DisplayCAL 3.2.9 BETA

[Florian Höch]

I’m happy to announce that I’ve just released DisplayCAL 3.2.9 Beta for anyone interested in trying it out. The list of changes is huge, so I’ll refer to the changelog.

In a nutshell:

• XYZ LUT profiles created from small patch sets (79 and 175 patches) with regularly spaced grids (3x3x3 and 5x5x5) now have improved accuracy due to an alternate forward profiling method that works better for very sparsely sampled data. Most presets now use 5x5x5 grid-based patch sets by default that provide a reduction in measurement time at the same or even better accuracy than the previously used small patch sets.
• Generating high resolution PCS-to-device tables will now take more advantage of multiple processor cores or CPUs, giving a nice computation speed boost.
• OLED owners battling with ASBL (automatic static brightness limiting) messing up the display response may want to try one of the new profiling patch sequence advanced options and see if it improves the situation (please read the respective changelog entry).
• Automatic output levels detection, and manual output levels selection as an advanced option.
• The visual whitepoint editor now supports all available display choices (i.e. Chromecast, Web Browser, Prisma and Resolve pattern generators).

Changes from 3.2.8 Beta:

• Add HDR madVR preset and change video preset to use OFPS patches.
• Calibration whitepoint targets other than “As measured” will now also be used as 3D LUT whitepoint target, allowing the use of the visual whitepoint editor to set a custom whitepoint target for 3D LUTs.
• Mark settings as changed when altering 3D LUT HDR parameters.
• Save and restore content colorspace 3D LUT settings with profile.
• Interpolate regular grid A2B cLUT to higher resolution.
• Fix: Prevent Argyll CMS from removing measurements with one or two zero CIE components by fudging them to be non-zero.
• Fix: Do not remove temporary folder during increasing effective B2A resolution because this may be an intermediate step.

Changes from 3.2.7 Beta:

• Changed regular grid patch sets to 3x3x3 and 5x5x5 sizes and fixed slight inaccuracy of shaper curve interpolation for XYZ cLUT profiles created from these sets, and made these sets the default for most presets.
• Profile loader (Windows): Check whether or not madVR resets the videoLUT and preserve calibration state if not.

Changes from 3.2.6 Beta:

• Added support for madVR’s ability to send HDR metadata to the display via nVidia or Windows 10 APIs (i.e. switch a HDR capable display to HDR mode) when creating SMPTE 2084 3D LUTs. Note that you need to have profiled the display in HDR mode as well (currently only possible by manually enabling a display’s HDR mode).
• Changed automatic output levels detection to use absolute luminance in cd/m2 instead of relative luminance (more reliable detection).
• Suppress errors related to WMI (note that this will break getting the display name from EDID and prevent individual Argyll CMS instrument driver installation).
• Profile loader (Windows): Work around Windows 10 Creators Update videoLUT loading bug.
• Profile loader (Windows): Fix inability to add profiles due to bug introduced in 3.2.6 Beta.
• Profile loader (Windows): Fix undefined variable if launched without an active display (i.e. if launched under a user account that is currently not the active session).
• Profile loader (Windows): Fix original profile loader instance not closing after elevation if the current user is not an administrator.

Download:

Via Zero Install (all platforms)
Standalone (Windows)

For people that had problems with bad clipping due to the underlying display response having concave regions (although I’m still questioning whether that was due to ASBL or badly setup display controls messing up the display response), there’s also a Beta of Argyll 1.9.4 available (currently only for Windows):

Argyll CMS 1.9.4 Beta 2017-04-18

• This topic was modified 6 years, 11 months ago by Florian Höch. Reason: Version 3.2.7
• This topic was modified 6 years, 11 months ago by Florian Höch.
• This topic was modified 6 years, 11 months ago by Florian Höch.
• This topic was modified 6 years, 11 months ago by Florian Höch.
• This topic was modified 6 years, 11 months ago by Florian Höch. Reason: Version 3.2.9

[t.g.v]

Having tested last beta on W 10 Pro latest version. with argyll cms beta.

everything worked wo any problem. the  complete process is about 1,5 x faster than last regular version.

thankyou for the good work.

tg

[Florian Höch]

Thanks for the feedback, much appreciated 🙂

[S Simeonov]

Florian,

in Displaycal 3.2.4 the default auto optimized patches are 425, in the latest beta are 175. Why is the difference?

[t.g.v]

Hy everyone.

Having tested 3.2.9 on W10 pro with Asus PB248/MSI GT-740. No problems at all. Have overwritten files from original displaycal-installation 3.2.4 with 3.2.9 files and pointed to Argyll 1.9.4 beta.

I have no problems with fewer patches as default set. I do not need 2 hrs measurement.

thankyou for your good work.

tg

[Florian Höch]

in Displaycal 3.2.4 the default auto optimized patches are 425, in the latest beta are 175. Why is the difference?

Less measurement time, with around the same accuracy in many cases. Of course, you are absolutely free to increase the amount of patches if you want increased accuracy of the resulting profile.

[S Simeonov]

in Displaycal 3.2.4 the default auto optimized patches are 425, in the latest beta are 175. Why is the difference?

Less measurement time, with around the same accuracy in many cases. Of course, you are absolutely free to increase the amount of patches if you want increased accuracy of the resulting profile.

So in short 175 patches=425 patches in terms of accuracy?

[Florian Höch]

So in short 175 patches=425 patches in terms of accuracy?

If the display is fairly linear, yes, probably. Most computer monitors, especially those with IPS/MVA/PVA panels, should fall into that category. TVs on the other hand may not, due to internal processing, YCbCr to RGB conversion and the like.

[S Simeonov]

So in short 175 patches=425 patches in terms of accuracy?

If the display is fairly linear, yes, probably. Most computer monitors, especially those with IPS/MVA/PVA panels, should fall into that category. TVs on the other hand may not, due to internal processing, YCbCr to RGB conversion and the like.

Thank you for the reply 🙂

[t.g.v]

… one slight problem w. 3.2.9:

profile-loader doesn’t stay at taskbar; additionally profile-loader reports having reset profile some hundred times.

that’s a new behaviour.

regards – tg

[Florian Höch]

profile-loader doesn’t stay at taskbar

Not sure what you mean?

additionally profile-loader reports having reset profile some hundred times.

That’s to be expected. Calibration state will be reapplied each time a change in the running processes is detected, see the changelog for more details.

[S Simeonov]

Will there will be exe file (installation file) for the beta or just the zipped files?

[t.g.v]

icon of profile-loader can’t be “fixed” in taskbar. it shows up up at “ausgeblendete symbole”.

[Florian Höch]

icon of profile-loader can’t be “fixed” in taskbar. it shows up up at “ausgeblendete symbole”.

You can choose which icons will be always visible in the taskbar options. Works fine for me.

[Florian Höch]

Will there will be exe file (installation file) for the beta or just the zipped files?

An installer will be available for the final release.

[Author]

Posts