Topics

HDR and PQ 200

Art Adams
 

Hi all-

I'm seeing more and more references to PQ 200 curves. In HDR, until fairly recently, diffuse white was generally targeted at 100 nits, much the same as in SDR (if we all shot with cameras made in the 1990s). Now, though, diffuse white has apparently been moved up the scale to 200 nits, because modern consumer TVs have been brighter than 100 nits for quite a long time and there were a lot of complaints that 100 nits was too dark.

My question: doesn't that shave off a stop of dynamic range? If I had three more stops, plus a nudge, to get from diffuse white to 1000 nits with the original curve, don't I now have one stop less?

--
Art Adams
Director of Photography
San Francisco Bay Area

Geoff Boyle
 

Don't the pq numbers refer to maximum levels?

Pq300 is 300 nits max pq1000 is 1000 nits max pq2000 is 2000 nits etc.

If it does work as Art suggests then yes, you'd lose a stop in the highlights and gain it in the shadows, maybe you could watch at home without a burka!


Geoff

On Wed, 27 Jun 2018, 19:30 Art Adams, <art.cml.only@...> wrote:
Hi all-

I'm seeing more and more references to PQ 200 curves. In HDR, until fairly recently, diffuse white was generally targeted at 100 nits, much the same as in SDR (if we all shot with cameras made in the 1990s). Now, though, diffuse white has apparently been moved up the scale to 200 nits, because modern consumer TVs have been brighter than 100 nits for quite a long time and there were a lot of complaints that 100 nits was too dark.

My question: doesn't that shave off a stop of dynamic range? If I had three more stops, plus a nudge, to get from diffuse white to 1000 nits with the original curve, don't I now have one stop less?

--
Art Adams
Director of Photography
San Francisco Bay Area

Art Adams
 

In Arri cameras they write this out as "ARRI 2100 PQ 1K 200", or Rec 2100 (HDR) Peak brightness (1000 nits) Diffuse white (200 nits).


On Wed, Jun 27, 2018 at 11:07 AM Geoff Boyle <geoff@...> wrote:
Don't the pq numbers refer to maximum levels?

Pq300 is 300 nits max pq1000 is 1000 nits max pq2000 is 2000 nits etc.

If it does work as Art suggests then yes, you'd lose a stop in the highlights and gain it in the shadows, maybe you could watch at home without a burka!


Geoff

On Wed, 27 Jun 2018, 19:30 Art Adams, <art.cml.only@...> wrote:
Hi all-

I'm seeing more and more references to PQ 200 curves. In HDR, until fairly recently, diffuse white was generally targeted at 100 nits, much the same as in SDR (if we all shot with cameras made in the 1990s). Now, though, diffuse white has apparently been moved up the scale to 200 nits, because modern consumer TVs have been brighter than 100 nits for quite a long time and there were a lot of complaints that 100 nits was too dark.

My question: doesn't that shave off a stop of dynamic range? If I had three more stops, plus a nudge, to get from diffuse white to 1000 nits with the original curve, don't I now have one stop less?

--
Art Adams
Director of Photography
San Francisco Bay Area



--
Art Adams
Director of Photography
San Francisco Bay Area

JD Houston
 

 it is a little more complicated than just losing a stop.

Peak brightness means you are displaying up to 1000 nits.
Peak brightness for SDR is 100 nits.

But for diffuse white for SDR, you are in essence talking about where to place the
90% white card reflection.  In SDR video, this might be at 91 IRE,  
AND it might be showing at 100 nits on the monitor.

In HDR, they are defining diffuse white as a 100% diffuse grey card
and placing it at 203 nits (with signal levels of 58% in PQ and 75% in HLG.

The mastering reference for SDR has not changed, although I understand that
some people might be tempted to use a higher number.  But QC for SDR signals is still based 
on feeding a 100% IRE signal to a 103 nits monitor (even though IRE is an analog standard, 
scopes still show it that way.)

Although you might think that many people have new flat panels showing higher brightness,
there are still 100 million older sets in the US that do not — even if they have migrated to 
the bedrooms, or the basement, they still need to get a good SDR signal, and mastering
for 100 nits is still essential for TV delivery. SDR video is a relative system anyway that was 
designed to work from 0 to 1.0 on ANY television brightness (with user adjustment available)

For HDR, ITU is putting the grey card (18% reflectance at 38% PQ and 38%HLG) as a common point between the
two HDR systems and they are placing that point for output at  26 nits.  This is for the convenience
of making HLG and PQ compatible and having HLG show correctly on SDR sets.

So in essence, they are saying an HDR signal should be brighter.
Colorists doing content have already been lifting mid-levels into the 20nit range for HDR1000.

So rather than focussing on what you might be losing (yeah about a stop sorta),
what you gain is increased clarity in the blacks by about a stop because the eye will actually see better
detail when shadow details are above 1 nit.   

Although OLEDs increase the dynamic range of blacks by quite a bit, they don’t really make dark
images any easier to see.  But doubling their brightness does.

I hope I haven’t confused too much from what is a simple question.

There are way more details than you might want to know in 
“Operational Practices in HDR Television Production”  ITU-R BT.2408-1   April 2018
an engineering document, not a cinematographers guide.


So this is a roundabout way of saying, for HDR, the recommendation is to give up some highlight range to make the 
image brighter.  Colorists are doing this anyway without any prompting from engineers.

For acquisition, you still have the option for non-live shows to place the exposure where
you want to either capture more highlight detail, or to capture less noisy shadows.


Jim Houston
Pasadena, CA 
Starwatcher Digital








On Jun 27, 2018, at 1:28 PM, Art Adams <art.cml.only@...> wrote:

In Arri cameras they write this out as "ARRI 2100 PQ 1K 200", or Rec 2100 (HDR) Peak brightness (1000 nits) Diffuse white (200 nits).

On Wed, Jun 27, 2018 at 11:07 AM Geoff Boyle <geoff@...> wrote:
Don't the pq numbers refer to maximum levels?

Pq300 is 300 nits max pq1000 is 1000 nits max pq2000 is 2000 nits etc.

If it does work as Art suggests then yes, you'd lose a stop in the highlights and gain it in the shadows, maybe you could watch at home without a burka!


Geoff

On Wed, 27 Jun 2018, 19:30 Art Adams, <art.cml.only@...> wrote:
Hi all-

I'm seeing more and more references to PQ 200 curves. In HDR, until fairly recently, diffuse white was generally targeted at 100 nits, much the same as in SDR (if we all shot with cameras made in the 1990s). Now, though, diffuse white has apparently been moved up the scale to 200 nits, because modern consumer TVs have been brighter than 100 nits for quite a long time and there were a lot of complaints that 100 nits was too dark.

My question: doesn't that shave off a stop of dynamic range? If I had three more stops, plus a nudge, to get from diffuse white to 1000 nits with the original curve, don't I now have one stop less?

--
Art Adams
Director of Photography
San Francisco Bay Area





--
Art Adams
Director of Photography
San Francisco Bay Area




Jim Houston
Consultant, Starwatcher Digital, Pasadena, CA

Art Adams
 

Jim, thanks. I learn a ton when you post. This makes sense. I guess I just wanted to confirm what I suspected, and also understand why it's happening. It's interesting that we're now focused on opening up shadows when for so long the holy grail was highlight detail. Fine by me.

I'm downloading the paper you linked to. Never assume it might be too much detail—I might not understand all of it, but I'm always interested in trying.

Thanks again.

--
Art Adams
Director of Photography
San Francisco Bay Area

Adam Wilt
 

In Arri cameras they write this out as “ARRI 2100 PQ 1K 200", or Rec 2100 (HDR) Peak brightness (1000 nits) Diffuse white (200 nits).

In the PDF that accompanies these LUTs, ARRI says: 

The 3D LUTs rendering diffuse white towards 200 cd/m2 are compatible with the BBC recommendations for creating HLG compatible HDR signals. 

HLG was designed with diffuse white at 50% (on a “traditional” 0%–100% WFM) but BBC practice is to expose for diffuse white at 73% for a brighter picture. HLG exposed in-camera this way — that is, with a camera in HLG-native mode — does indeed lose about a stop and a half off the highlights.

I pulled ARRI's PQ-1K-200 and PQ-1K-100 LUTS into LUTCalc (all hail Ben Turley! What would we ever do without LUTCalc?). It appears to me that the 200 nit LUT boosts “midtones” (in HDR, “midtones” include diffuse white, grin), brightening diffuse white without shifting exposure up or down. So you won’t lose a stop off the top; the LUT compresses highlights a bit more to allow for brighter “midtones”.

 

(This last one is an animated GIF flipping between the two LUTs; don’t know if it’ll display properly.)



Cheers,
Adam Wilt
technical services: consulting / coding / camerawork
Vancouver WA USA (no, not that Vancouver, the other one)

Adam Wilt
 

Jim said:

In HDR, they are defining diffuse white as a 100% diffuse grey card and placing it at 203 nits (with signal levels of 58% in PQ and 75% in HLG.

I said:

BBC practice is to expose [HLG] for diffuse white at 73% for a brighter picture…

We’re not in disagreement: the BBC practice is the ITU practice; I was talking about the 90% white reflectance level, not the 100% diffuse white.

Adam Wilt
technical services: consulting / coding / camerawork
Vancouver WA USA (no, not that Vancouver, the other one)

Art Adams
 

Interesting stuff, but I'm curious what reference anyone is using for 100% diffuse white... I didn't think such a thing was physically possible.

Thank you, gentlemen. I need to read more on this. And thanks, Adam, for the LUTCalc example. That is truly an awesome little program but not many people seem to know about it.

-- 
Art Adams
Director of Photography
San Francisco Bay Area

JD Houston
 


On Jun 29, 2018, at 2:09 AM, Art Adams <art.cml.only@...> wrote:

Interesting stuff, but I'm curious what reference anyone is using for 100% diffuse white... I didn't think such a thing was physically possible

It’s not.

Best you can get is about 99% with a ceramic called Spectralon.   But this is a Lab tool rather
than a field usable material. (in other words, it can get dusty/oily fast and lose it’s status).

Magnesium dioxide ceramic can get to about 98%. (Same Lab use)

The best titanium dioxide paints can get to 95%.

But really you don’t need to get 100%.  Just use a known reflector and figure out where it should be 
on the same curve.  So a paper chart should still work fine for setting exposure.

It is harder to get a properly lit card in many scenes that is close to a perfect average of the scene’s diffuse materials anyway
compared to say 90 vs. 92% reflectance paper.

I’ll ask the question too though, anyone use something other than the standard test charts?

Jim Houston
Pasadena, CA


Geoff Boyle
 

I’ve wanted to follow up on this but I’ve been too busy with house sales and purchases…

 

So, I’ve taken a Kodak Gray scale plus and exposed at what my meter says is correct.

 

I put it in Resolve in ACES using bog standard IDT and 709 ODT and it looks great on a calibrated 709 monitor.

 

I then change the ODT to P3-D65 ST2084 (1000 nits) and switch my monitor to PQ 1000 and I find I need to increase the offset in Resolve by exactly a stop to get what looks right to me.

 

Just switching between the two ends up with an image that looks too dark, I’m also looking at the junk around the chart 😊

 

So, empirically it looks like a stop brighter is about right.

 

Cheers

 

Geoff Boyle NSC FBKS

Cinematographer

Netherlands

www.gboyle.nl