would probably also call CalMAN flawed for color and tint adjustment.

My opinion is that ChromaPure and CalMAN probably take less understanding for a reasonable look at color, yet assuming you're using the same measurement device, there isn't reason to expect actual measurement differences. Mainly any differences across software comes from how the measurement information is presented. I believe ColorHCFR uses a different DeltaE calculation for color than the default of other two programs mentioned, so if you try to minimize DeltaE you may be led in different directions depending on the software you choose. Only ChromaPure seems to have a built-in custom colorspace option that HDTVChallenged prefers for setting color and tint, and as mentioned on the last page you can use a spreadsheet for calculating a custom colorspace reference for ColorHCFR or CalMAN. My main point is that the measurements from ColorHCFR will generally be as accurate as your measurement device, and any complication comes from how to go about interpreting the information.

EDIT: It was pointed out later by fight4yu that CalMAN can calculate a custom gamut, but I couldn't get the feature to work in a reasonable amount of time.

But all colorimeter drift...and even "new" i1D2 sometimes are not correct. My i1D2 definitely had a red-push...I do use HFCR + I1D2 in the beginning, but once you learn more, you will know you want more I am now using CALMAN+i1D3+ColorMunki-Spectro... (and I am still not satisfied.. .)

Interesting. First I've heard of this. Please supply links that have the test results showing these serious flaws.

Thanks,
Larry

Not true re HCFR. All the current calibration packages have been cross checked against each other (calman and HCFR for many years now) and found to be using correctt equations and add no additional error outside your probe's accuracy.

Exactly ... it's not that you can't use HCFR for the aforementioned tasks, it's just that you can't necessarily take the readings (particularly the primary/secondary delta lumas) at face value.

If you're not careful, HCFR might lead you in the wrong direction.

"Correct equations" don't help much if you're using the wrong reference points.

Hummm ... I hate to see the terms "custom" and "prefer" applied here when the real question is do we want to sacrifice 99.9% of the gamut just so we can say we hit the BT709 defined targets along the edge of the triangle, or do we "sacrifice" the 0.1% of the gamut along the edges so that the remaining 99.9% is (closer to being) correct?

The answer seems like a "no-brainer" to me.

BTW, I just found out that CALMAN do offer custom colorspace. You can choose to read in the current WHITE point, the current primaries, or even the current secondaries color as your GAMUT. You can also change the Y as well, if you choose to.

I objected to your use of the word flawed. HCFR does what it says it does and does it correctly. If the user is flawed that is another issue.

According to their forums there is a "gamut target editor", but I'm not sure how to get the "editors window". I wasted too much time trying to figure out how it's supposed to work. I'm not sure why they just don't put it in the "Gamut Target" area, create a default workflow with it, or have documentation.

I thought Tom Huffman had used something along the lines of 'custom colorspace' in the article for calculating Y based on xy, but it's probably been a few years, so I'm not absolutely sure that's how it was stated. I guess ChromaPure calls the setting "Calculated Target".

I'm not sure why there would be an issue with prefer, since effectively I was referring to your opinion. Your opinion is simply not in line with the general thinking, which is probably why these programs default the way they do. For what it's worth, my opinion on gamma is also not in line with general calibration practices either. The reason for prefer was merely to point out there are other opinions on the subject, such as general practice.


In trying to figure out current 'gamma calibration', I've decided that a significant part of calibration is mere nonsense, and often it's best to avoid any sort of logic. In many cases calibration simply means people should follow common industry practices, regardless where that leads.
(This portion is either 0.1% or 99.9% facetious)

Here are the issues:

1) HCFR is set up to reference BT709 (or SMPTE-C) primaries and secondaries.

2) There is no clear path or documentation about how to change these references to reflect targets appropriate for a real world display. If there is such a method, is so convoluted as to be completely useless. This is a design flaw.

3) A bit of a calibration fetish had developed on these forums: "You must minimize delta luma to have accurate color. Just look at my nifty bar-charts!!!"

You add all these things up and suddenly you're looking at your 'finely tuned display' wondering why everyone looks like green martians.

Then, you spend a day or two trying to figure out why "delta luma" and "delta E" are actually moving in opposite directions. All while freaking out that your eyes are seeing something complete different than what HCFR is saying. 'Did I get a broken i1D2? ... Should I go ahead spend 110% the cost of the display on a "better" colorimeter??? .... ARRRRRGGGGG!!!'

Eventually, you do the sensible thing and scrap all of your Color/Tint/CMS work that you did with HCFR/i1D2 and start from scratch using a proper color decoder pattern, color filters and your own eyes. In approximately 10 minutes, you arrive at optimum Color/Tint settings and people, places and things look normal again. Later, these settings will be verified as being the best possible settings for the display by using your colorimeter with appropriate "safeguards" and references in place.

All of which is just a long way of saying you can take 10 minutes to do color/tint and be directly on target, or you can waste days trying to chase down that last 0.5% of error that due to control "granularity" you're not going to be able to get rid of anyway.

Ok ... rant over.

LOL ... It just that when you use terms like "custom" and "prefer" you imply that one is just pulling you know what from you know where.

I think "angryht's" term "normalized" is probably better and more accurate.

The problem is that people seem to be stuck on Y/Cr/Cb/BT709 and forget that their displays actually operate in R'G'B'.


Classic.

Only if one subscribes to your philosophy regarding display calibration. I suggest you write your own software that does what you want it to do and leave the rest of us to our green martians.

Here is a summary:

- Color filters may have issues on a number of digital displays. For example the typical blue-filter method for setting color seems to tend toward a high setting on some digital displays compared with measurements.

- Filters built into the TV are likely better than color filters, but since professionals have reported that all built-in filters do not work correctly, it doesn't hurt to check their operation for anyone with a measurement device. There is a linked spreadsheet on the last page to use for a 'calculated target' or 'custom colorspace' that will allow users to look at color similar to using filters, and for tint try to get the dotted lines on the CIE Diagram from the secondaries to the opposite primaries near gray.

- Someone reading 'directly on target' might think such a thing would fit precisely with general calibration industry practices, which it doesn't exactly. The industry supports measurements, and I'm not aware of any software that defaults to what you're talking about. The actual ideal target is to calibrate to the same colorspace as displays used in video production, but not all consumer displays have adequate controls for that task. If a certain set of electronics do have the ability, the preferred method is to calibrate for a colorspace likely used for video production.

- I don't necessarily disagree with you regarding the limitations of many consumer displays, mainly I'm just pointing out that your position generally doesn't fit exactly with common practice. I agree that what matters from an end-user perspective is to look at the actual results, and I concur on the "nifty bar-charts" comment. In reading items like http://store.spectracal.com/support/...-s-vector.html I'm simply not sure how thoroughly calibrators have actually checked correlation between theory and typical practice. If typical practice is flawed, does calibration really hold much value?

Just in case you are interested, here is how in 4.3.2:

1) Tools --> Editor Panel --> Under Properties, pick Gamut Target Editor
2) Editor Mode --> Pick Persistent Gamut. Temporary gamut does not quite work.
3) Do a Copy Gamut. Then rename it to something else (GT30-gamut)
4) Now, you can click "edit whitepoint" to do a read (or edit) the whitepoint.
5) Or, you can now click read on RGB and it will take xy data.
6) Or, you can click "edit secondaries" and you can read secondaries xy data as well, instead of the derived values.
7) Or, if you really want to go total custom, you can edit Yn as well.
Now, the important thing is, once you are done with this Gamut, you need to "use" it. Go to Source Setting --> Advance Options and then pick this new gamut (GT30-gamut) you just create.

It will take a while for this to take effect, but once you are done, everything will now be based on this new gamut, including grayscale, DeltaLHS etc...

I do agree that there should be a workflow to do this. I modified the DUO version to create mine. which is one thing that CALMAN is good though. They give you a lot of flexibility to do whatever you like (be it right or wrong).

Anyone know in HCFR if (and how to) there is a way to enter corrections or offsets if you want to use a i1D2 that has been calibrated against a reference sensor?

And I'm pointing out that the "common (colorimeter based) practice" (at least with respect to primary/secondary Luminance) is going to lead you astray Unless you have a display that has perfect primaries in all three axes. This should not be controversial, it's simple mathematics: 1+1+1 = 3 kind of stuff.

PS: In fact the "common practice" will probably lead you in the opposite direction from where you need to go. I'll post an example of what I'm talking about later (running short on time.) I suspect you already know what I'm talking about, but may be wary of climbing all the way out on the limb with me.

I gave up on the colorimeter for fine tuning. The Blue only mode method gave me too much 'red' push. So, using a light meter, adjusted for 21% red and then confirmed with the colorimeter. I then adjusted the tint, using the 'blue only mode, to balance the seconadries blue output.
I then checked the accuracy of the colour gamut of the colorimeter by reading the output from a grade 1 broadcast monitor. The Spyder 2 colorimeter was the most accurate, only out by a couple of percent on the red and green targets.
This method seems to have provided the best colour performance.

On what display? I'm assuming some variety of plasma.

Maybe you're looking for the item under the 'Advanced' pulldown menu, 'XYZ coordinates adjustment matrix'.

I'm still not sure you get the standard position. Like I said, I think some typical calibration methods are complete nonsense, but here are the major points of the typical calibration position:

1) The ideal is to match the colorspace used for video production. (See Note1)

2) The general industry position assumes DeltaE corresponds with perception. If such an assumption is generally true, then reducing DeltaE comes closer to the ideal mentioned above. (See Note2)


Note1: Some displays do not have the capability of matching a pre-defined colorspace, so any 'calibration' attempt will be a compromise in one way or another. Your position falls into this category of being a compromise from the ideal.

Note2: My opinion is that the DeltaE calculation used in ColorHCFR does not correspond with perception, so reducing DeltaE for color in ColorHCFR does not generally tend to come closer to an ideal. I believe the default DeltaE calculation for color is different in CalMAN and ChromaPure.

When I go to the item on the left marked tools I get 8 icons, and none are listed that way.

It is the 3rd from the first row (It will say Editor's Panel).
When you click it, it will bring up like a "Create Profile" window. Then you highlight that window, and then click Properites (on the right side of the screen) then you can change it to Gamut Target Editor (default is Meter Profile Editor).

Let me know if you need some screenshot.

Here's what the tools item at the left shows.

This is interesting. This is what I have on mine. As you can see, I have that "editor panel"....

[IMG][/IMG]

I am using the latest, 4.3.2.266...

HDTVChallenged Quote:
Originally Posted by delphiplasma
This method seems to have provided the best colour performance.

On what display? I'm assuming some variety of plasma.

Hi,

This would be a plasma screen. HCFR usually specifies 'plasma' in the drop down menu selection, for the i1's. However, I find that the 'LCD' setting gives the nearest results for greyscale set-up. LCD mode reduces the over warm white temperature. I think the built in IR filter algorithm, used in HCFR, may have been correct for earlier i1 colorimeters, but probably due to a manufacturer change, the i1's have better IR filters?

I have HCFR 2.1 (which as far as I know is the latest version) and under the "Advanced" pulldown there is no listing for 'XYZ coordinates adjustment matrix'. The HCFR help info says "...use the basic calibration file building feature in the advanced menu" but I see no such feature listed in the advanced menu or in submenus to the 4 items that are listed there. The only 4 items under the advanced menu are: Patterns, Updates, Preferences and Language.