HDR Grading and output

[Luke Neumann]

I have been trying to put together a workflow for HDR grading and I think I have most of it figured out but was wondering if there was anyone that could clarify or share some tips on the matter. As of right now, this is my current workflow:


Export 16 bit DPX from RedCineX using the REC 2020 Color Space and HDR-2084 Gamma Space. Change Premiere Pro/Lumetri scopes to HDR and do a light grade in Premiere. Export as DNxHR 444 10 bit.


Things I'm not too clear on:

1. Is the DNxHR codec the best choice? I realize it will maintain a lot more information than most codecs but is it maintaining the gamma space?
2. Do you just grade to whatever HDR display you're using? I know Dolby has displays of 10,000 nits. What is the current standard? Do you aim for the median to try and satisfy the bulk of the models out there? If you grade PAST what some displays are, won't you have some harsh clipping? Safer to export a few versions that work with a few different values?

Again, any help is much appreciated.

 

[Marc Wielage]

My personal opinion is that HDR is premature and the standards are not completely nailed down, particularly between Rec2084 and Rec2020. I'm also skeptical that consumer HDR monitors are going to be reliable. Bear in mind they still can't provide Rec709 right out of the box, and that standard was created almost 20 years ago.

But you can read these articles:

https://www.linkedin.com/pulse/hdr-survival-guide-next-generation-home-dado-valentic

https://pro.sony.com/bbsccms/assets/files/cat/mondisp/articles/HDR_X300_explained.pdf

http://www.spectracal.com/Documents/Understanding-HDR.pdf

https://www.smpte.org/sites/default/files/Study Group On High-Dynamic-Range-HDR-Ecosystem.pdf

I think unless and until you have a Dolby Pulsar monitor, a Canon DP-V3010, or a Sony BVM-X300 monitor (each well over $30,000), it's just an intellectual exercise at this point. Note that none of these displays can still hit 100% of P3 color space. Now, imagine how well -- or how badly -- a $2000 consumer set will be able to handle the same thing. Barco and Christie laser projectors can handle it right now, but I suspect many will balk at the $250,000 cost of entry for those units.

Because there are multiple brightness levels for HDR, I'm not convinced it'll be possible to do a "one size fits all" correction that will fit for 100 nits or 1000 nits or 4000 nits. My fear of this moving target for HDR/wide-gamut standards is that it's possible a transfer today will not necessarily be optimal for the displays of the year 2020 or 2025 or beyond. I don't think HDR is important yet unless a particularly distributor (like Disney and Fox) demands both HDR and standard deliverables for a particular project. And they'll have to factor in the additional post costs involved.

My opinion is that it will always be possible to take log images and correct them to match virtually any dynamic range in the years to come. If the distributors want to pay for an HDR version in 2020, then I would plan to do another trim pass then and spend a couple of days creating that version. You will not have to go entirely back to scratch, assuming the color corrections survive and can be converted to forthcoming software a few years from now.