A colourists and display calibrators look at the iPad Pro M4 13” (2024 model)

A few months ago I purchased my first Apple product ever (purchased with my own money, Apple does not sponsor this, so they did not have any influence on this article) - The new iPad Pro 13" (2024) with its brand new dual OLED Display. However, this is not a review of the iPad Pro 13” (2024) and its user experience. There are already tons of excellent reviews out there. I bought it purely because of the display. I was looking for a small, portable display that would allow colour-critical tasks like look development and colour grading wherever I go. The new iPad Pro was pretty much the only option that came up. I used it for several months now and just by comparing it visually to grading displays like the Sony BVM-HX310 the colour accuracy of the latest iPad was impressive. The new Resolve 19 feature to stream the Resolve video output over the local network without going over a BMD server makes it the ideal choice as a wireless dedicated grading display that can be used right next to your laptop (as long as you have a Windows or Linux laptop with NVIDIA GPU because that is the prerequisite for the Resolve Video streaming on the encoding side…. topic for a different article….).

Using the iPad Pro 13" (2024) as video output Monitor via BMD DaVinci Resolve's local network streaming feature.

 After a few months of usage, I finally had the time to take some serious measurements on the iPad Pro 13” (2024) to confirm what I already suspected from the initial impressions. The TLDR is: The new iPad Pro 13" (and probably the 11" 2024 model too, since it is just a smaller version of otherwise the same device) is the best mobile colour-grading display on the market so far. Is it as good as a professional reference display? No, but you must look closely to see the differences. More about that later.

This is made possible by the "Reference Mode" feature which is hidden in the display section of the system settings in iPadOS (but can also be found on the MacBook Pro, so some of the things described in this article apply to the M series MacBook Pro devices as well, although I have not yet checked the colour accuracy of the MacBooks MiniLED Display). Once activated, the display brightness and white balance are fixed and can not be adjusted. In the background, Apple ColorSync assures that colours, gradation and peak brightness values from images and videos in any colour space no matter the application are translated and mapped correctly to the native gamut of the dual OLED display. Of course, as a professional colourist and display calibrator, I have been running my iPad Pro in Reference Mode all the time.

The Reference Mode also allows for calibration of the white point which Apple calls "fine-tuning" in the display settings of iPadOS. But that’s it, in terms of calibration features. It is impossible to change or influence the gamut mapping or the position of the primaries - you have to rely on the manufacturer having done everything right in this regard. Straight out of the box, I measured the iPad Pro’s red, green and blue Rec709 primaries and the white point with my JETI specbos 2501 HiRes spectroradiometer. The measured values were almost spot on to specification. Peak white was just 4nits too low which can be a result of me having used the iPad Pro already for some time (time of measuring September 2024, purchased in May 2024, manufactured in February 2024). But this is already quite good for a consumer device as a "reference" or "filmmaker" mode intended for colour-critical work. The first noticeable difference in direct side-by-side comparison against a calibrated Sony BVM-HX310 colour grading display would be a slightly more greenish-looking white point on a direct-out-of-the-box iPad Pro 13” (2024). This is a result of the different spectral properties of the iPad Pro’s dual OLED Panel compared to the BVM-HX310's dual LCD type panel which leads to differences in visual perception of predominantly white, grey and neutral colours despite both display's white points measuring in with the same CIE 1931 x, y chromaticity values on the spectroradiometer (a phenomenon called metameric failure).

The best practice is to properly calibrate the reference display's white point to the D65 specification and then visually match the white points of other displays with a different technology used in the same room and/or for the same project to get the best visual match. And that is exactly what I did before I started my experiment. I visually obtained a white point match for the iPad Pro (2024) by simply holding it side by side with a perfectly calibrated Sony BVM-HX310 displaying a 100nit peak white D65 solid and adjusting the RGB sliders on the iPad Pro until the white solid on the iPad Pro looked exactly like the white solid on the reference display. After measuring the resulting visually matched iPad Pro white with the spectroradiometer I then used the "fine tune" feature in the iPad Pro system settings to incorporate the small visual white point offset into the calibration so that the fine-tuned reference mode visually perfectly matches the BVM-HX310 reference display.

However, accurate positioning of the primaries, sufficient gamut coverage and an accurate white point are not everything there is to a well-calibrated display. Display manufacturers often use the "gamut coverage" term to promote the "accuracy" of their products. But colour accuracy does not automatically follow from a good gamut coverage. A display that merely "covers" 99% or even 100% of Rec709 or P3 is not necessarily calibrated to any industry standard spec used in image post-production nor does it mean it is colour accurate. It just tells us something about the physical properties of the display and its potential to be calibrated accurately. In other words, these technical specifications in the marketing material only tell us where the outer borders of the display’s colour space are but nothing about how all the other colours inside the colour volume are mapped. For perfect calibration of a display, it needs to be able to map input values to specific colour outputs defined by the different colour space targets e.g. Rec709 Gamma 2.4 - and it needs to do that with all of the colours within the 3-dimensional colour volume - and not just the outer boundaries and the white point. In fact: correct gamut mapping is ultimately much more important than whether or not a display “covers” the target gamut completely. To verify the accuracy of the colour gamut mapping we have to profile and verify the display post calibration by comparing the measurements of the verification profile against the colour space specifications.

Apple does not seem to like it when people can look into their engineering. Display calibration and colour management are 2 of the things Apple has somewhat hidden behind a black curtain. It is not easy to just go ahead and profile an Apple display with the tools I normally use. After figuring out a workflow I managed to run my display calibration verification colour patch set via DaVinci Resolve on the latest iPad Pro 13” (2024). Since it is the same patch set I normally use to verify the colour accuracy of all the displays I calibrated I can now objectively compare the new iPad Pro to professional Colour Grading Displays like the Sony BVM-HX310. For now, I did the verification in Rec709 Gamma 2.4 SDR. Therefore the timeline colorspace in the Resolve Project Settings was set to Rec709 Gamma 2.4. Apple ColorSync will take that parameter from the Resolve Project Settings and then map the Resolve timeline colours accordingly to the iPad Pro’s dual OLED display for it to behave as a Rec709 Gamma 2.4 SDR display - at least, until now, we could only trust it would do that. This is where it gets interesting. So far I have only heard subjective eyewitness testimony about the display accuracy of miscellaneous Apple devices and now my own subjective visual experience with the display of the iPad Pro (2024). I have not yet seen anyone bother with assessing the display accuracy of an iPad Pro objectively nor have I seen sophisticated calibration reports from any Apple display.

Results

To illustrate some of the findings from this experiment I used the excellent profile reporting features of Light Illusion ColourSpace - The software I use for all display-calibration and colour-science-engineering-related tasks.

The above image is a small summary of the results from the verification profile taken from a calibrated Sony BVM-HX310. These results represent what you can expect from a calibrated HX-310 in Rec709 SDR. The HX-310 will be the reference in this little case study (since it is the reference in the professional colour grading world anyway). The colour target, as mentioned earlier, was Rec709 with a pure 2.4 power law EOTF (aka Gamma 2.4). The target white point is D65 at a peak luminance of 100 nits. As can be seen from the data 1087 colour patches across the whole colour volume were measured in this verification profile. All 1087 measurements landed easily within 0-1 DeltaE 2000. As a rule of thumb, a DeltaE 2000 of 1 would be the visual threshold where a human could either just notice or just not yet notice a difference between the measured colour patch and a perfect reference colour patch in direct comparison. This rule does not hold up perfectly for all the colours in the volume since our visual system is more sensitive to certain colours and less to others but it helps to compare displays objectively. We also see that the average DeltaE 2000 of all 1087 colour patches is less than 0.2 which is extremely colour-accurate. Not one single colour patch was above DeltaE 2000 of 1. The patch with the highest deviation had a DeltaE 2000 of just 0.7. An almost perfect calibration result - but of course, this display has its price and we expect best-in-class performance from such a device.

In the second image, we see the same summary from the verification profile run on the iPad Pro 13" (2024) after I visually matched the white point of the iPad Pro to the white point of the HX-310 and incorporated this small adjustment into the iPad Pro via the white point fine-tuning feature. Since it was not possible to do anything else, the gamut mapping of the iPad Pro is still untouched. So we are seeing here how good the Apple factory profiling/calibration is, regarding gamut mapping. With the same verification colour patch set we have only 38 colours that came in with a DeltaE 2000 higher than 1. The point with the highest deviation still only had a DeltaE 2000 of 1.5. The average across all 1087 colour patches was 0.46 - which is excellent for a consumer device - in fact, it is in the same ballpark as a perfectly calibrated Sony PVM-A250 OLED.

In the graphs above we see the accuracy of the gradation and white balance of the HX-310 which is absolutely stellar. The EOTF is tracked almost perfectly and the the D65 white is maintained throughout the entire grey ramp.

The same graphs taken from the verification profile data of the iPad Pro 13" (2024) show in some places slight deviations from the optimum but still very good results and well within the accepted tolerances for professional colour grading displays. Apple ColorSync does exactly what it is supposed to do (at least in Rec709 Gamma 2.4 SDR, tests for other colour targets may follow in the future).

In the graphs above we are looking at zoomed-in parts of the 2D projection of the CIE 1931 diagram, focusing only on the white point and R, G, B primary ramps from the verification profile of the HX-310.  Overall near perfect accuracy. But some points don't seem to hit the primaries exactly but are still marked as green. To see what is going on there we need to look at a 3 dimensional illustration (see below)

Here we are looking at the 3 dimensional graph of the verification profile of the HX-310. The darkest colours are at the bottom and the brightest colours are at the top. Here it becomes clear that the darkest primaries are slightly tapered inwards into the colour volume which is normal for the dual LCD technology. As soon as the signal level for a primary reaches about 2-3% the colour measures in almost exactly at the colour space boundary - which is what we want to see. The primaries remain stable and in spec for the vast majority of the signal range. The reason these very dark patches are still marked as green (DeltaE 2000 < 1) is that in the DeltaE 2000 specification, deviations in such dark colours are regarded as not visually noticeable and in the case of the HX-310 the point with the highest deviation only had a DeltaE 2000 of 0.7.

Another view of the 2D projection of the CIE 1931 graph from the verification profile of the HX-310. This time with all the 1087 measured colour patches. We see a lot of "clumps" which are all individual dark to bright ramps of the same CIE 1931 x,y chromaticities - ideally, the points of each clump would be all perfectly on top of each other in this projection. But no display (neither probe) is perfect - hence the visible clumps.

Then, for comparison the zoomed-in sections of the CIE 1931 graphs from the verification profile of the iPad Pro 13" (2024), focusing on the white point and primary ramps. Slightly more deviations are visible compared to the HX-310 (please note that the zoom-in is very strong for illustration purposes). However, it becomes clear that some patches of the primary ramps are outside of the colour space boundary.

In the CIE 1931 3D graph, we can see that in the red and green primary ramps, the darkest patches seem to be skewed slightly outwards compared to what we could observe on the HX-310. So essentially darker primaries on the iPad Pro 13" (2024) would look a tiny bit more saturated than they should. I have to stress again that we are talking about tiny differences since most of the measurement points we are talking about are still within the DeltaE 2000 0-1 range. But since most of the colours with a higher than average DeltaE 2000 deviation seem to be concentrated towards the lower signal range and these colours are all skewed towards the boundaries of the target Colour Space, observers may notice very small differences when comparing images displayed side-by-side on the iPad Pro and the HX-310. But again: these differences are well within the tolerances for a professional colour-grading display.

Another view on the 2D projection of the CIE 1931 graph with all the verification data points taken from the iPad Pro 13" (2024). We can see that the "clumps" are a little bit larger than on the HX-310. Also, especially in the red corner, the clumps are elongated towards the red primary. The darker colours of the individual colour ramps are skewed towards more saturated tones while still mostly within the DeltaE 2000 <1 range and only very few points above DeltaE 2000 of 1. The core of the colour volume (area around the white point) is not affected by this as much as the more saturated colours are. So if you ever wondered why certain darker colours containing a fair amount of red look slightly more saturated on the iPad Pro (2024) compared to the Sony BVM-HX310 reference display - now you know.

Verdict

The accuracy of the factory calibration of the iPad Pro 13” (2024) is amazing. Although it does not reach the level of accuracy we see in a Sony BVM-HX310 (unsurprisingly, given the difference in price and target audience) it is extremely good for its price. And on top of that the iPad Pro is not just a mobile display - it does a lot of other stuff too (I have been told). Thanks to the new local network streaming feature in DaVinci Resolve 19 it is ideal as a mobile colour grading display. I can easily invite a DoP for a look development session (remote or local). Provided a decently quick and stable internet connection and the other viewers have the same iPad Pro model I have the certainty that everybody looks at the same images.

Could the display accuracy of the iPad Pro (2024) be even better? Yes, the dual OLED display technology has the potential to be calibrated even more accurately than it already is. It is purely limited by the lack of more sophisticated calibration features in iPadOS. However, the feature to calibrate the white point is helpful to better match the iPad Pro (2024) to a reference display.