Meaning of 16.5 stops of Dynamic Range

[Andrey Blanco]

What exactly RED means with 16.5 stops of dynamic range?

I saw some ungraded side by side real shooting situation test (not charts) showing Alexa, Red Dragon and F55 and they look very similar to me in relations to the Shadow Bright areas. I would say the Alexa show a little bit more info on the highlights. I think these cameras are really close on DR.

I know a lot of magic can be done in post with Resolve these days.
So, when they refer to 16.5 they means to the usable latitude we have available in post?

 

[John Marchant]

In my opinion, and there are others , the camera records about 16.5 stops of DR. What you see on an LCD with unarguable ease in RG4 is way less, probably about 11. Shadows are crushed substantially and stops are wasted into pretty highlight rolloff.

In RLF without arguments about 'what's a stop' and 'is that too noisy?' you see about 14.5. More stops than that usually start the arguments I've discovered...

If you get back to the edit suite and go digging to make the most of your shadows and highlights, you'll find over 16. IF the scene has it. So many just don't even contain 16+ stops to start with. What you'll see though, if visualising the range is all you aim to do, is a pretty flat image. 16.5 stops of capture is not always inspiring to look at IMHO unless 'worked' through smart grading, power windows etc.

Our display DR is simply not great enough that we can make on the spot judgement about this much dynamic range.

Often with an artificially lit scene its hard to create enough workable dynamic range to be revealing of the difference between a 14 and a 15 or 16 stop camera. IMO the greatest proof of Dragon's DR is it post flexibility, and the only people who get really tactile with that are colorists. And FWIW colorists seem to be a part of the industry most sold of Dragon's capability.

 

[Scott Crawley]

:hurray::hurray::hurray:

Well said.

In my opinion, and there are others , the camera records about 16.5 stops of DR. What you see on an LCD with unarguable ease in RG4 is way less, probably about 11. Shadows are crushed substantially and stops are wasted into pretty highlight rolloff.

In RLF without arguments about 'what's a stop' and 'is that too noisy?' you see about 14.5. More stops than that usually start the arguments I've discovered...

If you get back to the edit suite and go digging to make the most of your shadows and highlights, you'll find over 16. IF the scene has it. So many just don't even contain 16+ stops to start with. What you'll see though, if visualising the range is all you aim to do, is a pretty flat image. 16.5 stops of capture is not always inspiring to look at IMHO unless 'worked' through smart grading, power windows etc.

Our display DR is simply not great enough that we can make on the spot judgement about this much dynamic range.

Often with an artificially lit scene its hard to create enough workable dynamic range to be revealing of the difference between a 14 and a 15 or 16 stop camera. IMO the greatest proof of Dragon's DR is it post flexibility, and the only people who get really tactile with that are colorists. And FWIW colorists seem to be a part of the industry most sold of Dragon's capability.

 

[Andrey Blanco]

I understand dragon offers other advatnges (like lower noise levels, Better Crop Factor, better color Science, etc)


So These 16.5 stops of Dynamic Range in practice is something nice to have if we get in trouble and can bring details in Post. but not necessary something that will used all the time if we have a very well exposed image?

How Would you measure the Dynamic Range on this test?:

http://www.cinematography.net/UWE/index.html

 

[Scott Crawley]

It's not just for when you get into trouble, it enables creative choices and flexibility throughout production. For example; It allows the DP to shoot faces in relative shadow, and Producers to brighten them up later if they don't like them that way. If you look at the same scenario another way you might spend a little less time filling faces on the day if you know you can fine tune it later. (Not that that is the best M.O.)

Edit: you don't necessarily need 16 stops to be able to do that, but the more DR, the more bits at your disposal the further you can go, and cleaner.

I understand dragon offers other advatnges (like lower noise levels, Better Crop Factor, better color Science, etc)


So These 16.5 stops of Dynamic Range in practice is something nice to have if we get in trouble and can bring details in Post. but not necessary something that will used all the time if we have a very well exposed image?

How Would you measure the Dynamic Range on this test?:

http://www.cinematography.net/UWE/index.html

 

[Julio Quintana]

So These 16.5 stops of Dynamic Range in practice is something nice to have if we get in trouble and can bring details in Post. but not necessary something that will used all the time if we have a very well exposed image?

Try to shoot a naturally lit day interior and hold the highlights outside the window. You should introduce more light into the room or net/ND the window, but it helps tremendously to have a sensor that makes those windows recoverable in post.

 

[Mark K.]

Dynamic Range doesn't exist in isolation. How a camera spreads its DR across over/under-exposure latitude plays a huge part - for example if Dragon holds 7 stops of overexposure (9 stops of shadow) - then it won't be able to hold highlights any better than an Alexa with its 7 stops of exposure headroom, (but it will see miles into the shadows) - sp it all depends on that distribution.

 

[Meryem Ersoz]

Try to shoot a naturally lit day interior and hold the highlights outside the window. You should introduce more light into the room or net/ND the window, but it helps tremendously to have a sensor that makes those windows recoverable in post.

especially true in places with a ton of bright sunny days, like Colorado - goes double when shooting in reflective snow. also pretty much anytime you shoot daylight through car windows and want a natural, not blown, looking scene.

 

[Andrey Blanco]

So Buying a Red Dragon for London or Paris will be really pointless? not a lot of sunny days

especially true in places with a ton of bright sunny days, like Colorado - goes double when shooting in reflective snow. also pretty much anytime you shoot daylight through car windows and want a natural, not blown, looking scene.

 

[Andrey Blanco]

Now seriously, What I don't undersand is why looking at the test done here:

http://www.cinematography.net/UWE/index.html

in the interiors scene I can see from the ungraded files very easily Alexa getting better dynamic range than the Dragon

on the exterior scene they look very close to me probably because there is not anything to make reference to (like the grayscale bar on the studio)

http://blog.gunleik.com/2014/03/19/...d-scaling-dragon-alexa-f55-c300-pt-1-preface/

 

[Phil Holland]

Andrey, that test was done a bit ago on old tech and isn't exactly the best way to "measure" dynamic range. John, myself, and a few others measured step by step exactly what the camera sees to find the actual number an where it was placed. The difficulty here was at the time Dragon was "unwrangled" in terms of it's color science and hardware.


That basically is why there are 1000 reasons to do that test again now that we have:

- The New OLPF
- The New Calibration
- DRAGONcolor

Well 3 reasons at least.

As well as reading out the measured steps of DR. Doing this with all cameras is the only way to find out how much DR is available.

 

[Brian Merlen]

you do know this test was done using the old OLPF and not using the new color science..? For that alone I wouldn't put much faith in it until its redone or you get your hands on the actual camera... plus the range isn't the only selling point obviously...price, form factor, resolution, clients acceptance, all of this plays a critical role not just total dynamic range...plus technically you can run HDRX on Dragon to get even more anyway


edit phil beat me to it...

 

[John Marchant]

Also, I have a consistent bone to pick with one aspect of the way people conduct comparative testing. It crops up again in the UWE tests, as well thought out as they are.

Its fair to say that you need to make an aesthetic judgment on acceptable image noise. Its possible to be scientific on this, but that doesn't suit the nature of the choices and decisions people will make with footage outside of a lab environment.

One consistent report I've heard is that Alexa still has the same or more DR, but Dragon is way cleaner in the shadow... to me this is nonsense - you need to match perceived shadow noise as a baseline. If you're happy with the Arri's noise character, then rate the Red at an ISO where its noise character is comparable. THEN see where your DR gets you. Stating that its cleaner at the bottom but has no more stops available at the top is missing the point.

 

[Andrey Blanco]

Hey Phil, are you in Vegas? Would be Great with all those Cameras Together and experts, do some side by side tests.

Having for example Alister Chapman testing a F55, yourself a Red Dragon, Shane Hurlbut a C500 or Alexa, etc...

An truly Unbiased Camera Test

Andrey, that test was done a bit ago on old tech and isn't exactly the best way to "measure" dynamic range. John, myself, and a few others measured step by step exactly what the camera sees to find the actual number an where it was placed. The difficulty here was at the time Dragon was "unwrangled" in terms of it's color science and hardware.


That basically is why there are 1000 reasons to do that test again now that we have:

- The New OLPF
- The New Calibration
- DRAGONcolor

Well 3 reasons at least.

As well as reading out the measured steps of DR. Doing this with all cameras is the only way to find out how much DR is available.

 

[Phil Holland]

lol. Yes. We'll take a full day away from NAB today and strip down some of those cameras from various places on the floor.

There will be another test soon, but it won't be happening here for sure. Heck, I don't have any of my charts here.

 

[Mark Wuerthner]

I understand dragon offers other advatnges (like lower noise levels, Better Crop Factor, better color Science, etc)


So These 16.5 stops of Dynamic Range in practice is something nice to have if we get in trouble and can bring details in Post. but not necessary something that will used all the time if we have a very well exposed image?

How Would you measure the Dynamic Range on this test?:

http://www.cinematography.net/UWE/index.html

Its actually fairly simple. If you look at the DSC chart, the gray chip in the middle of the chart (the chip above and below the center black chip) is 18% gray. As you go up in overexposure, when that middle chip merges with the lighter toned chip next to it, the sensor had clipped the 18% chip and that is your overexposure amount of stops above middle gray (in this case, for dragon it clips between 6 and 6.5 stops.) For underexposure you do the same thing, except, clipping is not so obvious down in the darks. At that point you're dealing with noise so and the noise floor of the sensor is harder to distinguish. As a previous commenter noted people can have different interpretations of what is and isn't a "stop" down in the darks. Some people think that dragon has 8-9 stops below middle gray. If you were to see this image on a waveform monitor the "toe" of the monitor would be pretty flat. Whether you can pull info out of there I don't know. Looking at the images on CML alexa and dragon look fairly similar down there. Maybe dragon's cleaner? Or maybe because there is higher resolution the noise is smaller? Based on these images I see the same "amount" of noise on both but dragon noise looks smaller to me.

 

[Dr. Sassi]

Shortest answer*: Signal Noise Ratio. ;o) -- and how often it is sampled between the two extremes.

EPIC MX S/N RATIO: 66db

DRAGON SN RATIO: 80db

(*perhaps too technical)

 

[Nick Morrison]

Its actually fairly simple. If you look at the DSC chart, the gray chip in the middle of the chart (the chip above and below the center black chip) is 18% gray. As you go up in overexposure, when that middle chip merges with the lighter toned chip next to it, the sensor had clipped the 18% chip and that is your overexposure amount of stops above middle gray (in this case, for dragon it clips between 6 and 6.5 stops.) For underexposure you do the same thing, except, clipping is not so obvious down in the darks. At that point you're dealing with noise so and the noise floor of the sensor is harder to distinguish. As a previous commenter noted people can have different interpretations of what is and isn't a "stop" down in the darks. Some people think that dragon has 8-9 stops below middle gray. If you were to see this image on a waveform monitor the "toe" of the monitor would be pretty flat. Whether you can pull info out of there I don't know. Looking at the images on CML alexa and dragon look fairly similar down there. Maybe dragon's cleaner? Or maybe because there is higher resolution the noise is smaller? Based on these images I see the same "amount" of noise on both but dragon noise looks smaller to me.

To me, simple fact that Dragon has Alexa performance at:

1/3 Size
1/3 Cost
6K!!!!

Is a remarkable technical feat.

You can fly a 6K camera on a gimbal with the DR of a Alexa, and shoot RAW!

Exciting times...

 

[John Marchant]

Looking at the images on CML alexa and dragon look fairly similar down there. Maybe dragon's cleaner? Or maybe because there is higher resolution the noise is smaller? Based on these images I see the same "amount" of noise on both but dragon noise looks smaller to me.

Smaller noise, all things being equal IS less noise...

But check Gunleik's analysis of the UWE images. Its very clear that the way Dragon was tested it was producing much cleaner shadows than the Alexa. Opinion will vary on what that means I guess, but I've said what I think.

http://static2.oneclick.gunleik.com/2014/03/d4f0a781a8feecf576448d1512bf5aa1c56d686d.jpg <- view it full size...

(with Apologies to G for pulling one of his images out of context)

 

[Mark Wuerthner]

Smaller noise, all things being equal IS less noise...

But check Gunleik's analysis of the UWE images. Its very clear that the way Dragon was tested it was producing much cleaner shadows than the Alexa. Opinion will vary on what that means I guess, but I've said what I think.

http://static2.oneclick.gunleik.com/2014/03/d4f0a781a8feecf576448d1512bf5aa1c56d686d.jpg <- view it full size...

(with Apologies to G for pulling one of his images out of context)

Yes I've read Gunleik's piece on this. If you haven't seen this you might take a look at it:
http://community.sony.com/t5/F5-F55...ange-results/m-p/300212/highlight/true#M13774
Many Sony users take issue with the way Gunleik developed the f55 picture especially in regards to noise. And frankly, after reading their posts, I don't blame them. So, I'm not really sure what to make of all of that.I'm curious though about something you wrote earlier, about matching noise and then doing dynamic range tests. I don't know how this would work or what process you're talking about. If you could elaborate that would be great.