eeColor processor - ArgyllCMS - Page 2 - AVS Forum | Home Theater Discussions And Reviews
Forum Jump: 
 69Likes
Reply
 
Thread Tools
post #31 of 653 Old 04-03-2013, 03:44 PM
Advanced Member
 
Iron Mike's Avatar
 
Join Date: Oct 2006
Posts: 897
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 68 Post(s)
Liked: 413
Graeme,

can you elaborate on the HiRes mode Argyll supports for the i1Pro... from what I understand the spectral resolution is improved from 10nm to 3nm, there have been discussions whether this is possible with this device and if the data is usable or even worse than 10nm readings...

what were the results in your tests ?

Thanks.

- M

calibration & profiling solutions: Lightspace, Spaceman, Calman, Argyll, ColorNavigator, basICColor
profiling & calibration workflow tools: Display Calibration Tools
meter: CR-250, Klein K-10 A, i1Pro, i1D3
AVS thread: Lightspace & Custom Color Patch Set & Gamma Calibration on Panasonic 65VT60
Iron Mike is offline  
Sponsored Links
Advertisement
 
post #32 of 653 Old 04-03-2013, 04:16 PM
Senior Member
 
<^..^>Smokey Joe's Avatar
 
Join Date: Aug 2006
Location: Hobitville, New Zealand
Posts: 495
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 66 Post(s)
Liked: 94
I'm really not convinced by the Hires mode driver.

The optical limit is 10nm, this defines absolute nm accuracy.

Standard Spectro design dictates that several sensors(could be more than 3) should be averaged into output readings.

Because there are several sensors per region and you can report what they return doesn't mean they actually say anymore than the 10nm output.

The core calibration which can only be done by Xrite is based upon 10nm readings, not 3nm, where upon every sensor is still used but in averaging of groups.This is typically used to reduce noise.

Without higher level spectro references of say 0.5~1nm to prove the 3nm segments actually report greater detail than 10nm you have no idea what the readings actually mean. Very few actully have acess to 1nm or less spectros. (I have a 0.5nm190~1100nm device on a bench behind me, but it is designed to measure adbsorption and transmitance.) My jeti is capable, just, but I lack a reference lab setup, reference light source(TV's are to crap to correctly define levels).

The Konica Minolta website has a good description of spectro design, especially information and detail about what truely defines a spectro limit.
I'd advise all interested in this subject to read it.

Masterpiece Calibration Ltd
Lincolnvilleshire
New Zealand
<^..^>Smokey Joe is offline  
post #33 of 653 Old 04-03-2013, 05:00 PM
Advanced Member
 
gwgill's Avatar
 
Join Date: Jan 2013
Location: Melbourne, Australia
Posts: 926
Mentioned: 11 Post(s)
Tagged: 0 Thread(s)
Quoted: 250 Post(s)
Liked: 186
Quote:
Originally Posted by Iron Mike View Post

Graeme,

can you elaborate on the HiRes mode Argyll supports for the i1Pro... from what I understand the spectral resolution is improved from 10nm to 3nm, there have been discussions whether this is possible with this device and if the data is usable or even worse than 10nm readings...

Hi, the indications are that it improves the FWHM from about 25nm to 15nm. If you take a look at
http://www.argyllcms.com/doc/i1proDriver.html you can see that the color values in hi-res mode are at least very consistent with the normal-res mode, and the extra spectral detail captured hints at the possibility of improved accuracy, even for reflective measurements, which generally have much smoother spectra than display colors.
Quote:
what were the results in your tests ?

I've not been in a position to evaluate the differences objectively until just recently, but I'm hoping to make use of the JETI specbos 1211 currently at my disposal to make some measurements in the coming week.
gwgill is offline  
 
post #34 of 653 Old 04-03-2013, 05:53 PM
Advanced Member
 
gwgill's Avatar
 
Join Date: Jan 2013
Location: Melbourne, Australia
Posts: 926
Mentioned: 11 Post(s)
Tagged: 0 Thread(s)
Quoted: 250 Post(s)
Liked: 186
Quote:
Originally Posted by <^..^>Smokey Joe View Post

I'm really not convinced by the Hires mode driver.

The optical limit is 10nm, this defines absolute nm accuracy.

True, but the default FWHM is about 25nm, quite a way above the optical limit you quote.
Quote:

Standard Spectro design dictates that several sensors(could be more than 3) should be averaged into output readings.

While there is no doubt that averaging multiple readings will help mitigate random elements in the measurement process, nothing in the engineering of an instrument dictates that more than one sensor is required.
Quote:

Because there are several sensors per region and you can report what they return doesn't mean they actually say anymore than the 10nm output.

I'm not sure what you are intending to mean by that.
Quote:

The core calibration which can only be done by Xrite is based upon 10nm readings, not 3nm, where upon every sensor is still used but in averaging of groups.This is typically used to reduce noise.

That's not strictly true. The dark sensor value is actually calibrated at the raw sensor resolution, something that is needed to compensate for pattern noise. Wavelength calibration is a simple polynomial that is quite close to a straight line and therefore quite accurately interpolated (and in the i1pro2 the polynomial equations are explicitly given, and I am able to verify the accuracy of the interpolation). The emissive and white reference calibrations are indeed at 10nm resolution, but are also relatively smooth and therefore easily and accurately interpolated. The noise content depends mainly on the light level, and calibration values are typically made a light levels that minimize the impact of noise. There is no visual evidence of excessive noise at 3.3nm resolution, and of course the higher noise in the individual 3.3nm spectral values can have little impact on the resulting tri-stimulus values.
Quote:

Without higher level spectro references of say 0.5~1nm to prove the 3nm segments actually report greater detail than 10nm you have no idea what the readings actually mean.

I have a quite clear idea what it means, because the it's the result of stock standard sampling theory.
Quote:

Very few actully have acess to 1nm or less spectros. (I have a 0.5nm190~1100nm device on a bench behind me, but it is designed to measure adbsorption and transmitance.) My jeti is capable, just, but I lack a reference lab setup, reference light source(TV's are to crap to correctly define levels).

The Konica Minolta website has a good description of spectro design, especially information and detail about what truely defines a spectro limit.
I'd advise all interested in this subject to read it.

I've looked through it, and it's good as far as it goes, but I don't agree it has any critical bearing on the topic at hand.

I'm not at all critical of Gretag for designing the i1pro to be a 10nm instrument, given its primary application is reflective measurement, and pushing it up to 3.3nm resolution is an exploration of what's possible on my part.

Without some objective evidence, I wouldn't go as far as to claim that my implementation of a hi-res mode improves accuracy, but I'm not aware of any theoretical or experimental basis to think that it makes it worse, or is somehow meaningless. If I come to any conclusions by using the JETI as a reference, you can be sure that I'll publish the results.
gwgill is offline  
post #35 of 653 Old 04-03-2013, 08:36 PM
Senior Member
 
<^..^>Smokey Joe's Avatar
 
Join Date: Aug 2006
Location: Hobitville, New Zealand
Posts: 495
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 66 Post(s)
Liked: 94
You could be right, not the first time a manufacturer builds in head room to allow the overal device tolerance to be tighter unit to unit. The i1pro's are pretty good in that regard relative to their specs.

Proofs in the pudding of course!

I have an i1pro and Jeti1211BT, could do some validation tests.


Passing thoughts about this, remembering that in both cases all the sensors are being used, in the 3.33 mode you switch on each sensor is being reported, in the standard mode sensors might be averaged (or processed into groups as 10nm). The result could arguably be the same because the final transform to the standard viewer is constrained to XYZ. I suspect the sensor sensitivity and noise floor has more effect than switching to Hires/standard res.

Masterpiece Calibration Ltd
Lincolnvilleshire
New Zealand
<^..^>Smokey Joe is offline  
post #36 of 653 Old 04-04-2013, 03:21 AM
Advanced Member
 
gwgill's Avatar
 
Join Date: Jan 2013
Location: Melbourne, Australia
Posts: 926
Mentioned: 11 Post(s)
Tagged: 0 Thread(s)
Quoted: 250 Post(s)
Liked: 186
Quote:
Originally Posted by <^..^>Smokey Joe View Post

Passing thoughts about this, remembering that in both cases all the sensors are being used, in the 3.33 mode you switch on each sensor is being reported, in the standard mode sensors might be averaged (or processed into groups as 10nm). The result could arguably be the same because the final transform to the standard viewer is constrained to XYZ. I suspect the sensor sensitivity and noise floor has more effect than switching to Hires/standard res.

Yes - as I mentioned above, I would expect the noise in the XYZ value to be almost indistinguishable between normal and hi-res mode. It doesn't matter if you integrate over 10nm and then over the visual wavelength range or do it in one hit at 3.3 nm, all the sensor values get used, and have a similar (but not identical) weighting in the final result.
gwgill is offline  
post #37 of 653 Old 04-04-2013, 06:08 AM - Thread Starter
AVS Forum Special Member
 
zoyd's Avatar
 
Join Date: Sep 2006
Location: Planet Dog
Posts: 6,275
Mentioned: 89 Post(s)
Tagged: 0 Thread(s)
Quoted: 1316 Post(s)
Liked: 1055
Are you certain that the 10 nm reported values are a result of a 3 pixel co-add in firmware? If not then "hi-res" mode will bump your SNR and possibly do a better job at integrating peaked or wide gamut spectra. If it is a 3 pixel coadd then you are limited by the optical resolution and I wouldn't expect any difference between the two modes in producing tri-stimulus values.
zoyd is offline  
post #38 of 653 Old 04-04-2013, 06:25 AM
Advanced Member
 
gwgill's Avatar
 
Join Date: Jan 2013
Location: Melbourne, Australia
Posts: 926
Mentioned: 11 Post(s)
Tagged: 0 Thread(s)
Quoted: 250 Post(s)
Liked: 186
Quote:
Originally Posted by zoyd View Post

Are you certain that the 10 nm reported values are a result of a 3 pixel co-add in firmware?

100% certain. The raw CCD values are resampled (and wavelength calibration applied) using filters, and the sum of the filter coefficients is unity (as it also is for the hi-res filter coefficients.)
Quote:
If it is a 3 pixel coadd then you are limited by the optical resolution and I wouldn't expect any difference between the two modes in producing tri-stimulus values.

10nm mode is not at the optical resolution limit.

The tri-stimulus values are demonstrably different.
gwgill is offline  
post #39 of 653 Old 04-04-2013, 09:35 AM - Thread Starter
AVS Forum Special Member
 
zoyd's Avatar
 
Join Date: Sep 2006
Location: Planet Dog
Posts: 6,275
Mentioned: 89 Post(s)
Tagged: 0 Thread(s)
Quoted: 1316 Post(s)
Liked: 1055
Quote:
Originally Posted by gwgill View Post

100% certain. The raw CCD values are resampled (and wavelength calibration applied) using filters, and the sum of the filter coefficients is unity (as it also is for the hi-res filter coefficients.)
10nm mode is not at the optical resolution limit.

The tri-stimulus values are demonstrably different.

I remember comparing normal vs. hi-res mode using my i1pro rev D and there were measurable differences, especially in red, on my plasma although perceptually small. At the time I didn't know if they were due to resolution differences or the change in spectral coverage since that was another difference between the two modes. It will be interesting to see if those changes get you closer to what the Jeti is measuring.
zoyd is offline  
post #40 of 653 Old 04-04-2013, 03:10 PM
Senior Member
 
<^..^>Smokey Joe's Avatar
 
Join Date: Aug 2006
Location: Hobitville, New Zealand
Posts: 495
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 66 Post(s)
Liked: 94
Although this is slightly thread off topic, it is topical if -H mode switch is being used for the i1pro

The task here is to resolve if indeed the -H switch can return results of higher resolution, then if that is true, does it come with caveats like increased noise or some other loss greater than the gains.
True testing methods with this sort of thing is a problem outside lab setups where you have absolute control over all aspects.

First problem, the need of higher resolution spectro than the i1pro 10nm and 3.33nm -H mode.
Idealy a CS2000 would be the choice device, however I have a Jeti1211 which is highly regarded and research capable.

Light source, this is a problem. You might think that just measuring a display might be enough, maybe it is, but there is issues with this where I can show you an example of in the following image.

The problem with source light spectra is chance of spectra match ups. That is the peak by chance falls on the default 10nm point and therfore the difference between 3 and 10 is nothing.
So we need to control the sprectra, it just so happens I have a special filter for lab work sitting on my desk.

The following is this filter and it's transmission spectra, measured via a Cary 50 lab UV-vis spectra 0.5nm 190~1100nm (just happen to have one on my work desk)

Note: this graph is the Cary50 spectro, I have not compared the i1pro in this chart.
What I have compared is 10nm, 3nm and 0.5nm transmission spectra through this filter. The idea is to look for expected differences between the different spans.
It is note worthy that it shows something I expected, chances of differences and no difference in spectra peaks.
What is useful is the peaks in this filter show differences in regions of visible light and so viewable by the i1pro.



So atleast I have a useful light filter to generate repeatable peaks of light spectra outside the source.
Unfortunately the Cary50 light source is an internal lamp, so I can't use the same light for the Jeti or i1pro. However because I can now control spectra peaks with the filter I might be able to use something as simple as daylight, which is arguably more consistant than a display.

onwards and upwards.

Masterpiece Calibration Ltd
Lincolnvilleshire
New Zealand
<^..^>Smokey Joe is offline  
post #41 of 653 Old 04-04-2013, 04:41 PM
Advanced Member
 
gwgill's Avatar
 
Join Date: Jan 2013
Location: Melbourne, Australia
Posts: 926
Mentioned: 11 Post(s)
Tagged: 0 Thread(s)
Quoted: 250 Post(s)
Liked: 186
Quote:
Originally Posted by <^..^>Smokey Joe View Post

True testing methods with this sort of thing is a problem outside lab setups where you have absolute control over all aspects.

I have no idea why you think that is the case.
Quote:

Light source, this is a problem. You might think that just measuring a display might be enough, maybe it is, but there is issues with this where I can show you an example of in the following image.

"proof of the pudding.." - testing on real world displays is exactly what's needed to given an indication of any real world benefit.
Quote:

The problem with source light spectra is chance of spectra match ups. That is the peak by chance falls on the default 10nm point and therfore the difference between 3 and 10 is nothing.

One of the things the Konica Minolta website doesn't seem to cover is how the CCD samples are converted into wavelength calibrated spectral values. This typically makes use of re-sampling filters ("decimation" filters), and if they are chosen and designed correctly, there should be very little evidence of "spectra match ups", since the filter widths for 10nm and 3.3nm sampling are quite different (3.3/10 different!). To put it another way, even if a very narrow peak falls exactly on a 10nm and 3.3nm sample center wavelength, the amplitude will be different.
gwgill is offline  
post #42 of 653 Old 04-04-2013, 05:36 PM
Senior Member
 
<^..^>Smokey Joe's Avatar
 
Join Date: Aug 2006
Location: Hobitville, New Zealand
Posts: 495
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 66 Post(s)
Liked: 94
Because in the lab we measure against references first, known values and without contamination.
Then we bring in real world.

Although it is just as easy to make something fail in lab conditions.

By using a filter and known source I'm trying to remove a varible with a constant and also create peaks at pre known points irrespective of source which includes using a display.
In theory a higher resolution spectro like a jeti will iliminate these issues, but the real world bites.

Actually I don't really care how i1pros are designed per say, they work well as 10nm as it is, but I'd like to know if the -H switch has some usefulness to all users especially in calibration terms.
You have some proof with respect to your examinations, but I'd like to confirm proof for myself especially with peaked spectra.

Masterpiece Calibration Ltd
Lincolnvilleshire
New Zealand
<^..^>Smokey Joe is offline  
post #43 of 653 Old 04-05-2013, 04:06 AM
Advanced Member
 
gwgill's Avatar
 
Join Date: Jan 2013
Location: Melbourne, Australia
Posts: 926
Mentioned: 11 Post(s)
Tagged: 0 Thread(s)
Quoted: 250 Post(s)
Liked: 186
Quote:
Originally Posted by <^..^>Smokey Joe View Post

By using a filter and known source I'm trying to remove a varible with a constant and also create peaks at pre known points irrespective of source which includes using a display.
In theory a higher resolution spectro like a jeti will iliminate these issues, but the real world bites.
The interesting thing to do in a laboratory setting, would be to run a monochromator source through the i1pro in 10nm and 3.3nm modes, and plot the XYZ spectral sensitivity curves against the standard observer.

What I would expect to see is that in the transitional regions between the maximums and minumums of the curves, that the spectrometer curves have "bumps" between the spectro sample points, where the curve is above or below the standard observer curves. When narrow display primaries fall into these "bump" regions, you get tri-stimulus errors. I would expect/hope that as well as the 3.3nm mode "bumps" being narrower, that their deviation from the standard observer curves is smaller.
Quote:
gwgill is offline  
post #44 of 653 Old 04-05-2013, 03:46 PM
Senior Member
 
<^..^>Smokey Joe's Avatar
 
Join Date: Aug 2006
Location: Hobitville, New Zealand
Posts: 495
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 66 Post(s)
Liked: 94
Source is the main problem, hence why I'm going to try this filter(trace shown in graph in previous post.) It is a size and shape to fit a benchtop spectro, but big enough to sit in light path of the i1pro and jeti
Be nice to have a light box and controlled light source, but got nothing like that to work with (short of yanking a light out of a $50000 benchtop spectro tongue.gif )

If you examine the filter you can see some peaks, hollows actually since its a transmission spectra, I kinda wish it was reversed as the blocking areas define sharper peaks. I'm going to keeping fooking for something better though.

Anyway, first we need to get a method worked out which could also be repeatable by a second person. Also I haven't even grabbed your driver yet, btw you haven't got it ported to excel by any chance?

Masterpiece Calibration Ltd
Lincolnvilleshire
New Zealand
<^..^>Smokey Joe is offline  
post #45 of 653 Old 04-05-2013, 04:28 PM
Senior Member
 
<^..^>Smokey Joe's Avatar
 
Join Date: Aug 2006
Location: Hobitville, New Zealand
Posts: 495
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 66 Post(s)
Liked: 94
Ok I downloaded the lastest, but something happens I don't like at all.

It disables Microsoft Security Essentials, so I have a conflict.

Win7 64

Masterpiece Calibration Ltd
Lincolnvilleshire
New Zealand
<^..^>Smokey Joe is offline  
post #46 of 653 Old 04-05-2013, 08:18 PM
Senior Member
 
venkatesh_m's Avatar
 
Join Date: Oct 2007
Location: Penang, Malaysia
Posts: 398
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 35 Post(s)
Liked: 26
gwgill,

I hv the eyeOnePro and the Jeti1201 that I can use for the data gathering but am not sure on how I can proceed as I am not sure how I can get the eyeOnePro to show the data you requested.

The Jeti software can export the spectrum data if needed I think. However, I only have plasmas to test with. Might be able to test on some LED backlit LCDs as well, but it will take time.

Do let me know how I can help
venkatesh_m is offline  
post #47 of 653 Old 04-06-2013, 03:19 AM
Advanced Member
 
gwgill's Avatar
 
Join Date: Jan 2013
Location: Melbourne, Australia
Posts: 926
Mentioned: 11 Post(s)
Tagged: 0 Thread(s)
Quoted: 250 Post(s)
Liked: 186
Quote:
Originally Posted by <^..^>Smokey Joe View Post

Ok I downloaded the lastest, but something happens I don't like at all.

It disables Microsoft Security Essentials, so I have a conflict.

Win7 64

Sorry, I'm not sure exactly what you mean.
ArgyllCMS certainly doesn't have anything to do with anti-virus software.
gwgill is offline  
post #48 of 653 Old 04-06-2013, 08:01 PM
Senior Member
 
<^..^>Smokey Joe's Avatar
 
Join Date: Aug 2006
Location: Hobitville, New Zealand
Posts: 495
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 66 Post(s)
Liked: 94
Greame, no offence intended, but unfortunately I had several links of your site open and downloaded the 32 bit version and 64 bit version into a folder, then at the moment I unzipped the 64bit, security essentials was disabled with an error.
After cleaning the pc and reloading security essentials new version I found a "Settingsmodifier " trogan.

btw its not the first time I have found public sites carrying unknown nasties, it may have been from somewhere else and timebombed, but I'm just reporting it to you to atleast scan and clean sweep your site just in case.

Then I'll try again ok.


Back to other stuff.
I did some preliminary measures, found my filter not to be working as well as I would like as I need a better source light.

However my LCD display does have fairly peaked spectrum response so may be good enough after all.
Main issue I have found at this point is the sensitivity range difference between the i1pro and jeti to get them on a similar scale, after exporting to excel I had to devide the i1pro readings by 16000 to get the readings somewhere close.Bit of a fudge guess at the moment. But it gave an interesting comparison none the less. I and using jetiLIMES and i1Share at the moment.

CalMAN 5 is a bit limited in spectral readings, shame as it can use both, but it probably won't see your driver version, in comparison the jetiLIMES software is perfect for the job, unfortunately the i1pro is not ported to this software. I know the Jeti people did some comparisons a while ago between the i1pro and 1201 so they may have a LIMES version which can use the i1pro. I'll contact them and see if theystill have it and will allow me to use it.

otherwise they do have public SDK and dll available for the 1211 so you may be able to port the 1211 to your setup.

Masterpiece Calibration Ltd
Lincolnvilleshire
New Zealand
<^..^>Smokey Joe is offline  
post #49 of 653 Old 04-07-2013, 07:17 PM
Advanced Member
 
gwgill's Avatar
 
Join Date: Jan 2013
Location: Melbourne, Australia
Posts: 926
Mentioned: 11 Post(s)
Tagged: 0 Thread(s)
Quoted: 250 Post(s)
Liked: 186
Quote:
Originally Posted by <^..^>Smokey Joe View Post

Graeme, no offence intended, but unfortunately I had several links of your site open and downloaded the 32 bit version and 64 bit version into a folder, then at the moment I unzipped the 64bit, security essentials was disabled with an error.
After cleaning the pc and reloading security essentials new version I found a "Settingsmodifier " trogan.

btw its not the first time I have found public sites carrying unknown nasties, it may have been from somewhere else and timebombed, but I'm just reporting it to you to atleast scan and clean sweep your site just in case.

OK, so I've:

1) Scanned the MSWin V1.5.1 2 & 64 bit executable with AVG:
Clean.

2) Scanned my MSWin build system with AVG:
Clean.

3) Upgraded Microsoft Security Essentials on my Win7 64 bit test system, installed V1.5.1 64 bit.
Run Microsoft Security Essentials scan.

Clean.

4) It was submitted to https://www.virustotal.com/en/
see https://www.virustotal.com/en/file/b84a65039e6363a861a68e0fb98abd3071cd5a2942b636566457744638b8a96f/analysis/

Clean.

5) Many hundreds of copies of ArgyllCMS have been downloaded and installed, and I've had no reports of any problems with contamination by malware.

So I'm pretty sure the problem is at your end.

MSE does seem to crash when reading certain .zip files. You can avoid this problem by turning off it's scanning of archive files, manually de-archiving and scanning the contents.

Windows Defender on MSWin8 doesn't crash on these files.

It's not very reassuring that MSE has a bug like this though...
gwgill is offline  
post #50 of 653 Old 04-07-2013, 08:50 PM
Advanced Member
 
gwgill's Avatar
 
Join Date: Jan 2013
Location: Melbourne, Australia
Posts: 926
Mentioned: 11 Post(s)
Tagged: 0 Thread(s)
Quoted: 250 Post(s)
Liked: 186
Quote:
Originally Posted by <^..^>Smokey Joe View Post

otherwise they do have public SDK and dll available for the 1211 so you may be able to port the 1211 to your setup.

I've already added a driver for the JETI 1211 - dev. source code is here http://www.argyllcms.com/Argyll_dev_src.zip, selected MSWin 32 .exe's are here: http://www.argyllcms.com/dev_win32_inst.zip
gwgill is offline  
post #51 of 653 Old 04-08-2013, 12:16 PM - Thread Starter
AVS Forum Special Member
 
zoyd's Avatar
 
Join Date: Sep 2006
Location: Planet Dog
Posts: 6,275
Mentioned: 89 Post(s)
Tagged: 0 Thread(s)
Quoted: 1316 Post(s)
Liked: 1055
I did a couple more tests over the weekend in getting stable grayscale response. These two results were obtained using my own code with the hope of changing the target display transfer function without changing the gray scale chromaticity. To do this I forced the measured grayscale patches x and y coordinates to the input coordinates while letting the Y value change to match the overall transfer function desired:
Code:
X1=X0/Y0*Y1

where X1,Y1 are measured and X0,Y0 are inputs.

pre-lut 2.2 gamma



post-lut 2.2 gamma transformed to bt.1886 target



Yellow line measured, white line target, note that gray scale errors have not increased but transferred function is very close to targeted shape.
zoyd is offline  
post #52 of 653 Old 04-08-2013, 02:18 PM
Senior Member
 
<^..^>Smokey Joe's Avatar
 
Join Date: Aug 2006
Location: Hobitville, New Zealand
Posts: 495
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 66 Post(s)
Liked: 94
Greame, thanks for checking and sweeping out your system, sorry for any inconvience.
The trogan might of come from a certain large continent, we are expericing large spam counts from that direction at the moment.

Anyway dealt with now.

Back to business, Steffen from jeti got back to me and is quite interestied in this, but the LIMEs software can't be used for the i1pro at this point. Wants to be in the loop though.
Anyway I'll retry downloads and installs later today to get all the software/hardware dialled in.

Steffen mentioned that the Jeti Specbos outputs in W/m2 sr nm, if the difference in readings I saw from i1share were 10000 out it would indicate that the i1pro is outputing W/cm2 sr nm (Metres/centimetres), however since I found the difference seemingly about 16000 by trying guesstimates it could indicate the i1pro is not outputing absolute numbers.

So the question is, is this controlled via the driver or software?
we need both devices at same output.

Masterpiece Calibration Ltd
Lincolnvilleshire
New Zealand
<^..^>Smokey Joe is offline  
post #53 of 653 Old 04-08-2013, 04:40 PM
Advanced Member
 
gwgill's Avatar
 
Join Date: Jan 2013
Location: Melbourne, Australia
Posts: 926
Mentioned: 11 Post(s)
Tagged: 0 Thread(s)
Quoted: 250 Post(s)
Liked: 186
Quote:
Originally Posted by <^..^>Smokey Joe View Post


Steffen mentioned that the Jeti Specbos outputs in W/m2 sr nm, if the difference in readings I saw from i1share were 10000 out it would indicate that the i1pro is outputing W/cm2 sr nm (Metres/centimetres), however since I found the difference seemingly about 16000 by trying guesstimates it could indicate the i1pro is not outputing absolute numbers.

Typically Gretag/X-rite delivers mW/(m^2.sr.nm), while JETI returns W/(m^2.sr.nm), a factor of 1000 for emission measurements ("Luminance" mode). If you run the instruments to measure ambient light ("Illuminance" mode), there will be a factor of pi in there as well I think.

I get a pretty good match in levels using ArgyllCMS. I'm currently seeing if I can fix a calibration error in the hi-res i1pro mode that's adding about 0.5 delta E to the result compared to normal res-mode. The JETI 1211 and i1pro2 are agreeing to within about 0.8 delta E for white on the CRT I'm testing with.
gwgill is offline  
post #54 of 653 Old 04-09-2013, 12:52 PM - Thread Starter
AVS Forum Special Member
 
zoyd's Avatar
 
Join Date: Sep 2006
Location: Planet Dog
Posts: 6,275
Mentioned: 89 Post(s)
Tagged: 0 Thread(s)
Quoted: 1316 Post(s)
Liked: 1055
Quote:
Originally Posted by gwgill View Post

Typically Gretag/X-rite delivers mW/(m^2.sr.nm), while JETI returns W/(m^2.sr.nm), a factor of 1000 for emission measurements ("Luminance" mode). If you run the instruments to measure ambient light ("Illuminance" mode), there will be a factor of pi in there as well I think.

The spectrometers measure radiance and irradiance and these are converted via the color matching functions into luminance and illuminance. I have access to a variety of high resolution spectrometers as well as NIST calibrated radiance and irradiance targets if that could be helpful. I was going to look at taking a spare ocean optics spectrometer like this one and calibrate it for luminance but haven't had the time yet.
zoyd is offline  
post #55 of 653 Old 04-10-2013, 01:26 PM - Thread Starter
AVS Forum Special Member
 
zoyd's Avatar
 
Join Date: Sep 2006
Location: Planet Dog
Posts: 6,275
Mentioned: 89 Post(s)
Tagged: 0 Thread(s)
Quoted: 1316 Post(s)
Liked: 1055
Some additional measurements relevant to:

1. Longer term repeatability.
2. Comparisons between D3 and i1pro2.
3. Comparisons between collink and xicclu generated 3DLUTs.


I remeasured my display using the same 750 patch test set used for the original measurements in post #1. I have not changed the display settings and I looked at three test cases.

1. Original ArgyllCMS (collink) generated 3DLUT
2. Same as 1 except I used my i1pro2 instead of the profiled D3 to do the measurements.
3. A 3DLUT created by dispcalGUI, which uses a different ArgyllCMS tool (xicclu)

Very little difference was found in the mean behavior of all three test cases. Case 1 is separated by about 2 weeks with respect to the original measurements.


The profiled D3 tracked the i1pro very well and within it's estimated precision (0.4 dE 1-sigma). Outliers tended to be at low luminance (dark diamonds) as expected given the lower color precision of the i1pro2 at these levels.


The dispcalGUI generated output is statistically very close to the baseline collink generated 3DLUT performance.

dispcalGUI linear fit shows a 0.3 dE offset relative to the collink results


And here is a recap showing the reduction of color errors using the ArgyllCMS 3DLUT vs. no color correction.

zoyd is offline  
post #56 of 653 Old 04-10-2013, 02:06 PM
Advanced Member
 
fhoech's Avatar
 
Join Date: Apr 2013
Location: Stuttgart, Germany
Posts: 848
Mentioned: 36 Post(s)
Tagged: 0 Thread(s)
Quoted: 420 Post(s)
Liked: 278
I wonder if I should create 64^3 LUTs in dispcalGUI instead of the 65^3 ones. I've read in another thread each 65th point is fixed in hardware and can't be changed.

DisplayCAL - Graphical front-end for Argyll CMS display calibration and characterization
Current stable version 3.2.4 released 2017-02-19
Current development snapshot: 3.2.6 Beta (Windows/0install) released 2017-05-05 | Standalone | Changelog
DisplayCAL on Facebook
fhoech is online now  
post #57 of 653 Old 04-10-2013, 02:20 PM - Thread Starter
AVS Forum Special Member
 
zoyd's Avatar
 
Join Date: Sep 2006
Location: Planet Dog
Posts: 6,275
Mentioned: 89 Post(s)
Tagged: 0 Thread(s)
Quoted: 1316 Post(s)
Liked: 1055
Quote:
Originally Posted by fhoech View Post

I wonder if I should create 64^3 LUTs in dispcalGUI instead of the 65^3 ones. I've read in another thread each 65th point is fixed in hardware and can't be changed.

It's true that the 65th point (1.0) is fixed and not modifiable, but Graeme gets around that by mapping the 65^3 grid points into a 64^3 space using a 1D input curve that you can load into the eeColor box separately from the 3DLUT (but the LUT is still 65^3). In this way you can modify the end points. I don't know if you want to do that or not with your code since it probably has very little significance but it is technically the only solution. It would be nice if you could include an output switch for material with 16-235 levels, this involves remapping the 0-255 levels used to do the profiling into 16-235 coordinates and providing a smooth interpolation between 0-16 and 235-255.
zoyd is offline  
post #58 of 653 Old 04-11-2013, 01:13 AM
Advanced Member
 
gwgill's Avatar
 
Join Date: Jan 2013
Location: Melbourne, Australia
Posts: 926
Mentioned: 11 Post(s)
Tagged: 0 Thread(s)
Quoted: 250 Post(s)
Liked: 186
OK, it turns out that the current code isn't doing as well as it can for Hi-Res. mode, particularly with the i1pro2, which has a much bumpier diffraction grating, which makes up-sampling the calibration data a bit tricky. So I've implemented a way of doing this, and run a couple of simple, real world verifications tests.

See http://www.argyllcms.com/doc/i1proHiRes.html for the results. The summary is that yes, there is a slight improvement in colorimetric accuracy.

Source code used is here http://www.argyllcms.com/Argyll_dev_src.zip

MSWin 32 bit .exe's for key tools is here http://www.argyllcms.com/disptools.win32.zip
gwgill is offline  
post #59 of 653 Old 04-11-2013, 04:15 AM
Senior Member
 
<^..^>Smokey Joe's Avatar
 
Join Date: Aug 2006
Location: Hobitville, New Zealand
Posts: 495
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 66 Post(s)
Liked: 94
Greame, looks interesting.

I have the D model i1pro, once I clear the decks of crud i'Il get to task trying to work out how to use your drivers.
btw I found your 64x exe still has a hissy fit, both with MSE and nod32 from my work security system.
I could download the 32 bit stuff though, will this still run on a win7 64x rig?

I think those errors would grow towards low detection where noise ratio gets to high.
I was planning on repeating what you have there , but add 10% luminance step ramps for each aswell.The 10 readings per each point is probably enough, although that might depend on display.

For display noise or ability to recreate the same image repeatibly. Could use the jeti as gospil(even though one can't really),
where by use the specs as a guide, anything outside this is display noise. I suspect the measures per pixel for the spectral readings fall into the 2% at 100cd/m^2

Luminance accuracy ± 2 % @ 100 cd/m² and illuminant A

Luminance reproducibility ± 1 %

Chromaticity accuracy ± 0.002 x , y @ illuminant A

Color reproducibility ± 0.0005 x , y



CCT reproducibility

± 20 K @ illuminant A

Masterpiece Calibration Ltd
Lincolnvilleshire
New Zealand
<^..^>Smokey Joe is offline  
post #60 of 653 Old 04-11-2013, 04:30 AM - Thread Starter
AVS Forum Special Member
 
zoyd's Avatar
 
Join Date: Sep 2006
Location: Planet Dog
Posts: 6,275
Mentioned: 89 Post(s)
Tagged: 0 Thread(s)
Quoted: 1316 Post(s)
Liked: 1055
Quote:
Originally Posted by <^..^>Smokey Joe View Post

Greame, looks interesting.


I think those errors would grow towards low detection where noise ratio gets to high.
I was planning on repeating what you have there , but add 10% luminance step ramps for each aswell.The 10 readings per each point is probably enough, although that might depend on display.

For display noise or ability to recreate the same image repeatibly. Could use the jeti as gospil(even though one can't really),
where by use the specs as a guide, anything outside this is display noise. I suspect the measures per pixel for the spectral readings fall into the 2% at 100cd/m^2

The question was not about the absolute accuracy of either device, it was can you improve the relative accuracy of the i1pro against a higher resolution spectrograph using 3.3 nm sampled spectra. The answer is a convincing yes, although the improvement is small and may not hold under all measurement conditions, particularly if the display spectral gradients line up with bumps in the undersampled 10nm calibration caused by the grating.
zoyd is offline  
Sponsored Links
Advertisement
 
Reply Display Calibration

Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off