Steve:

I've gone through all the posts and phone conversations between us and some of the things I've read on your LS website at different locations and came up with the following list of things in one location you've said to consider for a quality LUT calibration. This is very raw. Pls provide feedback.

Kind Regards, JJ:

Light Space: Determining the LUT Calibration Quality … Raw Quality Related Statements

1. A good calibration system will provide both ways to overcome display and probe problems and the system understands what can, and what can't be calibrated, and does the best it can to calibrate what can, and doesn't allow what can't be calibrated to adversely affect the final result.

2. The aim of any calibration is to maximise the accuracy of the display within the limits of the display's capabilities, with a focus on the central core of colours within the display's required colour space. Light Space does this. Understanding this is very important, as many calibration systems focus on the extreme colour space colours (the primary colours) only. In the real world we see very few 'primary' colours, and focusing on them for calibration can often lead to very inaccurate final results, especially as many displays struggle with accurate primaries.

3. Additionally, calibrating at low-light levels can be very difficult with the more cost-effective probes available, with the probes introducing errors into the profile that do not accurately represent the display. Combine this with cheap displays that have a colour cast within their back-light, that shows in low-light conditions, but is overcome by the display as brightness is increased. Light Space understands the Display’s limitations and knows what can and cannot be calibrated. Such back-light cast colours cannot be 'colour corrected' out, although the profile data may cause other colour calibration systems to attempt to do so, causing inaccurate calibration results.

4. Use 1D-LUT Viewer to verify no 'kinks' in the shadow area that may introduce unwanted colour artifacts into the final calibration that’s introduced into the calibration by the probe's inability to accurately measure low-light levels.

4.1.1. If they exist, use the LUT Manipulation tools with LightSpace CMS as it is very easy to remove these errors, while keeping the underlying profile data.

4.1.1.1. For example, the LUT Manipulation 'Axis Blend' and 'Smooth' tools can be used to reduce the error.

5. The only way to check a LUT is to re-profile through it.

6. A correct profiling system, such as LightSpace CMS, will generate an accurate profile of any display, and from that show what is or is not possible. The only way to do this is with a set of very standard profiling data/image. From this all displays can be compared. It is key that any calibration system understands what can, and what can't be calibrated, and does the best it can to calibrate what can, and doesn't allow what can't be calibrated to adversely affect the final result.

7. Quick Profiling can be perfect for displays with good RGB Separation. 3D Cube profiling, Display Characterization, is for displays with poor RGB Separation. With Display Characterization, you'll need a minimum of a 10 or 11 point cube (11x11x11) for any level of accuracy ... but 17pt is the best!. Many people just think the calibration LUT size, but the profile LUT size can be as important. However, for many displays the approach taken by the Quick Profile version of LightSpace is perfect. Calibration has to be a single approach, and the profiling specifically needs to be a single contained operation. For example, LightSpace does not need to know any of the display target parameters before profiling. That is key to a good calibration system - and yet few understand the importance of that simple fact. After generating a profile from a display anything is possible as far as calibration goes. Making sure the profile is good is critical, and that is what LightSpace focuses on.

8. PCs and Macs use ICC profiles for display calibration. But they are 1D LUTs only. To do real calibration you need a 3D LUT box, such as Pluto or Davio, or HDlink Pro (when BMD fix the problem with that box). See the Light Illusion website for more info on ICC profiles (SpaceMan), and the Forums for info on the HDlink problem.

9. The best way to calibrate is not with reference material, but with the workflow and images outlined on the Light Illusion website. Images are just used to set brightness and contrast limits, while the actual calibration should be done with automated patches that cover the required gamma and gamut. That's it - nothing else required

10. A .bcs file is the profile data, not a LUT. You can only see that via the 'display' function (CIE and Gamma charts). You need to then use 'Convert Colour Space' to 'show' the profile data as a LUT relative to any given colour space.

11. If there are cross-colour issues they will show within the RGB Separation Graph when doing the first profile. From this you should see all you need to understand the initial profile, and then the capabilities of the final result.

12. Make sure there are no ICC profiles active within the LightSpace PC as part of your Pre-Calibration setup. Use SpaceMatch DCM to verify no active ICC profiles are present. The HDMI out from the PC must be clean, with no ICC profiles.


13. Color Separation, LCD Displays: First, most LCD screens with no attempt at calibration have good RGB separation. We hardly find any that don't have good Separation. Plasmas can be a problem - but even then it's not bad at all. ICC profiles are used within PC and Mac systems to try to calibrate their displays, so when using the HDMI out of a PC to do patches for calibration you must make sure there are no ICC profiles active - this can be a problem as they can be very, very invasive, especially if you have ever used a ICC calibration program. The attached can be used to check the output for active ICC profiles. Any ICC will be reactivated when the PC is rebooted, so be aware. The comment on reducing brightness is only needed if you are changing the colour temp (white point) there is NO other reason to do that, and it really should not be left to the LUT to do that, as that will change the overall contrast/gamma of the display, which is bad! Always best the get the white point set as accurately as possible using the available controls first.

14. Check for no ICC profiles active. You then need to set all display controls to default, and do a Quick Profile to see what you have. You are looking to have an even cube display when you make a Rec709 LUT from the profile. Not a perfect cube, but one that is even in shape, with no kinks. Like the 'before' cube (the middle one) here for example: *** Spam Blocker deleted link *** (Scroll down the page a bit to the 3 cubes). Ideally, the LUT should not touch the sides of the Cube - showing you have enough gamut for calibration. The Gamut profile (after the 3 cubes) shows what that should be like, again with a bit wider gamut ideally. And the Gamma as close as it can be to the target ideally. And the RGB Separation should look close - maybe a bit of variation in the shadows, but that's all. Basically a single line. When you have this with the display it can then be calibrated. Keep the LightSpace PC directly connect to the display - no Lumagen. To check how well it can be calibrated make the LUT from the final profile, after you have the above set correctly. Then load the LUT into the Active LUT menu and re-profile. The result should look like Cube 3 in the above images. The RGB Separation may change as the calibration may have to distort that to get the correct colours. But is should be obvious that the calibration is working.

15. It's probably worth saying something further about RGB Separation... After calibration the RGB Separation will nearly always be worse as the calibration corrects the underlying colours, which can distort the RGB Separation if the gamut is under range in any area - if the gamut is larger than the target, across the whole range, the RGB Separation will remain good, or at least not get any worse. But, a raw (uncorrected) display should have near perfect RGB Separation - not having that is a good way to know that the display's internal image processing is not ideal (actually, it means it's bad!). To check this try profiling a good PC display - a Dell or HP screen for example. As they have little internal image processing they show what a raw LCD display should really be like. This is quite an eye opener!

16. The point pattern in a LUT is the 'inverse' of the display properties. So, if a point has move further into the cube (move further 'inside' the cube) it is showing the measured point on the display had too much 'colour' (gamut/saturation), and the LUT is reducing it. If a point has moved towards the outside of the cube is showing the measured point had too little colour... Points squashed at the edge of the cube show where the display is under gamut overall (same as the points in the CIE diagram being 'within' the target gamut). I'll work on a crib-sheet page for the website! But, a way to see what different point movement means is to make a 'New' LUT (File/New) and then use the 'LUT Manipulation' tools to see what effect changing Saturation, etc, has on the cube. For full Display Characterisation there really isn't a way to know what size cube profile is needed - a rule of thumb is the worse the display the larger the point size required...

17. From phone discussion with Steve: Use “Out of Gamut” color conversion feature to create a LUT cube image to visually see what colors the Display cannot produce, because they are Out of Gamut.

18. From phone discussion with Steve: Use Bars and Black Reference Image viewed through your Verification LUT (VLUT, rec709) to check for clipping and scaling issues. Check for any consistency with issues found in Out of Gamut LUT Image.

Hey JJ, that looks good.
A couple of corrections -

4. (I modified) Use 1D-LUTs - should be Use 1D LUT Viewer...
7. 10 or 11 point is the absolute minimum - 17 is best!
(Quick Profiling can be perfect for displays with good RGB Separation - 3D Cube profiling is for displays with poor RGB Separation)
12. Use SpaceMatch DCM to verify no active ICC profiles.

But looks good!

Steve

As Steve said, loading LUTs into either version of the box is no problem except for having to rename LUTs to match those required by the ee software but it's no big deal.

I'm looking for the $300 box myself. Steve and I talked about this a year and a half ago. The stumbling block as I recall was HDMI licensing costs when the number of future sales is an unknown. You and I are at the front end of commercially producing 3D LUTs for the home theater market so it may take some time.

Buzz

Number 19. did not print all the way. Here it is:

19. From phone discussion with Steve: Use LUT modification subtraction feature to note differences between a Unity Cube (which represents a Calibration LUT, CLUT, with no modifications … which represents a perfect display) and your CLUT. Any points in the subtraction LUT not located in their “Unity position” represents a change needed for the display’s tristimulus value to conform to some color space standard … for instance, to rec709.

Thinking about it, point 19 is not necessary, as the LUT is already the inverse of the display, so is already showing the 'error'...

Subtracting the LUT (as it is an inverse of the display parameters) from a Unity LUT will result in the same LUT data as just 're-inverting the LUT. (You can use the 'invert' function to do exactly the same)

So either works.

But remember that the original LUT in the inverse of the display error, so both are showing exactly the same 'error' just with different views - one the inverse of the other!

Hope that makes sense.

Steve

Steve
Steve:

From your experience, when would you prefer a non-inverted vs. inverted Cal LUT. With little experience to draw from, I would expect a non-inverted CLUT Image to have some points outside the cube if colors are oversaturated in LS Pre-Cal Setup.

Kind Regards,

JJ

I only ever use the actual LUT to assess the display.
If the points are inside the LUT I know the display has more gamut than needed, and will probably calibrate well.
If the points are squished up against the cube edges I know the display is low on Gamut.
I would then reject the display....

I personally find no benefit in 'inverting' the LUT...

Down to each user I guess

Hi Guys,

I’ve list some things which are unique to the Light Space Pre-Calibration workflow process. What do you think?.

1. One can tell when the contrast levels are too low by the graying of white in the Bars and Black Reference Image. It can look like a scaling issue. This happened during trade offs among Back Light, Brightness and Contrast controls during Pre-Cal workflow to keep luminance at 100 nits. Contrast in the 70 -80/100 range … whites look healthy (Black Light = 0/10, Y=100nits). Contrast at 57 (Backlight =2/10, Y=100nits) … the whites were too grey/dull. Contrast can be too low.

2. 90 – 95% of D65 adjustments in LS’s Pre-Cal workflow can be done without use of RGB gain and bias controls. At this point in the workflow, in Live Feedback Mode, the gamut crosses were getting real close to being inside the gamut circle. The last 5 – 10% was still needed in order to be “just close enough” before starting the auto cal run. After the D65 bias and gain adjustments were finished, I used Chromapure to see what the grey scale dE’s were. They ranged from 6 -10% … i.e. plenty of margin! To answer Ted’s email from weeks ago, the focus is not on dE’s of 9%; rather, the focus in on the LS Pre-cal workflow process, which, if followed, will have, as in my case, D65 dE’s 6 to 10%. I sense LS software wants a lot of extra margin to play with in the beginning of the auto cal run … if so, one does not want to take away unnecessary margin during the Pre-cal workflow process. That trade off is what the gamut circle is all about. Get the crosses barely inside the circle, then your finished. This will leave plenty of margin to start the auto cal run. The unique result of the Pre-cal workflow process is a minimum amount changes to display controls will happen plus one will be “just close enough” plus enough margin is available for the auto calibration software to have the best chance for an accurate run. It just may turn out the dE’s were around 9% at the end of Pre-cal, as Ted noted in his email.

3. During LS Pre-cal Workflow: The Pre-Cal workflow requires bias (as opposed to gain) controls to be used first in making final D65 adjustments. This unique process can result in a lower gamma (~2.4) in the dark end and a higher gamma in the bright end (~2.2-2.3). This is because bias influences the bright end much more than gain influences the dark end. This actually complements the human eye in how it can distinguish “just noticeable differences” in dark versus bright environments. I’ve read calibrators usually force this condition during CMS if their display/projector is used in a dimly lit, light controlled room. All good …!

4. Selecting a wide color space during Pre-Cal workflow cannot be overly emphasized. There should be a list of some the popular displays and what display controls are used to select this … especially for the home theater market. It took me 1 week to figure this out because Sony likes to use vague words that can mean either color space or color depth. Usually forums help a lot.

Kind Regards, JJ

Good work JJ - much appreciated, and I am sure will help others a lot!

Steve,

Can the Display Port of a PC be used with an "DP to HDMI" adapter for Light Space? Or do you require the PC to have a physical HDMI port? Do you use the Display Port set up for your calibrations?

Kind Regards,

JJ

Display Port should be fine.

Diaplay Port, HDMI and DVI should all work as required, via adapters or direct.

Steve

Hi Steve,

HDCP handshaking btwn PC and Display needs to make all enhancement features selectable so they can be turned off as part of Pre-Cal workflow for a Profile cal run; otherwise, the Sony TV’s Digital Reality Creation Mode enhanced feature may be left on with no way to turn it off for a Profile cal run. Only a secure HDCP connection can make this enhancement feature selectable so it can be turned off. It’s entirely likely DRC was left on from the previous use of the HDMI port, say you watched a Blu Ray movie and you like DRC turned on. It stays on unless it's changed. Any Reset button does not affect DRC mode in my Sony TV. It can also be turned on automatically many other ways from previous uses of the HDMI port depending how Sony triggers unique calibration settings to be both stored and reactivated in/from memory. The only way to make certain all enhancement features are off is by controlling proper HDCP handshaking as part of a Pre-Cal workflow process. This will insure all enhance features are selectable and only then can these features be turned off by the calibrator prior to a Profile cal run.

Background: My Windows PC’s Display Port handshaking with the Sony TV (via a DP to HDMI adapter ... my PC does not have a physical HDMI port) does not allow Digital Reality Creation Mode enhanced feature to be selected in Sony's Custom Pic Mode. So if it was left on from the previous use of the HDMI port, it can’t be turned off. DRC is one of those enhancement features that must be turned off for Profile cal runs … especially Quick Profile.

Same is true with my Vista PC: it does not HDCP handshake with the Sony TV to make DRC Mode enhanced feature selectable so I can turn it off. The vista PC has a physical HDMI port built in. I did see this enhanced feature turned on during profiling, in Custom Pic Mode. I actually saw the control activation buttons used to modify DRC visibly washed out, but still visible. This means I cannot control them. I can't tell if DRC is on or off, but the washed out buttons showed it on. The only way to insure it’s turned off is through proper HDCP secure handshaking. This will make all enhancement features selectable and the calibrator can then turn them off. The reset button does not help.

Other observations:

1. You can turn off DRC with no HDMI cable connected; however this change is not permanent once you connect the HDMI cable between PC and TV.

2. I've noticed when my OPPO 83 Blu Ray player is connected to the TV, it allows the DRC to be selectable. I can choose to turn it on or off. My external pattern generator also does this. If fact, my external pattern generator triggers DRC Mode in all three Sony Pic Modes: Vivid, Std and Custom. Yes, Sony's internal calibration engine can be turned on in all 3 Pic Modes. Oddly, when no secure HDCP communication is going on to allow the DRC feature to be selected in Std and Vivid Pic Modes, DRC is not even a selectable feature … it’s not even listed on the feature menu.

3. The only way I can jury-rig the display to insure the DRC Mode is off prior to a Profile run is to connect my test generator to the TV’s HDMI port being used; this will make the DRC Mode feature selectable via a secure HDCP communications btwn pattern generator and TV and then I can turn it off. This change will hold when I disconnect the HDMI cable from the external pattern generator and reconnect the TV to the PC for the Profile cal run.


Kind Regards, JJ

Hi JJ,

HDCP should have no impact on the available controls within a TV.
It is just for content protection.
And PC with HDMI or Display Port will be HDCP compliant by default.

Are you meaning DDC/CI (EDID)?

Cheers,

Steve

Hi Steve,

Not sure what is controlling the ability to select the enhanced features on the Sony. My experience base is very limited in this area. Maybe the guys on this forum know.

Kind Regards,

JJ

Hi Steve,

I talked to the designer of the AccuPel 5000 pattern generator about the HDMI interface issue I'm having between the PC and the Sony TV. He said the issue is too complicated and could not offer a guess to what was going on. He also said PC companies are in the volume business, and conformance to all the video stds is not as big a demand from their customers as Price is ... so little incentive to conform and verify conformance to all video stds.

I'll talk to OPPO to see what they say. My guess, and my experience base is minimal here so I'm likely off base, is that some sort of minimal video stds needs to identified for the PC's HDMI interface to meet. The Sony Display knows what to listen for, on its HDMI interface, to trigger DRC ... and the PC isn't talking. AccuPel and OPPO-83 does this properly ... these are video products. It could also be something proprietary to the Sony Display and their HDMI interface ... but there are millions of Sony Displays out there ... all waiting to be LUT calibrated ... lol.

Kind Regards,

JJ