AVS Forum banner

201 - 220 of 233 Posts

·
Registered
Joined
·
9,850 Posts
Discussion Starter #202 (Edited)
ColourSpaceZRO Beta 1.0.0.899 (04 February 2021)

Release Notes


Enabled Patch Scale for in-built Test Patterns. (1.0.0.899)

Restricted Preset Patch Lists to match set Patch Scale values. (1.0.0.899)

Added additional options to Virtual Probe to generate near-perfect profiles. (Shows what a near-perfect display would look like in the various graphs) (1.0.0.885)

Added Sub-Range functionality when Patch Scale set to values other than 0-255 (1.0.0.885)

NOTE: New Download Instructions

Please read about the new download instructions here. (Please refresh your browser cache to see the new instructions)

Installation Notes


ColourSpace Installer will completely overwrite any previously installed version, not just update it.

It will not require to un-install older releases before installing a newer version.

Update Notes

While ColourSpace is working in beta, when the users will start ColourSpace, it will not pop-up notification to inform them about the availability of a newer beta.

However, I will update this thread but also refresh the form page with release notes text per each beta.



* IMPORTANT CHANGE FOR PATCH SCALE*

ColourSpace with any release before 1.0.0.885 worked only with 0-255 custom import patchsets (Manual Measure -> Presets).

The RGB Sliders were displaying 0-255 range of patches.

These 0-255 values were scaled according to the patchscale setting @ Settings of ColourSpace.

It means that 0-255 values scaled to 16-235 for the PGenerator with 16-235 as patchscale, and the PGenerator was generating 16-235 RGB triplets.

0-255 values scaled to 16-255 for PGenerator with 16-255 as patchscale, and the PGenerator was generating 16-255 RGB-Triplets.

When we wanted to use a specific 8-bit RGB-Triplet to create a custom patchset, we had to convert/scale the 16-235 RGB-Triplet values to 0-255 RGB-Triplets, since ColourSpace wanted a 0-255 patchlist.

That procedure was adding a rounding error due to the additional scaling of the custom patchset 0-255 to 16-235; it had a small deviation to the calculation of the target patch xyY for specific color triplets (not all).

That issue affected only manual calibration or verification of the calibration (not profiling of 3D LUT feature of HTL and higher license levels).

After ColourSpace 1.0.0.885 release, ColourSpace has disabled that patchlist scaling.

It means that if you want to calibrate for 16-235 levels, you have to use 16-235 triplets to your custom patchlist and use 16-235 Patchscale @ ColourSpace Settings.

If you want to calibrate for 16-255 levels, you have to use 16-255 triplets to your custom patchlist and use 16-255 Patchscale @ ColourSpace Settings.

In other words, your patchlist values have to match your selected Patchscale setting since ColourSpace has disabled any scaling to provide bit-perfect patch generation.

LightIllusion has updated the following related website pages:

Patch Scale Options

HDMI User Guide for Direct Profiling



SDR Calibration Instructions

Now the RGB sliders will lock to the range of the selected patchscale setting.

For 16-235 Patch generation:

When 16-235 Patchscale used @ ColourSpace Settings, the RGB Sliders in Manual Measure will lock to any selection between 16-235.

Using Target Luma 100 nit (max / Peak) 0 nit (min / Black), 235.235.235 selection of the slider will have target Y of 100 nits.

PGenerator will output the exact triplet you see to the RGB-Sliders, using 1.0.0.885 (or newer release)

About custom patchsets, users have to use RGB-Triplets with 16-235 values.

For 16-255 Patch Generation:

When 16-255 Patchscale used @ ColourSpace Settings, the RGB Sliders in Manual Measure will lock to any selection between 16-255.

For REC.709 with gamma 2.4 as target colorspace and Target Luma 123.336 nit (max / Peak) 0 nit (min / Black), 235.235.235 selection of the slider will have target Y of 100 nits, and 255.255.255 of the slider will have a target Y of 123.336 nits.

For REC.709 with gamma 2.2 as target colorspace and Target Luma 121.199 nit (max / Peak) 0 nit (min / Black), 235.235.235 selection of the slider will have target Y of 100 nits and 255.255.255 of the slider will have target Y of 121.199 nits.

PGenerator will output the exact triplet you see to the 16-255 RGB-Sliders, using 1.0.0.885 (or newer release).

About custom patchsets, users have to use RGB-Triplets with 16-255 values.



For Ted's Calibration Disk users (PGenerator / Blu-Ray / Media Files) Instructions:

Ted's patchsets will require ColourSpace 1.0.0.885 or newer.



These patchsets provide the capability for users of Ted's Bluray / Media Files / PGenerator to measure using ColourSpace all the CalMAN measurement options available to Ted's Calibration Disk.

It will also provide the capability to Ted's calibration disk users to compare different meter settings between ColourSpace / CalMAN, to see about how each software is handling their meter.

I have updated the patchsets for ColourSpace users; please contact me here to provide you a new download link (not via PM).

Ted's Patchlists for ColorSpace use Video (16-235) or Video Extended (16-255) RGB-Triplets for consumer TV/Projectors calibration. (It's included an additional folder with 0-255 patchsets for PC calibration users.)

Please set Patch Scale 16-235 @ ColourSpace Settings window for bit-perfect patch generation/calculation of charts/dE/color patch targets.

Patchscale 16-235 required for the color engine to render the charts, populate the dE and display with accuracy the color targets.

I have added also the Dynamic Range (Clipping with Peak) and 2-3-4-Point Grayscale patchsets, please set your Patch Scale 16-255 @ ColourSpace Settings only for that patchset, and change the Target Luma according to the 16-255 provided info.

About how to import a patchset, see a picture example here.



HDR10 Patchsets for ColourSpace users

I have updated for ColourSpace users HDR10 patchlist to verify the pre-post calibration with 4/5-Point Saturation Sweeps @ 100 / 203 nit of REC.2020 or D65-P3 using REC.2020 as a container or using ColorChecker @ 100 nits.



Select as Target Colorspace ST.2048 REC2020.

You can download the zip for these patchsets here.

When ITU's BT.2408 specifications were released, Diffuse White became 203 nits, so 58% of the full PQ signal (input) level.



PGenerator Settings for HDR10 with HD Fury

When you are using PGenerator, set the Pi3 Device to output RGB-Full and patchscale 16-235 @ ColourSpace Settings.

It will require to use of an HD Fury device (Linker, Integral, Vertex, etc.) to inject HDR10 metadata and RGB-Video 1080p24 AVI infoframe.

About HD Fury HDR10 Metadata and AVI infoframe injection, use the info provided to the PGenerator thread.



PGenerator Settings for HDR10 with LG 2020 OLEDs using HDMI Override (without using HD Fury)

LG 2020 OLEDs have a hidden HDMI override menu,

It will allow the configuration of the HDMI input metadata by overriding it.

It's the available configuration of Colorspace, EOTF, ST2086 metadata, also MaxCLL, MaxFALL.

While you send an SDR signal using PGenerator (with an HDR10 patchset), highlight the picture mode in the main menu, and press ''1113111'' from the TV remote.

The HDMI Override menu will appear.

Then set these options for HDR10 calibration/verification:



Configuration settings can be stored per HDMI input.

It will not require to have HD Fury when you have only a 2020 LG OLED to calibrate/verify HDR10. (for LG OLED 2018/2019, it will require HD Fury)



HDR10...but why 100 or 203 nits?

In the early years of HDR, the recommendation during scene capturing from cameras and colorists use 50% of the PQ as ''Diffuse White''.

50% of PQ is 100 nits.

When ITU's BT.2408 specifications were released in 2017, the recommendation for 'Diffuse White'' became 203 nits, so 57% of the full PQ signal (input) level.




It's the recommendations while capturing content and color-grade for HDR10.

I have included patchsets for both 100/203 nits.
 

·
Registered
Joined
·
9,850 Posts
Discussion Starter #203
I have updated the download link for the HDR10 patchsets.

You can find these two additional CMS patchsets if you re-download the ZIP, useful for HDR10 manual cal:

HDR10 CMS 50% REC2020 @ 100 nits
HDR10 CMS 50% REC2020 @ 203 nits

(50% of WRGBCMY @ 100 nits, aka 50% SAT / 50% STIM REC.2020 HDR10)
 

·
Registered
Joined
·
26 Posts
I have updated the download link for the HDR10 patchsets.

You can find these two additional CMS patchsets if you re-download the ZIP, useful for HDR10 manual cal:

HDR10 CMS 50% REC2020 @ 100 nits
HDR10 CMS 50% REC2020 @ 203 nits

(50% of WRGBCMY @ 100 nits, aka 50% SAT / 50% STIM REC.2020 HDR10)
Hi,
Could The CMS patchests HDR used for HDR 10 CMS calibration?
 

·
Registered
Joined
·
9,850 Posts
Discussion Starter #205

·
Registered
Joined
·
9,850 Posts
Discussion Starter #207
Okay,btw
Can LG OLED'S HDR 10 be manual cal CMS?
I don't think it will work because LG has a broken CMS processing of the regular TV CMS controls.

Display an HDR10 pattern with color gradation, adjust (without a meter) the CMS controls, and check for added issues after these adjustments.
 

·
Registered
Joined
·
1,372 Posts
The small inaccuracy was affecting the verification (when Video Levels patchsets was used, for full range was OK), not the manual CMS calibration.

The issue has been fixed with ColourSpaceZRO 1.0.0.885 release, so now, it has bit-perfect the custom patchlist import and target calculation.

Now the patchsets are not scaled.

About the Dark Skin patch used as an example, at the right side of the picture, below you see the HDMI capture of the PGenerator signaling. (ColourSpace patchscale 16-235 / PGenerator RGB-Full)
On the older version of ColourSpace, I encountered color discrepancy as mentioned by @thoth when I first calibrated at video range (with 0-255 patches scaled by CS), then calibrated a PC input on the same display at PC range (full 0-255 patches without scaling). To my surprise, I found I needed to make adjustments to RGBCMY CMS controls, 75%/75% patches were off, all in a similar deviation from the video range settings. At the time, I passed this off as the display doing some transposing between limited and full range inputs.

Now, with the new CS, I checked again on both video range and full range. Guess what - now the same CMS settings work perfectly for both.

So I remeasured using old and new ColourSpace to confirm my settings. Plots below were taken with TV set to limited range, and CS set to 16-235 patch scale. No CMS or other TV settings were changed between the two plots. The difference should be solely from any patch scaling on the old CS.

Old CS with 0-255 patch list, 16-235 patch scale
3089526



New CS with Ted's revamped 16-235 patch list, 16-235 patch scale
3089541



I still had to make sure the TV didn't transpose or scale between video range and PC range inputs. I repeated the same measurements with the TV set to full range, and CS patch scale to 0-255. Rather than post another plot - I ended up with identical plots to the second one above.

The net is when I was using the old CS, I ended up making adjustments to "chase" a few more points, which turned out to be less inaccurate.

Thanks Steve, for the reworked patch scale, and Ted for supporting it!
 

·
Registered
Joined
·
9,850 Posts
Discussion Starter #209
ColourSpaceZRO Beta 1.0.0.936 (27 February 2021)

Release Notes


Update solely to keep sync with main code base. (1.0.0.899)

Fixed update issue with Point Info when single point displayed. (1.0.0.930)

Added colour coding and text info for clipped patched in Presets. (1.0.0.930)

Internal changes to enable Variable Bit Depth (not active in this build) (1.0.0.930)

* This build has huge changes under the hood, in readiness for Variable Bit Depth capability. It is likely to have bugs we have missed during our testing, so DO NOT UPGRADE if that could be an
issue!

NOTE: New Download Instructions

Please read about the new download instructions here. (Please refresh your browser cache to see the new instructions)

Installation Notes


ColourSpace Installer will completely overwrite any previously installed version, not just update it.

It will not require to un-install older releases before installing a newer version.

Update Notes

While ColourSpace is working in beta, when the users will start ColourSpace, it will not pop-up notification to inform them about the availability of a newer beta.

However, I will update this thread but also refresh the form page with release notes text per each beta.
 

·
Registered
Joined
·
9,850 Posts
Discussion Starter #210
ColourSpaceZRO Beta 1.0.0.949 (02 March 2021)

Release Notes


Added Variable Bit Depth – 8bit, 10bit, 12bit, 14bit, 16bit.

Corrected Probe Matching patch value to be 240 (8 bit).

Fixed potential issue with duplicate points in profile data when updating a measurement.

Corrected rounding error with measurement profile data.

* This build has huge changes related to the new Variable Bit Depth capability. It is likely to have bugs we have missed during our testing, so DO NOT UPGRADE if that could be an
issue!

NOTE: New Download Instructions

Please read about the new download instructions here. (Please refresh your browser cache to see the new instructions)

Installation Notes


ColourSpace Installer will completely overwrite any previously installed version, not just update it.

It will not require to un-install older releases before installing a newer version.

Update Notes

While ColourSpace is working in beta, when the users will start ColourSpace, it will not pop-up notification to inform them about the availability of a newer beta.

However, I will update this thread but also refresh the form page with release notes text per each beta.
 

·
Registered
Joined
·
9,850 Posts
Discussion Starter #212
I understand this to have no practical impact on those using PGenerator, as PGenerator is expecting only 8 bit values. Correct?
Hi,

That feature has no impact on ColorSpace ZRO and PGenerator users.

Raspberry Pi 1/2/3 can generate 8-bit only.
 

·
Registered
Joined
·
9,850 Posts
Discussion Starter #213
1000 Point volumetric verification run using ColourSpaceZRO

Hey ColourSpace ZRO users,

As you know, ColourSpace ZRO provides the capability to load custom patchsets; however, the maximum number of color patches is 256 per file.

I have recorded a video to show a 'trick' for measuring 1000 color points with ColourSpace ZRO.

Download Link: It will require downloading the 10-Point Cube, 1000 patches for Extended Levels(16-255) patchlists from here.

I used a 10-Point Cube Anisometric sorted patchset (1000 patches) and spitted to 4 smaller patchlist of 250 colors.

We use 1000 point patchlist to evaluate the performance of LG OLED, after an SDR 3D LUT using ColourSpace HTL (or another license level with LUTs generation).

You can see some example of ColourSpace users with 1000 point verification results below:

1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17



After an SDR 3D LUT from ~5000 colorpoints following ColourSpace Guides for LG OLED, it will provide a verification of 1000p with all green or max 5-6 points above 1dE2000, usually around 1.1 - 1.2 dE max, no error above 1.5dE.

You can use these results as a 'reference' and see how a LUT for LG generated from CalMAN performs.

Regular verification runs available to other calibration software are very limited to the range of measured colors.

When you have a small patchlist, like ColorChecker Classic / SG, it's not possible to identify LUT generation problems, like 'SharkFins' from CalMAN, for example.

That specific 1000 color patchset can measure such problems since it will measure a lot of edge colors and many different levels of hue/saturation/luminance range for a volumetric verification.



You can find some references of the SharkFin problems a lot of CalMAN users have posted below:

1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26

'SharkFin' issue of CalMAN is a problem introduced to the full saturated primaries/secondaries; these colors are not measured during the very limited verification patchlist of ColorChecker SG or Classic because they are not including any full saturated color patch.





You can see the difference of the measured colors using ColourSpace ZRO 2D CIE Chart between 10-Point Cube - ColorChecker SG - ColorChecker Classic to the APNG (animated PNG file) below:



You can see the difference of the measured colors using ColourSpace HTL 3D RGB Cube normalized chart between 10-Point Cube - ColorChecker SG - ColorChecker Classic to the APNG (animated PNG file) below:



That 1000 color patchset has 16-255 triplets since LG internal LUT is 16-255 (Extended).

It will require to use PGenerator with patchscale Extended / 8-bit.

Extended = 16-255 patchscale.

I have recorded a video to see how to perform that verification run or 1000 points.

Users of different TV/projector units can use that 1000 color (10-Point Cube) verification patchsets also, but be sure that your TV/Projector is not clipping WTW. (above 100%, to allow headroom and see flashing bars in Contrast/Color Clipping patterns above 235)


Please enable the Subtitles to see the explanation of many different plotting options of ColourSpaceZRO.

Subtitles were added after the video recording.

You can view the video with 0.5x speed or use 'pause' to read the text info if required. (or translation of the subtitles)

The video is useful even for users without plans to run large verification run because it will explain many different plotting options.

The measurement speed is fast because I used a virtual meter with 0 sec of extra delay.

While taking real measurements, you can navigate the software and change plotting options since ColourSpace provides that capability.

I will make the 1000 point patchset available with 16-235 (Legal) or Full Range (0-255) after some days.

That 1000 color patch verification run can be used by any TV/Projector user to perform a very deep analysis of the post-calibration performance after a good manual calibration.

Users can post their results after a manual post-calibration, so we will know how each different display/projector performs without a 3D LUT profiling.
 

·
Registered
Joined
·
1,372 Posts
1000 Point volumetric verification run using ColourSpaceZRO
Ted, thanks for this.

Here are the post calibration results for Sony 65X950H, in Rec.709 gamma 2.2. Manual calibration by me.

1000 points
3107357


1000 points without tangents
3107358


Primaries and secondaries only
3107359


Points with dE > 1.0
3107356


Edit: plots updated after my display stabilized and I reran the measurements.
 

·
Registered
Joined
·
1,372 Posts
Here are the post calibration results for Sony 65X950H, in Rec.709 gamma 2.2. Manual calibration by me.
A question for the more experienced calibrators -

The CMS blue hue setting that gives me the plots above results in a noticeable band in the color ramps gradient (the vertical color ramps pattern on Ted’s iso). However, fixing the band visually requires setting the hue slider in a position that results in higher dE for several points in the 1000 point color checker. Essentially the tangent for those points are now in the wrong direction instead of “on axis”.

Which one do I want to prioritize? Lower dE or a smoother gradient in the test pattern?
 

·
Registered
Joined
·
43 Posts
Which patch/triplets are recommended when doing a two point HDR white balance on an LG CX? Also, is disabling the tone mapping recommended?
 

·
Registered
Joined
·
9,850 Posts
Discussion Starter #217
ColourSpaceZRO Beta 1.0.0.994 (15 March 2021)

Release Notes


Enabled Full Screen option for Stabilisation patches.
Fixed BG colour value save/load when Resolution >8 bit.
Fixed issue with Sliders Step value when Patch Scale set to Legal or Extended.
Changed Slider operation when set to % to better reference consistent integer value.
Changed Virtual Probe to work with Variable Bit Depths, and change label to Rec709 Legal.

NOTE: New Download Instructions

Please read about the new download instructions here.

Installation Notes


ColourSpace Installer will completely overwrite any previously installed version, not just update it.

It will not require to un-install older releases before installing a newer version.

Update Notes

While ColourSpace is working in beta, when the users will start ColourSpace, it will not pop-up notification to inform them about the availability of a newer beta.

However, I will update this thread but also refresh the form page with release notes text per each beta.
 

·
Registered
Joined
·
180 Posts
Updated to 1.0.0.994 this morning to try to remedy an issue I was having with v949 last night. When choosing to do repeat measures on a single colored patch value from a Preset CSV, the color value changes after the first measure, and first "repeat". It either changes to another color in the CSV or to a gray scale patch with the associated values ( eg 180 180 180) I am using Ted's Patchsets posted in this thread for use with the versions >1.0.0.885.

This occurs both on the internal patch and via Pgenerator.

It occured on both versions 994 and 949.

It occurs only on colors, not gray scale measurements.

Choosing the respective color in Manual Measure via the sliders will allow Repeat measure as usual, but using the preset CSV value causes the issue.

Occurs in all the color spaces I used, Rec709, REC2020, etc.

It can be remedied by going to manual measure and unchecking the three "linking" boxes at the bottom, then going back to the CSV and choosing to Repeat Measure.

Here are couple videos of the issue:


 

·
Registered
Joined
·
9,850 Posts
Discussion Starter #219
A question for the more experienced calibrators -

The CMS blue hue setting that gives me the plots above results in a noticeable band in the color ramps gradient (the vertical color ramps pattern on Ted’s iso). However, fixing the band visually requires setting the hue slider in a position that results in higher dE for several points in the 1000 point color checker. Essentially the tangent for those points are now in the wrong direction instead of “on axis”.

Which one do I want to prioritize? Lower dE or a smoother gradient in the test pattern?
Hi,

Sony CMS engine, which calculates the UI controls' adjustments to correct the signal, adds some errors under a specific combination of these control values.

If these problems are visible with gradation patterns, they will appear in actual video content also.

It's better not to have such errors visually in your picture; try to find out which control values are introducing that problem and reduce the adjustments of controls to prevent it.
 

·
Registered
Joined
·
9,850 Posts
Discussion Starter #220
Which patch/triplets are recommended when doing a two point HDR white balance on an LG CX? Also, is disabling the tone mapping recommended?
Hi,

HDR calibration works in LEGAL Levels (16-235), 235 is 10000 nits in HDR PQ.

When you don't have ColourSpace HTL (or higher), which allows you to have access and disable the ToneMapping using LG 2020 Template, then you are limited to calibrate with enabled ToneMapping and 2-Point RGB Balance.

Its better procedure to not stay and calibrate manually 2 specific patches but to display a 21-Point Grayscale sweep, overview at the end of the measurements the RGB Balance chart, display a Black Frame using Ted's Template for the panel to cool down and open the TV OSD menu to apply the values you believe that will reduce the RGB balance errors.

After applying these values, close TV OSD Menus and repeat the full grayscale sweep again.

Perform that multiple times until the results are better, but look only RGB balance, not luminance errors.

You can download the HDR10 Patchsets from here.
 
201 - 220 of 233 Posts
Top