Fork of HCFR Started - What's needed? - Page 4

I have a good understanding of the perceptual relationship between luminance and saturation. But you need to address the other part of the post, why do you want to know and/or minimize the perceptual difference between what your display produces and the wrong target. What you are saying is that if you can't calibrate to a reference target you want to calibrate to a non-reference target, why? What is the advantage of having a linear white point to color luminance ratio in the wrong color space if it increases dE in the right color space?

No, it would be minimising the difference between what the display produces and the correct target.

Let's say your display has an oversaturated red. Do you still want to set its Y to 21.2% of white? No you do not, for 2 reasons:

1. It will appear too bright because of the higher saturation. Its dE can be reduced by using a lower Y for red.

2. Nonlinear relationship between red and white.

And how exactly does it do that given the correct target doesn't enter into the formula? Again that formula doesn't calculate a target, it calculates the primary brightnesses of a perfectly linear display with the same xy primaries as yours, this is not something that you should then aim for and by themsleves the Y values given by it don't tell you anything useful about where the optimal setup is for your display given a particular set of controls.

The software already allows you to see the difference to the correct target, it's still not clear what you would want to do with the numbers that formula produces and what settings you would be adjusting while looking at them.

For both CMS and saturation/hue methods minimizing dE (which already includes the Y diff) to the standard targets is what you want, with possible caveats about which saturation % is best to do this at.

John

You know that's not the reason why the Y values change when you recalculate the gamut targets.

It has nothing to do with perception and everything to do with mixing.

So what? All I'm saying is the calibration software should ideally provide these Y targets so the user can minimise dE and/or make their display linear for their gamut.

What I'm saying is the correct target should be up to the user. So if the user wants linear mixing of colours, why not let them have it? If they want to reduce the Y for an oversaturated colour to get a lower dE and some increased linearity, then why not let them have that too?

Agree sort of but this is not a personal priority


Because there is only a finite amount of time and given I can't see the point of it and it doesn't solve any problem I have I'm not likely to spend any time on it, also I'm not keen to add extra information that is of questionable value to an already overcomplicated piece of software. So even if the code fairies supplied a clean patch I might reject it if it made the user interface busier or uglier...

John

Not at the moment, if there is any momentum then maybe, but for now the help we'll need is testing and patience.

John

please tell me when this is going to be done soon. like in a few months?

i just don't have $500-$700 for a CalMan or ChromaPure package, but would love to buy an i1 Display 3 for $250 to run with the updated HCFR (assuming it will support the Display 3).

really excited and will be following this thead closely.

thanks

i've just finished wrapping up the argyll api an exposing it as a c++ class, a bit more testing probably required but it's pretty much done and working well with an i1 lt, next is the integration with the main app which should take a few days then a bit of fiddling with the installer then the first test release.

So in real terms probably more than a week less than a month (at least for a test version with i1 display 3 support).

John

that's awesome.

will there me a new tutorial for dummies, or use existing dummies over at curtpalme?

$500 -$700 for calman or chromapure..I think your wrong with the prices.

$199 - you can buy your own colorimeter which you already have or going to purchase with the free HCFR

Chromapure $200 - then you can buy your own colorimeter or purchase it as a package with them which then add to price
(same as with CalMan)

yes, you're right. looks like around 450 if i go with either Calman/Chromapure with Display 3 bundle.

i was exaggerating.

JohnAd,

I just wanted to say thanks in advance for taking over development of HCFR and for all your hard work. I'm really looking forward to finally being able to upgrade my old i1 Display LT meter to the i1 Display 3 or ColorMunki Display!

Right on, but I would buy a used i1 pro instead of the colormunki. The i1 has a much better reputation and history. Some people already had a version of HCFR that supported the colormunki but readings were never reliable.

Here is another question, why are 2 (or was it 3) of the 6 primaries/secondaries on the Rec709 CIE reference points for gamut vary in HCFR compared to the other calibration software?
I don't recall which colors it was, but if you go to chromapure's site and look up the reference points then compare them to HCFR, some of the ones in HCFR are at different reference points on the CIE even when both are set to Rec709...

I'm referring to the NEW ColorMunki Display which is supposed to be the same hardware as the i1 Display 3 but crippled by the manufacturer to take slower readings and with a few less features. It's a fair amount cheaper than the i1D3. I have no desire/need to have a meter that costs $500+ (even used) to touch up my greyscale once a year.

Someone correct me if I'm wrong if you believe the i1 Display 3 will offer a distinct advantage over the ColorMunki Display when used with HCFR to calibrate a display. I'd prefer to save $80 if they will both yield the same result.

Link to new ColorMunki Display:
http://www.amazon.com/Xrite-CMUNDIS-...dp/B0055MBQOM/

I agree, if you are going to buy a super expensive meter, it's better to get a spectro meter instead of a colorimeter, so otherwise get something simple.

ColorHCFR shows reference points for the primaries on the CIE diagram, and they match the chart from http://www.avsforum.com/avs-vb/showthread.php?t=852536

just get an i1 Display LT for $110 and be done with it. that's what i'll do.

i don't have the money to spend on a meter that cost as much as my projector.

I have trouble with my I1 LT's drifting over time, so I end up buying cheap but new meters every 2 years as I can't trust the readings from the old one, and its safe to assume that new in the box the accuracy should be reasonable (for DIY calibration).

Yes, but since the i1 Display LT has been discontinued for almost a year now, any stock remaining has already started to drift from the factory calibration. The Display LT and Display 2 are going to become increasingly more difficult to find. Besides, they aren't particularly good meters to begin with and definitely aren't great for plasmas.

That's why being able to use the new ColorMunki Display ($170ish) or Display 3 ($250) with HCFR will be a godsend. Not only are they current meters, they are far superior to the LT/D2 and should drift much less over time (hopefully).

Speaking of which...

JohnAd, I will gladly throw you a donation once HCFR is working with these new meters. Just set up a site where we can do so.

Agree, the new Colormunki can be had for around $170 and has been as low as $140 on Amazon, and it is a superior meter than the Display LT and should not drift as much.
http://camelcamelcamel.com/Xrite-CMU...uct/B0055MBQOM

The Colormunki is 5 times slower than the Display 3 - but we will have to see when HFCR is ready for testing if it is useable. I'm OK if it takes 5 seconds for a reading instead of a second. If its timings are acceptable, the Colormunki & HFCR would be the best solution for the DIY calibration cheapskates!

On the topic of calibration for Y, I'm not really following the logic in the earlier posts in this thread. From what I recall, the current version of HCFR doesn't compensate for Y 100% or 75% saturation values, hence the need to use a separate spreadsheet. It seems we have not arrived at a conclusion what is the best way to treat this for the next rev of HCFR?? I would have thought this is the most significant missing feature in HCFR so should be prioritised, but I'm unclear why it is hard to do (workflow??) yet the other commercial packages can do it.

John, please post when you have a HCFR package ready for testing, and I'll try it out with my (1 year old, probably drifted) Display 1 LT and compare it to the previous version of HCFR. I will probably also buy the new Colormunki at some point so can test on that too.

i really like the ColorMunki price and its 'lasting longer', however i heard that it takes long to measure (which shouldn't bother most DIYers) and also have trouble measuring under 30IRE? is this true?

if i could get it at around 150 then it's really a bargain.

Here is the PDF that explains the differences between the Colormunki and the Display 3
http://www.xrite.com/documents/liter...lutions_en.pdf

Apart from the difference in bundled software (which is not an issue for HCFR users), the only hardware difference aside from speed seems to be the ambient light measurement which is luminance only for the Colormunki, and luminance and color for the Display 3. I'm not really sure if this difference is relevant for projector calibration in a dark room as this feature seems to compensate for the ambient light when calibrating.

Yeah, it's looking like the ColorMunki Display is probably the way to go for use with HCFR (once support has been added).

Another cost advantage to the colormunki is that X-Rite does not license 3rd party software vendors to use the colormunki in their software (not a problem for HCFR with the Argyll drivers) so you have to use the more expensive Display 3 (OEM version) with commercial software.

Also interesting to note on the Argyll web page for the Display 3 / Colormunki that it notes the Colormunki takes a "noticeably longer time" to take measurements. There is a customer comment on amazon who compared the colormunki with the display 3 and says the colormunki is only 2x slower, not 5 times, quoting a full calibration with the software taking 5 minutes on the colormunki and 2.5 minutes on the display 3. Assuming the software calculation time is tiny compared to the meter sampling time, then these results should translate to HCFR pretty well. As the new Display 3 reads significantly faster than the older Display 2 LT, it would be reasonable to expect the colormunki to be about as fast as the old Display 2 LT. But the meter will read slower under 30 IRE (as does the older Display 2 LT).

To smokarz comment about trouble under 30 IRE, I have seen no evidence that the colormunki is less accurate and as the devices internally are almost identical (only firmware changes for the speed and ambient light support more than likely), its unlikely there is a difference. In fact the newer X-Rite meters are more accurate under 30 IRE according to the review on the Chromapure website. The amazon customer comment/review when comparing the D3 with the Colormunki showed identical results.

Another useful article gives a comparison between the colormunki/D3 and other meters, and shows these new x-rite meters to be very good, comparible to $1500+ meters, and a definite improvement over the older D2 meters.
http://www.drycreekphoto.com/Learn/C...nHardware.html

Another good thread about the new x-rite meters: http://www.avsforum.com/avs-vb/showthread.php?t=1342422

So in conclusion, in comparison with the older Display 2 LT, the Colormunki Display will be more accurate (especially for lower IRE readings), drift less and have comparable speed for only a few bucks more than the old Display LT.

Aside from the meter discussions, it would be good to agree on an answer for HCFR for the following:

Having had a quick scan through the code I'm starting to see where I think the issues are. There are quite a few things that need to be fixed and changed and I was planning to go through all the code and calculations to confirm them. The biggest changes internally will be that we'll need peak white and black readings available and consistent throughout the code. So my ouline plan is

Meter support
Clean up, correct, extend and test colour maths library
Go through ui and simplify code and ensure each point's target is correct and that the displayed calculations make sense.

So essentially getting what's there working better before adding major new features in the UI. There will be plenty of releases along the way and this whole process will take a while but should be worth it...

Should still be on for a first release with the new meter code this week.

John

Since you're using argyll libraries, are you going to use the High Res mode for the i1Pro? As the i1Pro is not that great at low IRE's, even with the current crop of commercial software.

"A feature unique to Argyll when used with the Eye-One Pro, is the high resolution spectral mode. This returns spectral measurements at 3.333 nm spacing, rather than the default 10nm spacing, and also extends the range of wavelengths slightly. This high resolution may assist in giving better accuracy for "peaky" emissive sources such as illuminants and displays. The high resolution mode is selected by using the -H flag on the command line to dispcal, dispread, chartread, and spotread. It can also be toggled on and off within spotread using the h key. The extended range down to 350nm may give some information about an illuminants ultra violet content, although the accuracy of the readings between 350-380, and 730-740nm should not be relied upon."