I have a question, for which I haven’t found an exact answer (or I missed it… sorry…).
Why when I work in ACES: P3-DCI (simD60) are my highlights clipped before 100% of my scope?
Which doesn’t occur in P3-D65 or Rec709… I also notice that the dynamic ranges between the color spaces, are not exactly the same.
I think it’s because interpretation of white point, but also of spaces that are a little different… maybe ?
Anyone have an idea, and any workarounds? or is everything OK ?
Yes, that is expected behavior. I’ll probably explain this poorly but I’ll start the conversation…
The “sim” variants are to match creative white differences from your currently set white point of your display. When introduced, the ACES white (~D60) was a common setting for projectors where mastering was being done. So with the projector set to RGB P3D60, a code value triplet of [1 1 1] would result in a chromaticity of .0.32168 0.33767]. If then, for example, someone brought a Rec709 D65 monitor into the same mastering suite, and compared the images side by side, the whites would not match. This is the expected result because the Rec709 transform assumes an observer adapted white of D65 and performs a chromatic adaptation to D65, whereas the projector is set to assume D60 as what the observer is adapted to as “neutral”.
In order to produce D60 on a device that is set to a D65 white point (i.e. a display set up such that equal display code values yield CIE x,y chromaticities of [0.3127, 0.329]) then the red channel needs to be higher than green and blue to compensate for the bluer" D65 white. This is the intended behavior but it means that without compensation, as highlights increase, the red channel will hit the device maximum first and clip, resulting in a chromaticity shift as the green and blue channels continue to increase.
To avoid this clipping behavior, a slight scale factor is applied to darken the image slightly and allow the ODTs to “fit” D60 within the D65 calibration white point. Therefore, some channels will appear to be “unequal” on a signal waveform, but this is by design. If you want the waveform to show equal code values at [1 1 1], use a transform that is designed for that display’s white. Otherwise, expect unequal channels in order to be able make the D60 chromaticity.
Here’s 3 images to show the difference (using v2 transforms - but the effect is the same). Note that P3D65 and P3D60 are the same because its the signal being plotted, not the chromaticities. The setting of the display matched with the encoded white results in [1 1 1], otherwise, you get unequal channels.
As an add on, I would not usually recommend using the “sim” versions unless you know what you’re doing. They’re pretty much only there for very specific use cases and in general its a better idea to use an output transform that adapts white to the assumed observe adapted white for your display and viewing conditions.
Hi, Scott. If you don’t recommend using the “sim” versions, which ODT would you use on a P3 proyector calibrated for DCI white point? In Resolve there is just "P3 DCI (D65 sim) and P3 DCI (D60 sim) Running some tests the ODT that matchs other applications “P3 DCI” WP (including Resolve at timeline level) was "P3 DCI (D65 sim) I know this could sound quite vague, sorry. Many thanks
Well, you could certainly use it for a P3DCI projector, but the same logic applies for the mismatch as between D60/D65, just to an even larger extent with DCI. You should not be expecting channels to be equal code value at white or else that would mean they will come out at the DCI white chromaticity [0.314 0.351].
If you need neutral to be D60 or D65 and represented with equal RGB signal on your waveform then you should be setting your projector itself to D60 or D65 white (not DCI) and using the corresponding ODT.
Hello Scott and thank you for all the very thorough explanations and for taking the time!
I used the P3DCI_simD60, the director of photography wants to work in P3 DCI. My projector is indeed configured in P3DCI, which works well when I’ml working also P3DCI on DaVinci (mode RCM)
If I understood correctly, the whites are clipped earlier due to the RGB compensation and to the slight scale factor is applied to darken the image slightly and allow the ODTs to “fit” D60 within the DCI calibration white point(to sum up…). Personally, in ACES, I would tend to go towards P3D65, especially if later we want to do HDR (Dolby Vision) for example. But the director of photography wants to stick to P3DCI for cinema calibration. For me it’s a starting point. After when you want a master or a DCP it’s just a question of good colormanagement with input and outpout.
On your graph number 02, it says P3D65_simD60, so the “same thing” in P3DCI_simD60 ?
From what I’ve read in other conversations, there are no other options possible in ACES to have P3DCI. Either we go for simD60 or simD65, with a feeling of a colder or warmer image depending on what we want. I compared simD60 and simD65, and I prefer the image in simD60
It is a misconception that you need to use the green DCI white point in order to be compatible with a DCP.
It only produces unnecessary light loss and complications.
I explain this in detail here:
Best thing to do is to setup your projector to the primaries and white point you want to actually use. (E.g P3 D60)
It is also a misconception that you need to use D65 in cinema in order to get a better match for television.
In cinema you adapt to the white point to a greater extend because you have no surround.
