Blackmagic Forum

Hi,

I got a lot of Rec 709 clips to cut a feature doc from. I realize the footage does not contain HDR information, but would it make sense to set up the project in Rec 2020 / Dolby Vision anyway, to (1) perhaps get a wider color space and contrast in the grading process, and (2) be compatible with future requirements for deliverables?

Thanks so much!

IMHO not.
It's just like making UHD from HD, you won't gain information that wasn't there in the first place.

Thanks, Uli.

I wonder if during the grading process for 2020, aren't you pushing the blacks and the whites beyond what 709 could display, in effect expanding the dynamic range in post - even though it was not recorded on camera?

Similarly, wouldn't you be pushing, during grading, the color values into areas that weren't available in 709, i.e. the original footage?

I understand that up-rezzing does not increase information; but couldn't grading for 2020 increase the contrast, and color gamut, of originally 709 footage?

RobertN wrote:Thanks, Uli.

I wonder if during the grading process for 2020, aren't you pushing the blacks and the whites beyond what 709 could display, in effect expanding the dynamic range in post - even though it was not recorded on camera?

Similarly, wouldn't you be pushing, during grading, the color values into areas that weren't available in 709, i.e. the original footage?

I understand that up-rezzing does not increase information; but couldn't grading for 2020 increase the contrast, and color gamut, of originally 709 footage?

No, it doesn’t work that way. You will get larger files that doesn’t have any additional and useful information.

Plus, if you start with 8 bit and shove values around, you'll risk banding.

I think the idea makes sense. So does Dolby.

viewtopic.php?f=14&t=143886

Jim Simon wrote:I think the idea makes sense. So does Dolby.

viewtopic.php?f=14&t=143886

I don't think it makes sense at all, you can't create data where there is none by mapping it beyond its original color space, and I don't believe Dolby thinks so either.

Dolby is going the opposite direction, HDR to SDR. Their concern is whether there's subjective difference between Dolby's HDR to SDR process and a manual grade of an HDR master to an SDR standard. That's not at all what the OP is proposing here.

Thanks, you all.

Specifically: If I have a green value on my 709 footage, and push it in a 2020 grading session beyond a shade of green that 709 can display - have I not just "created data where there is none"?

I may be prey to a very deep misunderstanding. I would love to get a grip on this.

Thank you!

John Paines wrote:I don't think it makes sense at all, you can't create data where there is none by mapping it beyond its original color space, and I don't believe Dolby thinks so either.

Actually, you can do an SDR -> HDR conversion, but I would agree that (in general) you can't get a wider color gamut than what's already there. But you can tone-map & trim the highlights to get a brighter image for HDR. A much brighter image, as a matter of fact.

Kevin Shaw has a good tutorial on the process, either going from HDR -> SDR or from going SDR -> HDR. I know of documentaries and other shows that "repurposed" material to get HDR out of it, and when it's done well, it can look very good.

No question, going from HDR to SDR, particularly Dolby Vision to SDR, is a better way to go. Dolby Labs did a good presentation on this technique a few months ago, and they make a good case that it's possible to derive a better SDR out of HDR than you could in just grading an SDR deliverable directly from the same material.

I know, "brighter".... This has about as much appeal as colorization, but I guess we'll see lots of it, with the push to sell new TV sets, and reissue movies.

John Paines wrote:I know, "brighter".... This has about as much appeal as colorization, but I guess we'll see lots of it, with the push to sell new TV sets, and reissue movies.

Eh, I've done it and it can work. Kevin Shaw's tutorial is very good and explains the situation very well:

https://www.fxphd.com/product/introduction-to-hdr/

Marc Wielage wrote:Eh, I've done it and it can work. Kevin Shaw's tutorial is very good and explains the situation very well:

Thanks, Mark, for sharing your experiece. I'll sign up for the Shaw course once I'm through with all the Dolby videos.

May I ask whether you graded a whole SDR project in HDR, or just adjusted the SDR master? Did you work with a full HDR monitor, or did you map it down?

RobertN wrote:Thanks, Mark, for sharing your experiece. I'll sign up for the Shaw course once I'm through with all the Dolby videos. May I ask whether you graded a whole SDR project in HDR, or just adjusted the SDR master? Did you work with a full HDR monitor, or did you map it down?

One was a 4K SDR project I had graded, and another one was a flattened 4K file graded in SDR. We used a Dolby Vision-certified room in West LA with a Sony BVM-X300 display, and I used a combination of the automatic Dolby Trim mode and manual trims to create a reasonable tone-mapped version that actually looked very good. I let the specular highlights drift up close to 1000 nits, but kept the bulk of the program levels in some reasonable middle ground, like 200-250 nits. And I made sure that shadows were still deep and rich and so on. We did not have HDR-compatible scopes, so I just used the HDR scopes within Resolve. Everything went fine and we passed QC with flying colors.

So where does one place white/90% reflectance going from SDR to HDR at 1000 nits?

Good Luck

RobertN wrote:Everything went fine and we passed QC with flying colors.

Thanks, Mark, for the details on this. I was afraid there's no way around a true HDR monitor. But renting a suite just to do the HDR pass sounds like it makes (economic) sense.

"Diffuse white" is a fun question. Some say generally around 203, some say 300 or so, some say it's an unanswerable question of too many variables.

Sent from my SM-G960U using Tapatalk

rNeil H wrote:"Diffuse white" is a fun question. Some say generally around 203, some say 300 or so, some say it's an unanswerable question of too many variables.

Sent from my SM-G960U using Tapatalk

ITU-R BT.2408-0 recommends 203 cd/m^2 as the level of diffuse white for both PQ and HLG. But in the end, it's up to you.

Howard Roll wrote:So where does one place white/90% reflectance going from SDR to HDR at 1000 nits?

Eyeballs and gut reaction. I think it sort of depended based on scene content, but I looked more at the Dolby Vision picture than the scopes per se. I used the scopes for matching more than anything else. I wound up chasing the shadow level more than anything else -- it went out of whack a few times. This was before Dolby Vision 4.0, and I think a lot changed and improved since then.

Lots of free info at this link:

https://professionalsupport.dolby.com/s ... et-ddfa6=2

RobertN wrote:Thanks, Mark, for the details on this. I was afraid there's no way around a true HDR monitor. But renting a suite just to do the HDR pass sounds like it makes (economic) sense.

Yeah, the cost of the Dolby Vision/HDR display is a big deal, and it does give one pause. I'll say this: it's more economical now to buy the Dolby Vision tools and get licensed than ever before, so that part of it is very affordable now. There's at least one new Dolby Vision display coming out in the fall (the FSI XM312), and that's about $20K, which is more than $10,000 cheaper than they used to be. We're seriously leaning in that direction for our boutique facility, and we'll see where that goes in a few months.

You can gain some "eye pleasing effect" when creating HDR out of SDR master (brighter image stimulates eyes "more"), but overall process has big limitations (same as any eg. upscale). It's also import how your SDR master is stored. The higher quality format (DPX, EXR etc.) the better. If you have just 8bit SDR master then this can be a problem when going HDR.

Andrew Kolakowski wrote:If you have just 8bit SDR master then this can be a problem when going HDR.

May I ask what kinds of problems you anticipate with 8bit original footage (I am grading a project with old footage, pretty much all SDR/Rec 709)? Like, banding in gradients, and un-clean hard/contrasty edges, this sort of thing? Thanks!

Yes, you will "stretch" your data a lot when going to HDR, so 8bit will show its weaknesses definitely.
It reminds me all those early Blu-ray discs made from bad/old film scans and people screaming HD vs SD makes no real difference. Somehow no one comments it this way now, when most BDs are done from proper HD+ masters. Not sure if whole idea is worth the hassle. Create 1, watch it and compare to SDR if it's somehow more interesting to the eye.

Andrew Kolakowski wrote:It reminds me all those early Blu-ray discs made from bad/old film scans

Good point! Thanks!

moved to Resolve forum