Part I: Project settings
Part II: BT.2020 or P3-D65 Limited?
Part V: RAW
Part VI: The Grade
Part VIII: Why HDR Production Monitors Matter
“If you want to use an OLED for HDR mastering, I highly recommend using P3-D65 as your working color space. That’s the best choice in such pipelines. Don’t use BT.2020 with an OLED, only BT.2020 limited to P3-D65. I see this mistake all the time.”– Tim Yemmax, Colorist
YouTube Help Page
Colorists are virtually unanimous in recommending BT.2020 (P3-D65 limited) for grading HDR content and for good reason: DCI-P3 happens to be the color gamut most widely used by the film industry; Netflix requires deliverables be P3-D65 ST.2084; P3 is currently the largest color gamut that can be reproduced by mastering reference monitors; and even ubiquitous mobile devices support as much as 99% of the P3 gamut. In fact, it would be no exaggeration to point out that billions of people are carrying in their pockets a display with greater color accuracy and higher brightness than those found on the sets of most multi-million dollar productions. It goes without saying, however, that few colorists regularly upload anything but tutorials to YouTube; that precious few have ever uploaded a single HDR video in their life; and that relatively few colorists have any experience at all grading HDR professionally; furthermore, it’s no secret that not a few are even openly hostile toward high dynamic range video. We only bring this up because it may help to explain colorists’ stunned disbelief upon learning that on Google’s own YouTube help page, using the P3 color space is strongly discouraged. Curiously, neither on his English language YouTube channel nor on his German language one does Tim Yemmax recommend P3-D65 in his tutorials – even though video monitoring settings, color processing modes and a calibrated display are all mutually interdependent and need to match for accurate results.
No HDR Monitor? No Problem!
Initially, we thought it might be enlightening to conduct an online experiment to see whether anyone apart from a seasoned professional like Tim Yemmax could also instantly recognize when a YouTube video was carelessly graded in BT.2020 instead of the industry standard P3-D65. Naturally, we’d want to upload a few videos to YouTube to discover whether the color shifts were real. We also asked ourselves how someone so meticulous about proper color management could offer HDR masterclasses that promise to teach how to grade without an HDR monitor. And perhaps most importantly of all, we wondered if switching to P3-D65 would help revitalize our waning YouTube subscriber base…
Any serious discussion of color spaces must surely consider the acquisition format. That is, if the color processing mode is going to be P3-D65, would we be better off recording a smaller color space like S-Gamut3.Cine? Or should we instead be capturing the widest available gamut – S-Gamut3 – thereby preserving as much of the sensor’s native color space as possible and future-proofing our timeless masterpieces of cinematic art for posterity? Whether the sensor in the a7s III can even see the full color range that S-Gamut3 is capable of recording is another story altogether. If instead of XAVC, you’re shooting ProRes RAW, it’s sensor native data and no color space or log curve is applied – only the color space metadata is saved. The color space is applied to the RAW data during the de-Bayer process, so it shouldn’t matter which of the two gamuts we choose for recording – but which one is optimal for HDR grading isn’t entirely clear. Which gamut do DPs shooting Netflix Originals prefer? The Camera Production Guides that spell out the settings and best-practices for capture with Sony cameras on Netflix 4K Originals from the Venice on down, whether XAVC or RAW, recommend either S-Gamut3/S-Log3 or S-Gamut3.Cine/S-Log3, adding that S-Gamut3.Cine/S-Log3 is the most common color space used on FX9, F55, FS7, F5 and FX6 productions. We’ve also been informed that key mastering engineers at Sony Pictures in Culver City recommend selecting the S-Gamut3.Cine/S-Log 3 settings in Resolve.
Art Adams had this to say about choosing gamut:
“While the temptation to use full SGamut3 is probably overwhelming to some, it’s best to ask yourself (1) when will the extra color be displayable, (2) will your project still be marketable when that happens, (3) are you shooting anything that takes advantage of that color space, and (4) do you have the talents of a professional and expert colorist at your disposal. If not, SGamut3.cine is clearly the better choice. If your project has a long shelf life and would look great with rich saturated color then SGamut3 will protect all that, but you won’t be doing the Rec 709 or P3 grades yourself: you’ll need skilled professional help.”
Then there’s this advice from Sony Professional:
“There is a lot of confusion regarding S-Gamut3 and S-Gamut3.Cine. Essentially the difference is S-Gamut3 is the native (very wide) color space of the camera, and is wider than REC-2020. It is very good for archiving as a digital camera negative, as transcoded / debayered up to16 bit code values. However, it is much more involved to grade footage than if shot using S-Gamut3.Cine. Whereas S-Gamut3.Cine is natural color reproduction with minimum grading needed in comparison to SGamut3, it is still a wide color space, beyond DCI-P3, and much wider than REC-709. We recommend shooting S-Gamut3.Cine most of the time. The Look Profiles, and 709 3DLUT’s provided in the camera, and in Sony’s RAW Viewer are designed specifically for S-Gamut3.Cine, not for S-Gamut3. If you choose to shoot S-Gamut3 you need to convert to S-Gamut3.Cine with the 3DLUT provided below, “SLog3SG3toSG3Cine.cube”, then in addition apply Look Profiles or other type 3DLUT provided with the camera and in Sony’s RAW Viewer. In other words you need to apply two 3DLUT’s. If you are experienced at Color Grading, you can indeed shoot S-Gamut3, but as mentioned above it is much more work than if shooting (and thus grading) S-Gamut3.Cine.”
Where content is going to be color graded, it is recommended to use source material that best retains the native color gamut and original dynamic range of the content at the time of its origination in order to create high quality Dolby Vision content.
Where content will be directly edited and then aired with just some minor color corrections or even without a color grading process, capturing in a format that matches to the color gamut and dynamic range of the delivery requirements can enable a simpler workflow.– Dolby Vision
When it comes to grading, the very first requirement is inarguably a mastering monitor large enough to assess UHD image quality at a viewing distance of 1.5 times picture height. A rule of thumb is the smaller the monitor, the better things look – and only when they’re really bad do problems become apparent. The second condition is a monitor capable of being calibrated to P3 D65, PQ / ST.2084. “That’s child’s play,” you’re probably thinking, “just adjust the settings on the LG CX!” That there is a hidden menu on the LG CX with Colorimetry, EOTF, Mastering Peak and Mastering Color, MaxCLL and MaxFALL, P3-D65, Rec.2020 as well as various flavors of SDR goes to show just how wildly popular LG OLEDs have become among both colorists and filmmakers, but information about how exactly to configure the settings is all but non-existent and we found that changing the colorimetry or the mastering color had no discernible affect on the picture whatsoever.
Not only that, but regardless of whether HDR Rec.2020 PQ or Rec.2020 PQ P3-D65 were selected on the color management page of Resolve, our TV obstinately detected BT.2020 as the incoming signal.
What’s really going on when we grade HDR Rec.2020 PQ (P3-D65 limited) on a Mac connected to an LG CX via an UltraStudio 4K Mini HDR? We reached out to Blackmagic support in Singapore and received this reply:
“In this case, you would still be outputting in a Rec.2020 gamut, just that the signal is hard limited in a P3 space within the Rec.2020 gamut. I believe that the UltraStudio 4K Mini would suffice.” – Zach He, DaVinci Resolve Specialist
“What’s challenging about [the LG CX] for HDR though, is that it doesn’t work in any other HDR color space other than Rec.2020. While the display itself covers most of the P3 gamut, HDMI doesn’t technically support using P3 color primaries. So you always have to work in Rec.2020 for HDR… With other color spaces or look LUTs, you’ll need to spend an additional $1,300.00 for the Teranex Mini SDI to HDMI 8K.”– Samuel Bilodeau
While the LG CX display itself is said to cover as much as 96% of the P3 gamut, apparently the HDMI protocol doesn’t technically support using P3 color primaries – which must be what the warning on the YouTube help page is referring to when it says that P3-D65 is not a supported format for delivery to consumer electronics. If that is indeed the case, in order to make the display compliant with other color spaces, you’d need to invest an additional $1,295.00 into something like the Teranex Mini SDI to HDMI 8K ($1,600.00 here in Vietnam). So, we’re looking at a total expenditure of around $4,000.00 including an LG OLED TV and UltraStudio 4K Mini HDR with Thunderbolt 3 – all for a non-monetized YouTube channel with fewer than fifteen hundred subscribers of whom only 7% watch our content on a regular basis!
According to Ted Aspiotis, “The P3D65 selection in the HDMI Override menu is just a cosmetic option; it’s the same as REC.2020. LG hasn’t enabled it yet, and it’s pending to be enabled in the future (or never, as its over two years of waiting already).”
Tyler Pruitt adds, “You can calibrate the TV to HDR 2084 P3 gamut using Calman. A lot of facilities are doing this and usually they do one picture mode, say HDR cinema mode to 2020 HDR, and a second HDR picture mode like HDR Filmmaker mode to P3 HDR. This is a very common practice when using LG TVs in postproduction for HDR.”
Evidently, there are hordes of hapless filmmakers out there who sincerely believe that they’re grading HDR Rec.2020 PQ P3-D65 – when they’re not!
Since an X-Rite ColorChecker was used in all of our tests, and as there is next to no information available anywhere about color charts and HDR video, we thought we’d take this opportunity to share just a few of our observations. The ColorChecker is used to maintain color consistency from the set to the edit. Since the earliest days of film, one of the challenges the cinematographer faced was communicating with the telecine colorist how the picture on the set was supposed to look and an 18% gay card was used to let the film dailies timer know where exposure should be placed and what color the dailies ought to be. Because film colors are fixed and digital colors are not, a gray card no longer tells us anything about the correct color. That’s where color charts come in handy. They’re also useful in workflows that employ LUTs and plugins calibrated for properly exposed, white balanced, color corrected images. Due to the limitations of inks and printing methods, test charts in general are not printed at 100% saturation: some are printed at 50% and to be read properly, the vectorscope is set to 2X magnification. Even for Rec.709, reflective cards are already pushing the boundaries of what is possible with pigments, which is why they are really only useful for HD colorimetry. So, we’re finally parting ways with the bad, old ColorChecker and making corrections to skin tones using the vectorscope and either the color board of Final Cut Pro or Offset in the Primaries – Color Wheels of DaVinci Resolve.
We tested every conceivable combination, from S-Gamut3 and S-Gamut3.Cine to DaVinci Wide Gamut, HDR Rec.2020 PQ and Rec.2020 PQ (P3-D65 limited), full versus video levels as well as creating a bunch of mind-numbingly boring still lives and portraits of yours truly to get to the bottom of the P3-D65 controversy. Constructing elaborate still lives, shooting, editing, grading and rendering dozens of videos is a time-consuming process and our task would have taken an eternity with anything less than the incomparable M1 Max MacBook Pro. For the utmost in quality, we shot ProRes RAW which was then transcoded to ProRes 4444 in Apple Compressor (the process is blisteringly fast with the M1 Max MacBook Pro) and imported into DaVinci Resolve Studio 17. Until Resolve supports ProRes RAW, ProRes 4444, which can be considered visually lossless, will have to suffice.
We discovered that the colors of the much-maligned a7s III are better than we imagined. And even though we didn’t get answers to all of our questions, we confirmed that there’s a whole lot more to grading P3-D65 than just selecting an option on the color management page of DaVinci Resolve. While conceding that there most certainly are differences between P3 and Rec.2020, in the end, it’s your grading skills that will ultimately determine whether your videos attract eyeballs or not and recording XAVC instead of RAW or using a camera with poor dynamic range might very well have a greater negative impact on the final results. And if a video graded in Rec.2020 and one graded in P3 look identical, it very well could be because they are!
Update, 03.06.2022: Ted Aspiotis explains how to set up your LG TV for P3-D65 ST2084.
Here are the project settings HDR Rec.2020 PQ (P3-D65)
You can get even more granular if you like. Select either S-Gamut3.Cine/S-Log3 or S-Gamut3/S-Log3 according to how the source footage was captured.
Note 23.10.2022. We categorically recommend Rec.2020 PQ (P3-D65 Limited).
For the past six months, we’ve been uploading Rec.2020 ST2084 (P3-D65 Limited) videos to YouTube, but it appears that many are still operating under the assumption that Rec.2020 videos constrained to P3 should not be uploaded to the video sharing platform. In order to settle this question once and for all, we recorded a billboard with super bright, highly saturated colors. The signage was so bright that, in order to prevent clipping of the highlights, it was necessary to stop the lens all the way down to f/13 at ISO 800.
In DaVinci Resolve Studio 18, working with RedCode RAW, we added a Gamut Limiter at the very end of the color pipeline to hard clip any colors beyond P3.
When switching the Gamut Limiter in DaVinci Resolve on/off, changes could be easily seen in the vectorscope, indicating the presence of colors that exceed P3, but the picture itself looked the same with the limiter on or off, because, like all displays, the XDR miniLED of the MacBook Pro cannot reproduce colors beyond P3. We also struggled to see any discernible difference at all between the original clip and the one uploaded to YouTube when viewing on either the MacBook Pro or on our LG CX.
We’d also like to offer these words of advice to those creating HDR videos for YouTube. Firstly, in project settings, on the color management page, it is important to set the correct value for “HDR mastering is for X”, which lets you specify the output, in nits, to be inserted as metadata into the HDMI stream being output, so that the display you’re connecting to correctly interprets it. In the case of the LG CX, that figure would be 700. Secondly, as the full screen brightness of OLED displays is < 20% of peak brightness, the average picture level (APL) of your grade should probably not exceed 200 nits. Lastly, we strongly recommend inserting MaxCLL and MaxFALL metada so that your videos look correct on displays with differing brightness and contrast capabilities.