Originally Posted by Dominic Chan
That’s strange. I was able to open the other three using 3.1.4, but not this one.
I've run a full grey scale calibration (EOTF target Gamma 2.2) and primary/secondary saturation sweep after clearing previous greyscale and CMS settings and reverting to HCFR 22.214.171.124. I'm seeing similar differences between GDI(0-255) and GDI(16-235) that I saw previously. I'm also seeing large differences in the saturation sweep when sourced with external patterns (GCD) vs. HCFR generated patterns. I'll describe those in a bit, but first I'd like to post my PC/GPU settings. I'd appreciate any advice if I'm doing something wrong.
My laptop has two graphics controllers: integrated Intel HD Graphics 4000 and NVIDIA NVS 5200M. I'll focus on the NVIDIA settings as that is the controller which I have selected for my HDMI connection. I'll list what I think are the pertinent settings from the NVIDIA control panel, what options are available, and which I have selected (indicated by a * and in bold).
2. Apply the following settings.
Desktop color depth: Highest(32-bit)*
, Medium(16-bit), Lowest(8-bit)
Output color depth: 8 bpc, 12 bpc*
Output color format: RGB*
Output dynamic range: Limited, Full*
Video->Adjust video color settings
2. How dow you make color adjustments? With the video player settings, With the NVIDIA settings*
Advanced->Dynamic range: Full(0-255)*
I have a few questions about how to set these options for greyscale and CMS calibration with HCFR:
1) Should I use YCbCr44 rather than RGB output?
2) Should I use 8 bits rather than 12 bits color depth?
3) Should I use the video player color settings rather than the NVIDIA settings?
Now for the greyscale calibration results. I ran an 11-point greyscale calibration using madTPG as the renderer and HCFR as the pattern generator. After calibration I measured the greyscale with the following rendering options: madTPG, GDI(0-255) and GDI(16-235. I also measured with patterns sourced from the GCD via FireTV/Kodi. Here's a summary of the results:
HCFR/madTPG: flat RGB levels across all luminance levels except for a blue deficiency at 15% which my 11-point controls don't allow me to correct. DeltaE < 0.5 across the board except for 15%.
HCFR/GDI(0-255): mostly flat RGB response with some channel separation at low IREs. DeltaE < 0.5 except for spikes at 15%, 25% and 55%.
HCFR/GDI(16-235): exponential decay RGB curve with >500% Y error at 5% luminance. DeltaE linearly drops from 6.0 to 0.3.
FireTv/Kodi/GCD: mostly flat RGB response, very similar to the GDI(0-255) measurements at 10% and above. Slightly increased levels at 5% luminance. Green channel is slightly depressed relative to Red and Green. DeltaE generally good, below 2.0 except at 95%.
I have a few questions/thoughts about these results.
1) Based upon the external source (GCD) results I'd infer that the madTPG and GDI(0-255) measurements are valid and the GDI(16-235) measurements are invalid. Is that a reasonable conclusion?
2) Why are the results different for the low IRE (<30%) between GDI(0-255) and madTPG?
As they both have the same source generator (HCFR) I'd expect the same results.
3) Which is a better renderer for grey scale calibration, madTPG or GDI?
Finally to the saturation sweeps. I measured a 5-point saturation sweep (0%, 25%, 50%, 75%, 100%) for all three primaries and three secondaries with two different sources. I made no changes to the CMS settings in my TV before the measurements. Here is what i see in the Rec709 CIE diagrams:
1) FireTv/Kodi/GCD source:
a) The red and green primaries are slightly larger than the Rec709 standard, likely due to the expanded color space of my qunatum dot display.
b) The green and magenta curves are slightly rotated off of the Rec709 targets.
c) All 5 measurements are reasonably spaced for all primaries and secondaries and generally agree with the REC709 targets.
2) HCFR/madTPG source:
a) The 0% and 100% measurements align with the Rec709 targets well with red and green closer to the Rec709 levels than the GCD sourced patterns.
b) The middle three measurements (25%, 50%, 75%) are compressed toward the top with the 25% measurements at the 50% Rec709 targets, the 50% measurements above the 75% targets and the 75% measurements very near to the 100% target.
1) Does the madTPG saturation sweep point to a problem with the source (highly unlikely as I assume that HCFR has been thoroughly vetted), or a problem with my set up? If my setup, what could be the issue?
From the manual source (GCD) I conclude that my TV is capable of generating a Rec709 compliant color response but I'm confused as to why I'm not able to use HCFR generated patterns to measure this.
I've attached a number of files which show my results:
After Grey Scale calibration - madTPG.jpg : measured RGB and deltaE curves using madTPG renderer, after grey scale calibration using madTPG renderer
After Grey Scale calibration - GDI (0-255).jpg : measured RGB and deltaE curves using GDI renderer with 0-255 range, after grey scale calibration using madTPG renderer
After Grey Scale calibration - GDI (16-235).jpg : measured RGB and deltaE curves using GDI renderer with 126-235 range, after grey scale calibration using madTPG renderer
After Grey Scale calibration - GCD.jpg : measured RGB and deltaE curves using GCD patterns, after grey scale calibration using madTPG renderer
Color measures after Grey Scale calibration - GCD.jpg : measured primary/secondary saturation sweep using madTPG renderer
Color measures after Grey Scale calibration - madTPG.jpg : measured primary/secondary saturation sweep using GCD patterns
grey scale.zip contains:
After Grey Scale calibration - GCD.chc
After Grey Scale calibration - GDI (0-255).chc
After Grey Scale calibration - GDI (16-235).chc
After Grey Scale calibration - madTPG.chc
color measures.zip contains:
Color measures after Grey Scale calibration - GCD.chc
Color measures after Grey Scale calibration - madTPG.chc