Topics

Dynamic Range Test - surprising results


Axel Gimenez
 

I devised a test to measure the dynamic range of a camera. I tested the Alexa Mini and managed to get only 12 stops out of the 14+ stops listed on the ARRI website. What's going on?
 
For my test, I kept the camera settings, EI, shutter speed and T-Stop, ARRI rec709 constant and adjusted the intensity of the light hitting an 18% gray card.
 
Using a spot meter to measure the light hitting the gray card, I adjusted the light intensity in 1/3 stop increments.
 
At each 1/3 stop increment, I captured a couple of seconds of footage of the gray card and noted the IRE value on a waveform monitor. 
 
That effort produced the attached jpg. Does this make sense or am I doing something wrong?
 
Thanks,
Axel


Geoff Boyle
 

Hi Axel,

 

You’re doing nothing wrong.

The manufacturers, all of them, live in a different universe.

Your results reflect those of the CML tests https://cinematography.net/CineRant/2018/07/30/personal-comments-on-the-2018-cml-camera-evaluations/ these were my personal comments on the camera tests.

 

You can download the raw files from the tests and see for yourself all of the main cameras exposed at +/- 7 stops and see what they actually do.

 

If you are talking about what can be measure buried in noise, then you can extend the DR to what manufacturers claim. Real world use shows what you have found is correct.

 

cheers
Geoff Boyle NSC FBKS
EU based cinematographer
+31 637155076

www.gboyle.nl

www.cinematography.net

 

 

From: cml-mentor@... <cml-mentor@...> On Behalf Of axeltg@...
Sent: 15 December 2019 07:02
To: cml-mentor@...
Subject: [cml-mentor] Dynamic Range Test - surprising results

 

I devised a test to measure the dynamic range of a camera. I tested the Alexa Mini and managed to get only 12 stops out of the 14+ stops listed on the ARRI website. What's going on?

 

For my test, I kept the camera settings, EI, shutter speed and T-Stop, ARRI rec709 constant and adjusted the intensity of the light hitting an 18% gray card.

 

Using a spot meter to measure the light hitting the gray card, I adjusted the light intensity in 1/3 stop increments.

 

At each 1/3 stop increment, I captured a couple of seconds of footage of the gray card and noted the IRE value on a waveform monitor. 

 

That effort produced the attached jpg. Does this make sense or am I doing something wrong?

 

Thanks,
Axel


Axel Gimenez
 

Thank you Geoff! Glad to know I’m on the right track. Reading your linked post, I have another question (rating cameras) - I’ll create a new thread for clarity. 


Merritt Mullen
 

Isn’t the 14+ stops what is captured in Log-C / raw? Rec.709 gamma would bury some of the shadow information to get clean, solid blacks — I think you’d find those last two stops there, sliding into the noise floor.

David Mullen, ASC
Los Angeles


Geoff Boyle
 

David is of course right, however all the CML tests are done raw.
If you don't use noise reduction then 12 stops is accurate.

Geoff Boyle NSC FBKS




On Mon, Dec 16, 2019 at 3:42 AM +0100, "Merritt Mullen via Cml.News" <mdmullen1=verizon.net@...> wrote:

Isn’t the 14+ stops what is captured in Log-C / raw? Rec.709 gamma would bury some of the shadow information to get clean, solid blacks — I think you’d find those last two stops there, sliding into the noise floor.

David Mullen, ASC
Los Angeles


Axel Gimenez
 

Hi David,

hmmm. So I shot the footage in Pro Res 4444XQ Log-C with the Rec.709 added on top. I was thinking the same as what you wrote but I noticed that adding or subtracting the rec.709 from the log-c footage only shifted where it appeared on the waveform monitor but didn't add anything meaningful to the footage. 

I discovered this by mistake. I was feeding log-c to the waveform monitor and it was clipping well below 100 IRE. Then when I added rec.709 the clipped input jumped up to 100IRE as I expected.

Perhaps it is different if you shoot in actual raw instead of a log-c prores?

After the holidays, I will test and share.

Thanks,
Axel


Igor Trajkovski
 

I remember Art Adams had interesting article on PVC about Alexa's DN.
Should be same for Mini too.

Here the link - material for comparison of insights:


Art Adams
 

>I remember Art Adams had interesting article on PVC about Alexa's DN.

That’s an old one, so I hope it’s mostly accurate. 😊 I know a lot more now.

I recently shot some tests comparing noise on a Mini at 3.4K to full frame 4.5K on a Mini LF and found that the increased number of photo sites pushes perceived noise down to the point where the Mini LF can be rated at 1/3-1/2 stops faster than the Mini.

And I still easily see 14 stops in Log C, and without counting additional stops that aren’t effectively usable.

 

_______________________________________________________
Art 
Adams
Cinema Lens Specialist
ARRI Inc.
3700 Vanowen Street
BurbankCA 91505
www.arri.com 

 
aadams@...
Get all the latest information from www.arri.comFacebookTwitterInstagram and YouTube.






This message is confidential. It may also be privileged or otherwise protected by work product immunity or other legal rules. If you have received it by mistake, please let us know by e-mail reply and delete it from your system; you may not copy this message or disclose its contents to anyone. Please send us by fax any message containing deadlines as incoming e-mails are not screened for response deadlines. The integrity and security of this message cannot be guaranteed on the Internet.



Art Adams
 

>That effort produced the attached jpg. Does this make sense or am I doing something wrong?

 

The Rec 709 standard is only meant to reproduce six linear stops of dynamic range on a display. Highlights and shadows are compressed with that LUT applied, because there’s no way to show 14 linear stops of dynamic range on a Rec 709 monitor.

 

Do the test again in Log C and you’ll see different results. Also, one thing I’ve done in the past is to wrap the chart in black wedding veil so you can see when the pattern disappears. That tends to reveal noise reduction. There are several cameras out there that advertise 16 stops of dynamic range, but those last two or three stops are good for tonal values only. There won’t be any detail.

_______________________________________________________
Art 
Adams
Cinema Lens Specialist
ARRI Inc.
3700 Vanowen Street
BurbankCA 91505
www.arri.com 

 
aadams@...
Get all the latest information from www.arri.comFacebookTwitterInstagram and YouTube.






This message is confidential. It may also be privileged or otherwise protected by work product immunity or other legal rules. If you have received it by mistake, please let us know by e-mail reply and delete it from your system; you may not copy this message or disclose its contents to anyone. Please send us by fax any message containing deadlines as incoming e-mails are not screened for response deadlines. The integrity and security of this message cannot be guaranteed on the Internet.



Igor Trajkovski
 

I remember Art Adams had interesting article on PVC about Alexa's DN.
Should be same for Mini too.

Here the link - material for comparison of insights:

https://www.provideocoalition.com/alexa_dynamic_range_its_all_in_how_you_use_it/


PS: Sorry if it is a double post. Said it will "bounce" if not signed so.. here signed :)


Igor Trajkovski
Multimedia Artist
Macedonia, Europe


Axel Gimenez
 

Hi Art,

Both you and David have suggested doing the test again with Log-C and I will. I've noticed the 6 linear stops in rec.709. I think I'm going to expand my dynamic range test to think about how to "lay" the stops of the camera into rec.709 as that is the final color space I finish too. Although, I assume it changes once you start finishing in HDR for theatrical release, etc. 

Question for you: in a previous post you talk about over-exposing skin tones 4-5 stops - what happens to your highlights? Or are the skin tones the brightest thing in your scenes?


Merritt Mullen
 

Film negative captures more stops of dynamic range than a film print can display, because a projected image needs a certain amount of contrast and deep blacks to look good. Rec.709 is similar in that the number of stops that are displayed are less than is captured in log/raw in order to get a good-looking image within the brightness range of a Rec.709 monitor.

If you display all 14-stops of dynamic range within the 10-stops of a print or in Rec.709, you’ll get a low-contrast image. With an HDR monitor, you can display all 14-stops and have a rich image with deep blacks and hot whites.

The advantage of capturing more stops of information than can be displayed is that you have more latitude to correct the image up or down, plus selectively pull detail out of the shadows or bright highlights when needed.

David Mullen, ASC
Los Angeles

Both you and David have suggested doing the test again with Log-C and I will. I've noticed the 6 linear stops in rec.709. I think I'm going to expand my dynamic range test to think about how to "lay" the stops of the camera into rec.709 as that is the final color space I finish too. Although, I assume it changes once you start finishing in HDR for theatrical release, etc.


Geoff Boyle
 

Sign your messages Axel or they will start bouncing.

 

Name, job & location are the rules, it really does make it easier for people to answer.

 

cheers
Geoff Boyle NSC FBKS
EU based cinematographer
+31 637155076

www.gboyle.nl

www.cinematography.net

 

 

From: cml-mentor@... <cml-mentor@...> On Behalf Of axeltg@...
Sent: 18 December 2019 05:46
To: cml-mentor@...
Subject: Re: [cml-mentor] Dynamic Range Test - surprising results

 

Hi Art,

Both you and David have suggested doing the test again with Log-C and I will. I've noticed the 6 linear stops in rec.709. I think I'm going to expand my dynamic range test to think about how to "lay" the stops of the camera into rec.709 as that is the final color space I finish too. Although, I assume it changes once you start finishing in HDR for theatrical release, etc. 

Question for you: in a previous post you talk about over-exposing skin tones 4-5 stops - what happens to your highlights? Or are the skin tones the brightest thing in your scenes?


Axel Gimenez
 

Sorry Geoff, was not aware. Issue corrected. Thanks.
--
Axel Gimenez
Commercial Auteur
NYC
https://axelgimenez.com/


Art Adams
 

>Question for you: in a previous post you talk about over-exposing skin tones 4-5 stops - what happens to your highlights? Or are the skin tones the brightest thing in your scenes?

 

That’s an artistic choice. If someone walks through a bright patch of light, or walks across a room to stand in front of a brightly lit window, it wasn’t a big deal to let people burn out (as long as it looked correct). Most other cameras couldn’t (and still can’t) handle that kind of extreme without showing terrible chroma clipping and very coarse highlight roll-off. Alexa, from the beginning, tolerated this kind of thing beautifully, with a soft roll-off that looked much like film.

 

I wasn’t rating the camera 4-5 stops faster, but simply allowing people to become brighter or darker within a frame to the same extent that I did in film, without worrying about technical video issues that prevented such things when working with earlier cameras, where I’d have to light much flatter and control highlights to a greater extreme due to technical limitations.

 

In general, I always popped people 1/3-2/3 stop in my commercial work because I liked them to stand out a bit more.

_______________________________________________________
Art 
Adams
Cinema Lens Specialist
ARRI Inc.
3700 Vanowen Street
BurbankCA 91505
www.arri.com 

 
aadams@...
Get all the latest information from www.arri.comFacebookTwitterInstagram and YouTube.






This message is confidential. It may also be privileged or otherwise protected by work product immunity or other legal rules. If you have received it by mistake, please let us know by e-mail reply and delete it from your system; you may not copy this message or disclose its contents to anyone. Please send us by fax any message containing deadlines as incoming e-mails are not screened for response deadlines. The integrity and security of this message cannot be guaranteed on the Internet.