A comparison of 3DLUT solutions for the eeColor box - Page 25 - AVS Forum | Home Theater Discussions And Reviews
Baselworld is only a few weeks away. Getting the latest news is easy, Click Here for info on how to join the Watchuseek.com newsletter list. Follow our team for updates featuring event coverage, new product unveilings, watch industry news & more!


Forum Jump: 
 62Likes
Reply
 
Thread Tools
post #721 of 747 Old 05-20-2015, 07:26 AM
 
all clear's Avatar
 
Join Date: Sep 2008
Posts: 341
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 166 Post(s)
Liked: 69
so in other words
if you have a tv that is already close to reference you will not see any major improvement with more patches?
it all depends on how accurate they are from the beginning right.

i use the i1D3 with a KRP-500A Plasma.

an comparision between the i1D3 and the K10-A would be great.
zoyd has one here that you probably already know about.
i1Display Pro - Klein K10-a tracking

a more interesting test would be how much the differences are between lets say 3000 patches and 9000 on different tvs.
compare the accuracy of an F8500 plasma to a Pioneer 9G Kuro as an example.
it will be shown right away in the charts what tv that benefits more from using more patches.

Last edited by all clear; 05-20-2015 at 07:46 AM.
all clear is offline  
Sponsored Links
Advertisement
 
post #722 of 747 Old 05-20-2015, 07:48 AM - Thread Starter
AVS Special Member
 
zoyd's Avatar
 
Join Date: Sep 2006
Location: Planet Dog
Posts: 5,858
Mentioned: 65 Post(s)
Tagged: 0 Thread(s)
Quoted: 984 Post(s)
Liked: 826
Quote:
Originally Posted by all clear View Post
to translate that to real world performance
do you see any difference in PQ with these two luts?
No, and you wouldn't expect to since both sets of residual error distributions are far below perceptual limits.

Quote:
Originally Posted by all clear View Post
an comparision between the i1D3 and the K10-A would be great.
zoyd has one here that you probably already know about.
i1Display Pro - Klein K10-a tracking
And one between the i1D3/CR-100 here. There is no difference in perceived image quality between the two. The only advantage would be if you have a highly non-linear display that requires a very large data set to characterize, the CR-100 will do it about twice as fast as the i1D3 for the same level of precision. Also, the CR-100 will measure below 0.001 cd/m^2 but that has no implication for LUT color errors.
zoyd is online now  
post #723 of 747 Old 05-20-2015, 07:56 AM
 
all clear's Avatar
 
Join Date: Sep 2008
Posts: 341
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 166 Post(s)
Liked: 69
Quote:
Originally Posted by zoyd View Post
Also, the CR-100 will measure below 0.001 cd/m^2 but that has no implication for LUT color errors.
thats one thing that i have wondering about
MLL on this KRP is unmeasurable with the i1D3 and probably also with the K10-A.

is there any 3d LUT colordata from 0 to lets say 0.002cd/m2?
i just wondering how much below 0.002 that gets blanked out by the i1D3 in the 3D LUT that effects the low end details.

will i be able to measure more patches below 0.002 with an K10-A thats the question?
all clear is offline  
post #724 of 747 Old 05-20-2015, 08:06 AM
Advanced Member
 
Light Illusion's Avatar
 
Join Date: Aug 2010
Posts: 621
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 52 Post(s)
Liked: 283
On most displays as you get down to the last 5 to 3% or so of range the back-light illumination becomes the strongest factor in the displays colourimetery.This cannot be correct, unless you lift the blacks and make them 'grey' - which is not at all ideal.
(That obviously depends on the display technology in use, but you get the idea).

A fair bit more data on the comparison, and the results on the test performed, can be seen here: http://www.lightillusion.com/calibra...omparison.html
It is interesting, as it is a direct set of results, and I've tried to include PQ information too.
In this instance there was definitely PQ issues that the 'data' didn't really show.

Steve

Steve Shaw
LIGHT ILLUSION

Light Illusion is offline  
post #725 of 747 Old 05-20-2015, 08:22 AM - Thread Starter
AVS Special Member
 
zoyd's Avatar
 
Join Date: Sep 2006
Location: Planet Dog
Posts: 5,858
Mentioned: 65 Post(s)
Tagged: 0 Thread(s)
Quoted: 984 Post(s)
Liked: 826
Quote:
Originally Posted by all clear View Post
thats one thing that i have wondering about
MLL on this KRP is unmeasurable with the i1D3 and probably also with the K10-A.

is there any 3d LUT colordata from 0 to lets say 0.002cd/m2?
That depends on the LUT in use and the display. The lowest control point on the eeColor (assuming you are going to leave black alone) is video level 20, which for your display calibrated to 120 cd/m^2 and gamma 2.4 would be ~0.008 cd/m^2. So measurements below that will be of little use, especially the chromaticity of those measurements.

Quote:
i just wondering how much below 0.002 that gets blanked out by the i1D3 in the 3D LUT that effects the low end details.

will i be able to measure more patches below 0.002 with an K10-A thats the question?
Most patch sets won't have many (if any) measurements this low except to establish black.
all clear likes this.
zoyd is online now  
post #726 of 747 Old 05-20-2015, 08:37 AM
 
all clear's Avatar
 
Join Date: Sep 2008
Posts: 341
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 166 Post(s)
Liked: 69
Quote:
Originally Posted by Light Illusion View Post
A fair bit more data on the comparison, and the results on the test performed, can be seen here: http://www.lightillusion.com/calibra...omparison.html
It is interesting, as it is a direct set of results, and I've tried to include PQ information too.
interesting
so you can reduce banding when using Lightspace over Calman ?
that was new.

im probably going to get LightSpace HTL here soon
if you compare the final 3D LUT result from using Lightspace to ArgyllCMS will there be any as you know about?

i have some banding at the low IREs now when looking at the grayscale ramp with the 3200 patches 3D LUT created by ArgyllCMS
but its not visible on real content.
all clear is offline  
post #727 of 747 Old 05-20-2015, 08:47 AM
 
all clear's Avatar
 
Join Date: Sep 2008
Posts: 341
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 166 Post(s)
Liked: 69
Quote:
Originally Posted by zoyd View Post
The lowest control point on the eeColor (assuming you are going to leave black alone) is video level 20, which for your display calibrated to 120 cd/m^2 and gamma 2.4 would be ~0.008 cd/m^2. So measurements below that will be of little use, especially the chromaticity of those measurements.
Most patch sets won't have many (if any) measurements this low except to establish black.
im not using the eeColor box yet but i will get it in a few weeks.
i use madvr as for now but it should be the same.

ok so level 20 is the lowest reading
why isnt level 17 to 20 measured?

isnt level 17 the lowest level you want to read to avoid black clipping?

interesting how you came up with the numbers for luminance at level 20 if you start from 0
asume i will measure level 17 with 120cd/m2 and gamma 2.4
what would the luminance be there?

Last edited by all clear; 05-20-2015 at 09:20 AM.
all clear is offline  
post #728 of 747 Old 05-20-2015, 09:37 AM - Thread Starter
AVS Special Member
 
zoyd's Avatar
 
Join Date: Sep 2006
Location: Planet Dog
Posts: 5,858
Mentioned: 65 Post(s)
Tagged: 0 Thread(s)
Quoted: 984 Post(s)
Liked: 826
Quote:
Originally Posted by all clear View Post
im not using the eeColor box yet but i will get it in a few weeks.
i use madvr as for now but it should be the same.
The madVR LUT is 256^3 so it has 4x as many control (adjustment) points per axis as the eeColor.

Quote:
ok so level 20 is the lowest reading
no, video level 20 is the lowest control point, you can read any level you want.

Quote:
why isnt level 17 to 20 measured?
You can measure any level you want in the profile.

Quote:
isnt level 17 the lowest level you want to read to avoid black clipping?
You don't need to measure the level 17 response to avoid clipping.

Quote:
interesting how you came up with the numbers for luminance at level 20 if you start from 0
video level 20 is 4 levels above video black. (4/219)^2.4 * 120 = 0.008

Quote:
asume i will measure level 17 with 120cd/m2 and gamma 2.4
what would the luminance be there?
(1/219)^2.4 * 120 = 0.000289811 cd/m^2. Your plasma might not even be able to dither this particular shade above it's black.
zoyd is online now  
post #729 of 747 Old 05-20-2015, 01:19 PM
 
all clear's Avatar
 
Join Date: Sep 2008
Posts: 341
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 166 Post(s)
Liked: 69
Quote:
Originally Posted by zoyd View Post
Here are the measured differences between a 3500 point profile and a 10,000 point profile (LS/CR-100 profile, ArgyllCMS LUT)
Zoyd
how do you build up this graph?
can an amateur also do it ?

im thinking of doing a new 3D LUT calibration with 9200 Patches.
if i can build up the same graph with the KRP as you have with the F8500 plasma it can be interseting to compare the two tvs.
that is if i can put in the errors from my current 3D LUT with 3200 patches?
all clear is offline  
post #730 of 747 Old 05-20-2015, 01:58 PM - Thread Starter
AVS Special Member
 
zoyd's Avatar
 
Join Date: Sep 2006
Location: Planet Dog
Posts: 5,858
Mentioned: 65 Post(s)
Tagged: 0 Thread(s)
Quoted: 984 Post(s)
Liked: 826
Quote:
Originally Posted by all clear View Post
Zoyd
how do you build up this graph?
can an amateur also do it ?
The graphs are built from verification runs of random patches with the LUT active and building them is easy if you have the dE values, you can even use Excel to build histograms. In order to get the dE values though you have to calculate them yourself like I do, or use something like Argyll's colverify utility to get text output of all the dE values. I think colverify will also plot a histogram for you but it won't overlay multiple histograms.

Quote:
im thinking of doing a new 3D LUT calibration with 9200 Patches.
if i can build up the same graph with the KRP as you have with the F8500 plasma it can be interseting to compare the two tvs.
that is if i can put in the errors from my current 3D LUT with 3200 patches?
I don't think there will be any difference in the minimum achievable average dE values from either panel. The limitations appear to be a function of working with an 8 bit video pipeline.

If you send me a .ti3 output file from DCG I can overplot it with one of my "best ofs".

Last edited by zoyd; 05-20-2015 at 02:11 PM.
zoyd is online now  
post #731 of 747 Old 05-20-2015, 02:16 PM
 
all clear's Avatar
 
Join Date: Sep 2008
Posts: 341
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 166 Post(s)
Liked: 69
Quote:
Originally Posted by zoyd View Post
If you send me a .ti3 output file from DCG I can overplot it with one of my "best ofs".
here is the latest from 3 days ago with 3135 patches

Edit: ok i get it
i fix it later or tomorrow

Last edited by all clear; 05-20-2015 at 02:31 PM.
all clear is offline  
post #732 of 747 Old 05-20-2015, 02:20 PM - Thread Starter
AVS Special Member
 
zoyd's Avatar
 
Join Date: Sep 2006
Location: Planet Dog
Posts: 5,858
Mentioned: 65 Post(s)
Tagged: 0 Thread(s)
Quoted: 984 Post(s)
Liked: 826
That appears to be a profile of the native display response, you need to run a random test set with the LUT from that profile active. What's typically called a "verification run".
zoyd is online now  
post #733 of 747 Old 05-20-2015, 03:19 PM
 
all clear's Avatar
 
Join Date: Sep 2008
Posts: 341
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 166 Post(s)
Liked: 69
i got it to work but all i got was an html file
no ti3 file
http://privat.bahnhof.se/wb192876/Me...1%2000-22.html

Last edited by all clear; 05-20-2015 at 03:34 PM.
all clear is offline  
post #734 of 747 Old 05-20-2015, 03:46 PM - Thread Starter
AVS Special Member
 
zoyd's Avatar
 
Join Date: Sep 2006
Location: Planet Dog
Posts: 5,858
Mentioned: 65 Post(s)
Tagged: 0 Thread(s)
Quoted: 984 Post(s)
Liked: 826
That looks great but you'll need more measurements to get better statistics on overall LUT performance. In DCG you can load this chart in the profiling section and then do options->measure testchart to get a .ti3 output. You also have to enable the madVR LUT. Do this by adding the switch -V to the dispread command line [options->set additional commandline arguments]
zoyd is online now  
post #735 of 747 Old 05-20-2015, 05:40 PM
 
all clear's Avatar
 
Join Date: Sep 2008
Posts: 341
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 166 Post(s)
Liked: 69
result
http://privat.bahnhof.se/wb192876/ma...XYZLUT+MTX.ti3

but what i noticed with this test chart was that it doesnt have a 0 black patch
is that correct or did i miss it ?
many of them was low but i think they were at the 2-3% levels

Edit: i had to start over because i was measuring the 3135 patches chart at first.
you have to select this chart under profiling not under verification. (just like you said )
to much info at once i think.

so how good or bad is this 7 year old KRP-500A from a scale 0-10 do you think ?

Last edited by all clear; 05-20-2015 at 06:02 PM.
all clear is offline  
post #736 of 747 Old 05-20-2015, 06:07 PM - Thread Starter
AVS Special Member
 
zoyd's Avatar
 
Join Date: Sep 2006
Location: Planet Dog
Posts: 5,858
Mentioned: 65 Post(s)
Tagged: 0 Thread(s)
Quoted: 984 Post(s)
Liked: 826
It has 1 black patch (545). But it's still the native response. Did you set the -V switch for dispread? The log will have the actual command line that was run.
zoyd is online now  
post #737 of 747 Old 05-20-2015, 06:11 PM
 
all clear's Avatar
 
Join Date: Sep 2008
Posts: 341
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 166 Post(s)
Liked: 69
so must i run it again or lol ?

just look at the picture

Edit:It seems that the log says -M?
its getting late here in sweden but i think i can have another go
Attached Thumbnails
Click image for larger version

Name:	V.jpg
Views:	43
Size:	371.9 KB
ID:	727906  

Last edited by all clear; 05-20-2015 at 06:15 PM.
all clear is offline  
post #738 of 747 Old 05-20-2015, 06:22 PM - Thread Starter
AVS Special Member
 
zoyd's Avatar
 
Join Date: Sep 2006
Location: Planet Dog
Posts: 5,858
Mentioned: 65 Post(s)
Tagged: 0 Thread(s)
Quoted: 984 Post(s)
Liked: 826
zoyd is online now  
post #739 of 747 Old 05-21-2015, 02:49 AM
Advanced Member
 
Light Illusion's Avatar
 
Join Date: Aug 2010
Posts: 621
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 52 Post(s)
Liked: 283
Ok - Delta-E distribution results from a 21^3, 17^3, and 10^3 calibration with LightSpace.
So 9261, 4913, and 1000 points...
(All verifications were performed with 21^3 profiles to maximises the verification accuracy)

21^3


17^3


10^3


(I have had to re-scale the graphs again to make them match visually)

As can be seen there is very little difference between the 21^3 and 17^3 results.
What is more interesting is there is virtually zero PQ difference between all three.
With a home TV I would expect the 10^3 calibration to show some PQ variations though!
Plutotype likes this.

Steve Shaw
LIGHT ILLUSION


Last edited by Light Illusion; 05-21-2015 at 02:55 AM. Reason: Additional info
Light Illusion is offline  
post #740 of 747 Old 05-21-2015, 10:36 AM
 
all clear's Avatar
 
Join Date: Sep 2008
Posts: 341
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 166 Post(s)
Liked: 69
Zoyd: here you go
i hope its right now

Reseted Pioneer KRP-500A with 3151 Color Patches
http://privat.bahnhof.se/wb192876/ma...XYZLUT+MTX.ti3
all clear is offline  
post #741 of 747 Old 05-21-2015, 10:54 AM - Thread Starter
AVS Special Member
 
zoyd's Avatar
 
Join Date: Sep 2006
Location: Planet Dog
Posts: 5,858
Mentioned: 65 Post(s)
Tagged: 0 Thread(s)
Quoted: 984 Post(s)
Liked: 826
zoyd is online now  
post #742 of 747 Old 05-21-2015, 10:58 AM
 
all clear's Avatar
 
Join Date: Sep 2008
Posts: 341
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 166 Post(s)
Liked: 69
Quote:
Originally Posted by zoyd View Post
yes that's better, what were the tone curve settings for the LUT?
phew
Gamma was set at custom at 2.3
all clear is offline  
post #743 of 747 Old 05-21-2015, 11:42 AM - Thread Starter
AVS Special Member
 
zoyd's Avatar
 
Join Date: Sep 2006
Location: Planet Dog
Posts: 5,858
Mentioned: 65 Post(s)
Tagged: 0 Thread(s)
Quoted: 984 Post(s)
Liked: 826


So I overplotted an F8500 result using a similar number of patches and they are very close but the Kuro has a little bit narrower (better) shape. The only area the Kuro did worse in was along the blue axis which reaches a maximum error of 2.24 at the primary location.

KURO
Code:
% < 1.5   % < 1.0
98.2094   97.2452
dE00 avg,stdev,max,worst 10%
0.33436705   0.28581043    2.2489707   0.96563032
F8500
Code:
% < 1.5   % < 1.0
99.8661   97.8581
dE00 avg,stdev,max,worst 10%
0.37397657   0.22572266    1.5570987   0.87249034

Note that this compares the precision of the two calibrations (how well each can reproduce the target colors). The absolute accuracy will depend on what you have referenced the colorimeter to.
all clear and ConnecTEDDD like this.

Last edited by zoyd; 05-25-2015 at 10:24 AM.
zoyd is online now  
post #744 of 747 Old 05-21-2015, 12:04 PM
 
all clear's Avatar
 
Join Date: Sep 2008
Posts: 341
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 166 Post(s)
Liked: 69
great work thanks!

as for gamma im not quite happy with it.
even when the 3D LUT is loaded i get a gamma at 2.4 for the low IREs 5%-10% while the rest of the curve is flat at 2.3
i dont know if that gamma error is shown in these results or not.

as for the curve at the chart it reminds more of the curve you got with 10000 patches right?
all clear is offline  
post #745 of 747 Old 05-21-2015, 12:24 PM - Thread Starter
AVS Special Member
 
zoyd's Avatar
 
Join Date: Sep 2006
Location: Planet Dog
Posts: 5,858
Mentioned: 65 Post(s)
Tagged: 0 Thread(s)
Quoted: 984 Post(s)
Liked: 826
Quote:
Originally Posted by all clear View Post
great work thanks!

as for gamma im not quite happy with it.
even when the 3D LUT is loaded i get a gamma at 2.4 for the low IREs 5%-10% while the rest of the curve is flat at 2.3
i dont know if that gamma error is shown in these results or not.
They're in there but you won't see much of a shift in the histogram unless your average gamma is significantly different than your target.

Quote:
as for the curve at the chart it reminds more of the curve you got with 10000 patches right?
It's actually only 2661 patches. The reason it's almost the same as the 10k run is because the starting conditions for the two calibrations were different. I started the 10k with higher errors so that I get some measurable differential between a 3k and 10k run.
zoyd is online now  
post #746 of 747 Old 08-24-2015, 10:36 AM
AVS Addicted Member
 
Join Date: Sep 2002
Location: Wetumpka, AL
Posts: 16,163
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 642 Post(s)
Liked: 337
Quote:
Originally Posted by zoyd View Post
I've compared the most recent versions of dispcalGUI (DCG)/ArgyllCMS, LightSpace, and CalMAN using the eeColor on a plasma display
Thanks for all the hard work running these test.

I'd like to find out if since the last test was run in April, if there had been any improvements in Calman's LUTs program where it would essentially produce the same LUTs as DCG/Argyll.

I don't expect you to run the test every few months but perhaps we can bribe you to rerun it sooner rather than later.

I prefer using a pattern generator, otherwise, I'd give DCG/Argyll a spin.
JimP is offline  
post #747 of 747 Old 08-24-2015, 10:52 AM - Thread Starter
AVS Special Member
 
zoyd's Avatar
 
Join Date: Sep 2006
Location: Planet Dog
Posts: 5,858
Mentioned: 65 Post(s)
Tagged: 0 Thread(s)
Quoted: 984 Post(s)
Liked: 826
I did check one of the 5.5.2 betas in July and there were no improvements compared to the April performance, the release notes also do not mention any improvements in this area.
JimP likes this.
zoyd is online now  
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