or Connect
AVS › AVS Forum › Display Devices › Display Calibration › ee/ColorBox, LightSpace, and 3D LUT Calibration
New Posts  All Forums:Forum Nav:

ee/ColorBox, LightSpace, and 3D LUT Calibration - Page 5

post #121 of 253
Quote:
Originally Posted by ConnecTEDDD View Post

The same Output.... Do you think someone who has bought QD 780 don't know how to setup a comparison?

I don't "assume", I ask. wink.gif

ss
post #122 of 253
Quote:
Originally Posted by sotti View Post

I think the amount of drift here is likely being exaggerated a bit. The amount of drift even over several hours may take a dE of 0.2 and move it to 1.5. But when you turn it back on it might be 0.8, and then warm up to 0.6. It's not as if the display constantly floats further away from the calibrated values. It's not as if you start with a dE of 5 calibrated it to 0.3 and by the end of the day it's back to a dE of 8.


I am trying to understand but as I see it there is no way to know at the time of meter reading for a single point whether the normal error fluctuation you describe above is at its midpoint or an extreme.

Is there not a danger that points may be incorrectly assumed to be in need of adjustment if only one reading is taken?

I accept that warming up a display helps but surely there is still a non constant output from any display that needs to be accounted for somewhere.
post #123 of 253
I have measured the repeatability of single measurement profiles with the D3 meter over periods of several hours, which include variations in the display output as well as tearing down and setting up the equipment, and the mean repeatability error over a test set which included 750 colors optimally spaced for such a measurement was 0.41 dE94.
post #124 of 253
Quote:
Originally Posted by zoyd View Post

I have measured the repeatability of single measurement profiles with the D3 meter over periods of several hours, which include variations in the display output as well as tearing down and setting up the equipment, and the mean repeatability error over a test set which included 750 colors optimally spaced for such a measurement was 0.41 dE94.

Many thanks for that data Zoyd.

Was there any pattern to the repeatability (eg certain areas of the Cube better / worse than others or was 0.41 dE representative of all points checked?
post #125 of 253
Quote:
Originally Posted by zoyd View Post

and the mean repeatability error over a test set which included 750 colors optimally spaced for such a measurement was 0.41 dE94.

Also with a plasma tv ?
post #126 of 253
Quote:
Originally Posted by work permit View Post

Thank you both for your responses.

By "do it manually", do you mean using a calibration disk and stepping through each color pattern? For me, that would get pretty old pretty fast, and so I was interested in a more "automated" setup. I never had a good experience using a laptop as a color pattern generator. From reading the documentation on spacematch, it appears to resolve some of the issues I've had in the past, specifically with the windows color management interfering with the generator. Does spacematch allow for custom refresh rates (ie 1080p/24)?

Are there any other goodies to get, and issues I should be aware of?

What specific issues did you have with windows color management?

I have tested my graphics card against AVSHD generated patterns and am pleasantly surprised by the results.
I use the free Xrite tool to advantage during the process.
post #127 of 253
Quote:
Originally Posted by PE06MCG View Post

Many thanks for that data Zoyd.

Was there any pattern to the repeatability (eg certain areas of the Cube better / worse than others or was 0.41 dE representative of all points checked?

The 0.41 dE is a global average and good indicator of the influence of systematic (non-random) errors like reproducing the same set-up, display drift, probe drift, etc. Specific luminance levels have higher or lower dE reproducibility due to the influence of random (sensitivity limit) errors. So for example a single measurement of a dark color might have a reproducibility of 1.5 dE in that distribution while brighter colors will get closer to 0.1 or 0.2


All the measurements were done on a plasma.
post #128 of 253
Quote:
Originally Posted by PE06MCG View Post

Is there not a danger that points may be incorrectly assumed to be in need of adjustment if only one reading is taken?

1) When dealing with random numbers, any single reading is likely to be representative of the actual value as any other reading taken in isolation.

2) We are talking about randomness that is typically below human perception. So it really doesn't matter too much if it's 0.2, 0.8 or 1.5, you're not going to be able to differentiate between the measured color and the target color anyway.
post #129 of 253
Quote:
Originally Posted by sotti View Post

1) When dealing with random numbers, any single reading is likely to be representative of the actual value as any other reading taken in isolation.

2) We are talking about randomness that is typically below human perception. So it really doesn't matter too much if it's 0.2, 0.8 or 1.5, you're not going to be able to differentiate between the measured color and the target color anyway.

Don't get me wrong Joel, I am in awe regarding the degree of accuracy the meters and any associated software can produce but point number 2 in your answer is really the criteria most of us are interested in.
Certainly the Display and Camara manufacturers and probably program Producers take advantage of this human eye frailty.
post #130 of 253
Quote:
Originally Posted by sillysally View Post

^^^^

Ted what settings are they using in the oppo's, are they using source direct, RGB,4:2:2, 4:4:4, ect?
Can't read Chinese.

ss

sillysally,

set the Oppo to 444 (if you want the Oppo to do the upsampling), the VT50 can handle that signal - I've tested it with the Mini (some of the older TV's can't handle 444)...

the real question is if the color box is using and outputting 444 or 422 and then also if the 444 upsampling should be done by the eeColor box or the Oppo... u wanna have it done by the superior chipset...

- M
post #131 of 253
Quote:
Originally Posted by ConnecTEDDD View Post

Generally the 1D LUT or 3D Cube calibration using the Blu-Ray as a Source is more accurate that any pattern generator of any price range because All Blu-Ray Players Outputs are different and not so accurate as we know.

For Example, as we all know currently the OPPO Blu-Ray Players are the Most Reference Players right now, but see that comparison someone has run using your QD 780 Singal Generator vs. OPPO-BDP-105 & BDP-103.

http://www.my-hiend.com/vbb/showthread.php?6671-【訊源】OPPO-BDP-105-BDP-103-儀器實測分析

Calibrations using the Ted's LightSpace Calibration Disk (for 3D LUT with LightSpace only) or AVSHD/GDC/etc (for 1D LUT with CalMAN/ChromaPure/HCFR) will fix problems that players are introducing.

That works very well when your blu ray player is your only source. Unfortunately my typical setup includes a dune player, a roku, an apple tv, and a cable box (in addition to a blu ray player). While ideally each input would havenits own full calibration, I personally found its easiest to calibrate the display to a a single reference and tweak basic contrast/color/brightness as needed in the avr.
post #132 of 253
Quote:
Originally Posted by work permit View Post

That works very well when your blu ray player is your only source. Unfortunately my typical setup includes a dune player, a roku, an apple tv, and a cable box (in addition to a blu ray player). While ideally each input would havenits own full calibration, I personally found its easiest to calibrate the display to a a single reference and tweak basic contrast/color/brightness as needed in the avr.

I think what you do is the most practical for a home setup. It's what I do. That's why I think the "error budget" you specify in the calibrations is important. What you describe is just one more thing that pulls the image off its calibration. The closer to the spec you start the more room you have for drift and different sources.
post #133 of 253
Quote:
Originally Posted by work permit View Post

That works very well when your blu ray player is your only source. Unfortunately my typical setup includes a dune player, a roku, an apple tv, and a cable box (in addition to a blu ray player). While ideally each input would havenits own full calibration, I personally found its easiest to calibrate the display to a a single reference and tweak basic contrast/color/brightness as needed in the avr.

yeah Ted is pointing out the best case scenario of a single chain which will almost yield the best possible results (you can see the results achieved by LS on his Kuro in a previous post)... for a multi chain setup (mine is similar to yours with tons of media players and other source devices spread across every room) possibly including devices that simply can't be calibrated (DVR | cable box | etc) simply use a reference pattern generator (just gotta make sure it is supported by the application)...

but as a side note, for media players:

if you want to calibrate that chain (in case a media player has a stronger distortion) u can use the m2ts files from Ted's disk and play them through the media player to calibrate that chain via LS and store that LUT in a different slot... it will only be DIP mode (time based) in LS and therefore take longer, but that's pretty useful...
post #134 of 253
That is a very good idea. For 1d calibrations i use the avs hd 709 in my blu ray and a ripped version on the dune to test how far off each source is from reference. For 3d calibrations i can see using teds disk to check for 3d deviation in the blu ray player and the dune player. Unfortunately for the online streamers all i have is the video test pattern on netflix.
post #135 of 253
Quote:
Originally Posted by ConnecTEDDD View Post

The problem with CalMAN approach (generally that type of method) of multi measure and realtime correct each color point is that each correction interacts to other color points also. Whats why the final calibrated results after performing an AutoCAL is different with the results of re-measuring again the same display exact after the end of the AutoCAL.

If you compare that 2 reports you will notice differencies.

This is a known and reported issue that CalMAN's AutoCAL proccess that users have already reported, it's not something new.

Sorry but my understanding is If this is true, that the calibration of one color interacts with other colors in the cube, the Light Space concept can´t work.
post #136 of 253
Quote:
ArgyllCMS is a 3DLUT approach, equivalent in it's math to both CalMAN and LightSpace, performs just as well, has the additional capability of doing CIECAM02 appearance matching, and it's free. The .icm profiles it uses to link source and target color spaces are just a data path for exactly the same XYZ measurements and manipulations that LightSpace uses.</</div>

That is a very incorrect statement.
All that is similar is that all the systems generate 3D LUTs for final calibration - that way those LUTs are generated from the measured/profile data is vastly different.

An ICC profile can contain a whole lot of stuff - and it may contain a 3D LUT, some matrix and 1D LUTs. This is exactly what a profile is, it describes a transform from an input colour space to an output colour. But it does this in a very different way, and with very, very different colour mathematics.

The hard bit is not in running images through a LUT, it's generating the LUT from the probe data, and LightSpace uses propriety algorithms to do this in a very unique way, that has no relationship or association to the way any other system works.

To be fair, there are some potential limitations with ICC profiles, as they force you to use a certain structure. For example, you need to fold the conversion to pcs space (LAB or XYZ) into the profile which can be quite messy, and the internal 3D LUT/Matrix/1D LUT structure is not ideal in structure order for the best application for calibration.

That's not to say other systems can't be used for calibration - they obviously can (although we would say within limits) - but it is not correct t say they are in any way the same!
There is absolutely no common code or algorithm between LightSpace and any other calibration system.

Just saying.

wink.gif

Steve
post #137 of 253
Quote:
Originally Posted by fredman2005 View Post

Sorry but my understanding is If this is true, that the calibration of one color interacts with other colors in the cube, the Light Space concept can´t work.

Sure, it works. I have tested it several times wink.gif

From the profiling data LightSpace pre-calculates the non-linear errors of the display and counters them within the 3D LUT.

When you measure a specific colour, reading the displayed value, you are obviously seeing the display error for that colour as you know the source colour and the measured value. When you do that with all colours in the profile, and assess all the points by comparing the individual colour errors, you can calculate the non-linearity for each and every point within the cube colour data.

This is possible only because the cube profile data is fully complete 'before' the calibration is performed.

Another reason why separating profiling from calibration is so important.
post #138 of 253
Quote:
Originally Posted by fredman2005 View Post

Quote:
Originally Posted by ConnecTEDDD View Post

The problem with CalMAN approach (generally that type of method) of multi measure and realtime correct each color point is that each correction interacts to other color points also. Whats why the final calibrated results after performing an AutoCAL is different with the results of re-measuring again the same display exact after the end of the AutoCAL.

If you compare that 2 reports you will notice differencies.

This is a known and reported issue that CalMAN's AutoCAL proccess that users have already reported, it's not something new.

Sorry but my understanding is If this is true, that the calibration of one color interacts with other colors in the cube, the Light Space concept can´t work.

To give some more detail about the LightSpace Software, because i believe maybe a lot of users here think that is a new software that entered the HT market late 2012....

Lightspace comes from professional market and it's used @ post production houses from 2009 only for 17-Point Cube (4913 Color Points) Profilings since that is the Stardard Cube Size for Critical Color Applications @ Professional World.

Color Scientists from Professional World has decide for that numbers of Color Points (4913) before a lot of years, this is the Default Cube Resolution @ Post Production Studios, they don't accept lower color points.

Think that they have decide for that 4913 Color Number for the Post Production Displays they are using which the cheapest model costs around 20.000$.....these post porduction displays have the best technology available and they have very linear/stable response.

The Question here is why they used so many colors as a default and not 778 or 1500 as they working only with the best available displays?

If the professional displays of 20-50.000$ needs 17-Point Cube Profling, Our Consumer World Displays will need 17-Point Cube as a minimum!

LightSpace gives you the option to run a profiling from 3/4/5/6/7/8......17-Point Cube but only the 17-Point is recommended.

I have Evaluate THX CineCube HD also, It has 2 Options for Cube Resolution: Quick Profiling (1000 Color Points) and Full Profiling (4913 Color Points).

Seperating Profiling from Calibration is the way that the 3 major Software Companies from Professional World are using, THX CineCube HD, FilmLight TruLight & Light Illusion LightSpace CMS.

LightSpace's Chief Colour Scientist is the same person that developed the Colour Pipeline for Industrial Light & Magic (ILM: George's Lukas Motion Picture Effects Company), WETA Digital (Peter's Jackson Visual Effects Company) & Park Road Post (Peter's Jackson Post Production Facilities). Their Chief Colour Scientist was also responsible for the colour science in TrueLight.

LightSpace is already installed at the most popular Post Productions Facilities around the world, including Warner Bros, Universal Studios, Walt Disney, Dolby, BBC etc...

I found your statement 'the Light Space concept can´t work' very funny after reading your post, since I have experienced the difference of using LightSpace (using it 4 months already).....also used CalMAN/ChromaPure for the last 3 years.
Edited by ConnecTEDDD - 3/31/13 at 6:48am
post #139 of 253
Quote:
Originally Posted by work permit View Post

That is a very good idea. For 1d calibrations i use the avs hd 709 in my blu ray and a ripped version on the dune to test how far off each source is from reference. For 3d calibrations i can see using teds disk to check for 3d deviation in the blu ray player and the dune player. Unfortunately for the online streamers all i have is the video test pattern on netflix.

The problem is that the Consumer Player are not so accurate so it will distort the final calibration because of their inaccurate output.

These are examples of Consumer Blu-Ray Players that tested using Quantum Data HDMI Analyser:

http://www.my-hiend.com/vbb/showthread.php?6671-【訊源】OPPO-BDP-105-BDP-103-儀器實測分析

http://www.hometheaterhifi.com/images/stories/2011/september-2011/panasonic-dmp-bdt210-blu-ray-player/panasonic-dmp-bdt210-hdmi-benchmark-tables-lg.jpg

http://www.hometheaterhifi.com/images/stories/2011/november-2011/lg-bd670-blu-ray-player/lg-bd670-hdmi-results-lg.jpg

http://www.hometheaterhifi.com/images/stories/2011/december-2011/samsung-bd-d6500-blu-ray-player/samsung-bd-d6500-hdmi-analysis-lg.jpg

http://www.hometheaterhifi.com/images/stories/2011/july-2011/samsung-bd-d5500-blu-ray-player/samsung-bd-d5500-hdmi-test-results-lg.jpg

http://www.hometheaterhifi.com/images/stories/2011/june-2011/sony-bdp-s580-blu-ray-player/sony-bdp-s580-hdmi-results-lg.jpg

As you see , all players are changing the signal at their output, so calibrating 3D Cube using a Disk as a Source is the most accurate way of calibrating a consumer setup calibrating and correcting errors to the whole chain.

A 100.000$ Lab Grade Pattern Generator will aware all these player problems, you can't see or fix them. The only way currently to Fix these problems is to use Ted's Calibration Disk - LightSpace Software Combination, if you need the best possible results from your current Blu-Ray Player you are using as a source.
post #140 of 253
Ted,

Does the eeBox support multiple devices or do I have to buy one eeBox per device ? I wanted to calibrate a projector and a plasma tv.
post #141 of 253
Quote:
Originally Posted by Ranger View Post

Ted,

Does the eeBox support multiple devices or do I have to buy one eeBox per device ? I wanted to calibrate a projector and a plasma tv.

Color Box has 1 Input - 1 Output & 6 Different LUT Memories.

Fro Those who have more devices, you have buy a HDMI Splitter 1x2, to feed from 1 Source 2 Different Displays. wink.gif
post #142 of 253
Quote:
Originally Posted by ConnecTEDDD View Post

Color Box has 1 Input - 1 Output & 6 Different LUT Memories.

Fro Those who have more devices, you have buy a HDMI Splitter 1x2, to feed from 1 Source 2 Different Displays. wink.gif

So for 2 different displays, do I still get 6 different LUT memories per display or will eeBoX use LUT Memory 1/2 for projector, LUT Memory 3/4 for plasma tv, etc ? Trying to understand this color box works. Thanks.
post #143 of 253
1x HDMI IN and 1x HDMI OUT, 6 memories for LUTs for the HDMI OUT. So you have to switch between the memories of the eeColorBox.

Have a look at the user guide: Click !
post #144 of 253
Quote:
Originally Posted by Ranger View Post

Quote:
Originally Posted by ConnecTEDDD View Post

Color Box has 1 Input - 1 Output & 6 Different LUT Memories.

Fro Those who have more devices, you have buy a HDMI Splitter 1x2, to feed from 1 Source 2 Different Displays. wink.gif

So for 2 different displays, do I still get 6 different LUT memories per display or will eeBoX use LUT Memory 1/2 for projector, LUT Memory 3/4 for plasma tv, etc ? Trying to understand this color box works. Thanks.

eeColor has 6 LUT memories for that 1 Input - 1 Output.

You can use a HDMI Switcher to the eeColor's Input to have more Input Devices and an HDMI Splitter at the eeColor's Output to have more Displays connected.

The 6 LUT Memories are global, you can manage the selected memory from the eeColor's Remote.
post #145 of 253
Thread Starter 
Quote:
Originally Posted by Ranger View Post

So for 2 different displays, do I still get 6 different LUT memories per display or will eeBoX use LUT Memory 1/2 for projector, LUT Memory 3/4 for plasma tv, etc ? Trying to understand this color box works. Thanks.

There are six LUT slots total. Assuming the same content source, you would need an HMDI splitter to send the processed LUT signals to two displays.
post #146 of 253
Thread Starter 
3 replies in two minutes. Question answered. biggrin.gif
post #147 of 253
Thread Starter 
If any of you are going to NAB, Light Illusion will be there.
post #148 of 253
Quote:
Originally Posted by buzzard767 View Post

There are six LUT slots total. Assuming the same content source, you would need an HMDI splitter to send the processed LUT signals to two displays.

Thanks guys. I am a bit slow this morning.
Just to clarify, once the projector and plasma have been calibrated with LS and both profiles loaded into the eeBox, the eeBox will detect what display I am using at the moment and apply the correct profile. Is that correct ?

Buzz, I am just up the road from you. biggrin.gif
post #149 of 253
Quote:
Originally Posted by Ranger View Post

Quote:
Originally Posted by buzzard767 View Post

There are six LUT slots total. Assuming the same content source, you would need an HMDI splitter to send the processed LUT signals to two displays.

Thanks guys. I am a bit slow this morning.
Just to clarify, once the projector and plasma have been calibrated with LS and both profiles loaded into the eeBox, the eeBox will detect what display I am using at the moment and apply the correct profile. Is that correct ?

Buzz, I am just up the road from you. biggrin.gif

You have to manually select each memory.
post #150 of 253
Thread Starter 
Quote:
Originally Posted by Ranger View Post

Thanks guys. I am a bit slow this morning.
Just to clarify, once the projector and plasma have been calibrated with LS and both profiles loaded into the eeBox, the eeBox will detect what display I am using at the moment and apply the correct profile. Is that correct ?

Buzz, I am just up the road from you. biggrin.gif

No. A 17x17x17 point color profile is made from each display. The profiles are then used to produce look up tables. The LUTs are loaded into the eeBox and from that point you must select which LUT you want to process the signal using the remote control.

Edit: You must use the HDMI splitter so that the selected LUT is going to the proper display.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Display Calibration
AVS › AVS Forum › Display Devices › Display Calibration › ee/ColorBox, LightSpace, and 3D LUT Calibration