Hello everyone,
I've recently joined the QD-OLED world with several of the latest SmallHD Quantum monitors, and I'm struggling to get consistent calibration results. I'm working on two 32" and two 27" units, but despite following the SmallHD calibration guide (the guide doesn't mention that the Quantum calibration wizard actually requires a two-step characterisation) the results have been less than ideal.
The core issue is that the white points never perceptually match between units, even though quick profiles show low delta-E values and everything appears correct in ColourSpace. This raises a question: even if the instrumented results look good, do I still need to perceptually match the white points? And if so, why would that be necessary when these monitors use the same panel? Are the narrow spectral peaks of the primaries causing increased unit-to-unit metameric differences? I assumed the metameric behaviour would be similar across panels with comparable narrow-band emissions.
Before I commit to full volumetric calibrations—which will take quite some time across multiple units—I want to make sure I'm not overlooking anything fundamental. My workflow is as follows:
· Warm up each monitor for 45–60 minutes.
· Set patch size to custom 3%, full range, no stabilisation.
· Run the 1000-patch pre-roll CSV (8-bit full range), at 1 second per patch (adding ~16 minutes of additional warmup).
· Switch to 10-bit range and set stabilisation to 0.250s FW. (I'm not sure stabilisation is even needed on this panel type—thoughts?)
· Set CR-100 exposure to 500 ms max integration time, multiplier 2, Intelligent Integration enabled below 0.100 nits.
· Use probe sync to confirm a manual 48 Hz setting is valid for 24 fps patches, and run auto-delay. Although auto-delay typically gives 0.109 s, I extend it to 0.250 s for safety.
· Create FCVM with the CR-250 as the reference probe (auto-exposure, low light averaging), matching the CR-100. Both probes warm up during pre-roll.
· Use an AJA ColorBox as the pattern generator, bypassing everything. When sending unity bypass to the LUT box, ColourSpace throws an error at 10%. All eight of my ColorBoxes show the same behaviour, even after factory resets, so I suspect a ColourSpace integration bug. (I'll create a separate post on this once I confirm others haven't seen this issue.)
In any case, this doesn't seem to affect measurements: using Resolve as the TPG yields the same suboptimal results, and the monitor's colour-picker returns the correct 10-bit values with both TPGs, so the signal path seems fine.
· Run the characterisation using Grey Primary & Secondary Ramp + for the first step of the Quantum calibration wizard ("bright pass"). Save the profile and generate a LUT for DCI-P3, D65, gamma 2.4. (SmallHD recommends building a custom 2.4 gamma space rather than 2.6.) Peak Chroma, no gamut mapping, normal levels.
· Load the generated LUT and verify with the same quick profile—looks good.
· Continue with the wizard, which displays a white patch and asks for a manual luminance reading to set peak brightness. This is usually just under 1000 nits. I enter the value and proceed to the second characterisation step, targeting ~250 nits.
· Repeat probe matching and quick profile at the lower peak luminance, build the LUT, load, and verify again. Then read the dimmer white patch, enter its value, and complete the wizard.
My understanding is that the monitor derives its other colour spaces and brightness levels from these two P3 calibrations. When I test different colour pipes (Rec.709, P3, etc.) and run verification profiles—including probe matching—the results all come back excellent. Instrumentally, everything is "green."
However, when placing multiple units side-by-side and feeding them content or colour bars, there is a clearly visible white-point mismatch. I didn't have time to check a wide range of colours perceptually, but the white point difference is immediately noticeable.
Has anyone experienced similar behaviour or has insight into what might be causing this? Any ideas or suggestions would be hugely appreciated (especially from Steve)
Many thanks in advance!

Have lined up a few of these and find they do match, in fact the panels were almost identical. However, I did them by hand not with a LUT, and needs some extra steps if mixing SDR and HDR!

With a LUT this video will explain how to get it right, but an odd process that I've used before and works https://youtu.be/Ux7WYaFX-Kc?si=VkJ327Jy09lX0iz8

Give it a try and share how you get on.

In simple terms the SmallHD workflow is flawed.
Making any changes AFTER a LUT upload invalidates the LUT.
No idea if that is the issue, but it is a flawed workflow.

Steve

Update on my previous note on the manual calibration process, back at my desk so I can look up the notes.

First, and normally a good call, upgrade to the latest Firmware I think this functionality was added!?

The trick is to enable 'Manual Calibration Sets' then set up a set for each colour space you are intending to calibrate, at least one for SDR and another for HDR. Then in the colour pipe, select the 'Manual Calibration Set' for that colour space. This allows you to make the usual luminance, bias and gain adjustments for each colour space as you would on a regular production monitor.

Let me know if you get stuck, happy to help.

Cheers

Victor

vlut
Thanks, Victor — I'll review the video you shared.
I had already come across the Calman link referenced in the SmallHD documentation, but not this ColourSpace-specific one. It's unclear why these videos are unlisted on SmallHD's YouTube channel (I'm curious how you even located it), which effectively prevents users from discovering them through normal search.
In any case, it's useful to know that you achieved a satisfactory calibration either via the "manual calibration" approach (which is essentially just post-LUT manual trimming) or previously through the LUT-based workflow.
Steve
Aside from the manual trims Victor mentioned — which, in principle, shouldn't be necessary if the LUT-based calibration is executed correctly — the wizard itself exposes no additional adjustment parameters.
The SmallHD workflow simply requests a measurement of a 3% white patch at the panel's maximum output (or at roughly 250 nits on the second pass). My understanding is that this value is used to derive the luminance scaling for the monitor's various brightness modes (e.g., fixed-nit presets such as 100 nits for studio environments), with the system performing the internal remapping.
I also suspect that SmallHD generates the alternative colour-space and EOTF profiles from this single calibration state, provided the display's native gamut and peak luminance exceed the requirements of the selected targets. This seems to be the intended design, though that's my interpretation.

As said, any alteration of any setting after a LUT has been loaded will break and invalidate the LUT.
It is that simple.
https://lightillusion.com/calibration_issues.html#calibration_adjustments

Steve

Why don't you use the internal LUT of the SmallHD?

Export a reset 3D LUT and upload it using the SmallHD menu, activate it, measure White, and set it to about 120 by measuring White, and then profile using primary only, generate the 3D LUT, and upload it into the SmallHD (BMD 33p LUT file format).

Measure the White and adjust the backlight until you have the nits you want (100, for example).

Re-upload a reset 3D LUT and profile with a 17-point cube.

Resolve iTPG does not support 10-bit TPG; set 8-bit FULL in ColourSpace and 10-bit output in Resolve's project settings.

I don't know if AJA as a hardware TPG supports 10-bit.

Use 8-bit, you will not have any issues.

After calibrating the first (master) monitor, display the White patch on both monitors, match the Y for the second monitor, and then use the second monitor's RGB balance menu to perceptually match the calibrated White of the master monitor.

When they look identical, measure the White on the second monitor, look at the xy coordinates, modify the REC.709 colorspace, and enter those xy coordinates, so this will be the target colorspace for the second monitor calibration.