Jump to content

Underexposure as a Tool


AJ Young

Recommended Posts

 

As far as I know, film works a tad differently and the ISO of film is actually dependent on the grain structure of the emulsion (smaller silver halide vs. larger silver halide) and doesn't result in a dramatic change in dynamic range between the shadows/highlights of different film ISO's. However, "re-rating" the ISO of a stock (under/overexposing) will yield the same results.

 

 

Film film ISO is 'very different' from how Digital Film ISO is calculated. For Film film, a densitometer is used to determine an exposure to yield a .1 log reading above the film base + fog of the development process. The time for development was such that the resulting gamma was .5.

 

While I don't know what may have been required of cinematography students, for still photography students taking advance exposure classes, going through a 'zone system' procedure, using densitometers and various development processes, one was required to determine an effective 'ASA/ISO/Exposure Index'.

 

There are several specs for how to determine a Digital Film Exposure Index, or ISO rating... but the specs include has sort of 'vague' what looks good to the engineers in terms of picture quality.

 

I think ARRI engineers spent some time analyzing Film Film characteristics and developed/rated their camera's response so as to minimize transitioning from Film Film to Digital Film.

 

I use the terms 'Film Film' and 'Digital Film' to emphasize that the goal of the digital cameras, either stills or motion pictures, are to 'replace Film' and were not really designed to match analog video characteristics. Few Film Film photographers and cinematographers where all that familiar with vidicon tube responses, limitations, and the measuring equipment relative to 'video', and so, information and understanding of that class of equipment had to be acquired.

 

There has been some seepage of video terms into Digital Film, but not that much. One now speaks of Dynamic Range, rather than 'latitude', or one speaks of Signal to Noise, SNR, in terms of dB values, rather than 'film grain'(*), etc.

Link to comment
Share on other sites

  • Premium Member

I still find it hard for me sometimes, especially after a long day of shooting, to switch into digital mode; Digital sensors have a maximum photon threshold, and film has a minimum photon threshold.

 

And, because of this, I find myself doing far more tests with digital cameras, as they tend to reach into the darkness...

Link to comment
Share on other sites

I think the key was how I reinterpreted the data in the grade through the ProRes file. I pushed and pulled parts of the new exposure to shape the dynamic range because the image was underexposed in the ProRes file. If it were RAW, rating it to 6400ISO wouldn't give me the same results in comparison to a combination of gain/gamma/lift to shadow drop off and highlight roll off.

It sounds like you are using a similar technique to stills photographers, who use ISO invariance as a tool. They deliberately underexpose their images to protect highlights, and then selectively brighten parts of the image in post, rather than the blanket adjustment of changing ISO. The technique works very well with Nikon cameras which exhibit very little difference in noise between the two techniques, but less well with Canon cameras, which apparently show markedly more noise when the ISO is changed in post, rather than in camera.

 

I'd imagine that there are similar differences between digital cinema cameras as well. Certainly, the Alexa's highlight response is similar enough to that of film to make aggressive protection of highlights unnecessary.

Link to comment
Share on other sites

  • Premium Member

Yes, this is how I shoot on my Nikon, use the lowest ASA rating and underexpose to protect the highlights but have a clean enough signal to bring up the shadows. Unfortunately since I shoot both raw and jpeg, it means my jpegs are also dark and thus need to be color-corrected. I guess there must be a away to create a custom profile for jpegs that have a brightness boost.

Link to comment
Share on other sites

  • 3 weeks later...

Miguel

 

As an aside I was just last night shooting with the new -ish Panasonic varicam LT.. which has this mysterious double native ISO setting adjustment .. 800 and 5000..

I'm not sure of the full science behind it, but from what I've read the Varicam has two analog circuits that allow two native ISO's behind each pixel. So when you switch the iso from 800 to 5000, the camera switches circuits to deliver a similar signal to noise ratio of 800 ISO. I saw it in action two weeks ago I 2nd'ed on, and it was awesome! We only used a bounce from a window for an interior scene. 5000 ISO allowed one source to illuminate the whole room, and even in shadows we didn't see any noise.

Link to comment
Share on other sites

I'm not sure of the full science behind it, but from what I've read the Varicam has two analog circuits that allow two native ISO's behind each pixel. So when you switch the iso from 800 to 5000, the camera switches circuits to deliver a similar signal to noise ratio of 800 ISO. I saw it in action two weeks ago I 2nd'ed on, and it was awesome! We only used a bounce from a window for an interior scene. 5000 ISO allowed one source to illuminate the whole room, and even in shadows we didn't see any noise.

 

 

Yes quite amazing.. I was shooting high speed 50fps.. up a mountain in the night ..lit by fire/torches... but there were alot of them it has to be said..

Link to comment
Share on other sites

  • 3 years later...

My apologies for drudging up an old topic, but I wanted to update my experimentation with this method:

I've since used this technique on two other projects and loved the results. Essentially, I underexpose the image by a certain number of stops intentionally and then recover it in post for a desired look. This was partially inspired by Martha Marcy May Marlene and Birth, but also inspired by how film responds to light (poor shadow detail in comparison to how much highlight detail film, particularly Vision 3, can handle)

The idea behind this method is to control how the camera's dynamic range is distributed. The general theory with ISO and digital cameras, particularly in RAW, is that changing the ISO doesn't make the sensor more or less sensitive, it just reads the RAW data differently (there are a few exceptions, of course). When a cinematographer deviates from the native ISO, they are under/overexposing the image and the new image has the dynamic range re-mapped by the debayering process. However, I wanted to control how that dynamic range was redistributed. Is this something we can do? Something we can test?

I recently finished an experiment and would love for everyone to check my work, to make sure what I'm experimenting with is actually correct. You can find the detailed (and very nerdy) experiment here: http://www.ajyoungdp.com/articles/blog/OverUnder01/

Here's a tease with all of the over/underexposures from the experiment in one image:

OverUnder-Thumb.jpg

What do you think? What are your thoughts? Am I off base here?

Link to comment
Share on other sites

  • Premium Member

Since film negative has a logarithmic response to light, heavy underexposure puts more shadow and midtone detail in the flatter (and grainier) toe of the characteristic curve. With digital, the log version is applied to the raw linear sensor readout so rating the camera faster doesn't make it lower in contrast if recording raw, in theory. However, in your case, by recording Log-C ProRes and underexposing, you were emulating more of a film response by pulling up information from the bottom of the log curve.

Link to comment
Share on other sites

Just so I can understand it right: when changing ISO of the RAW data, does the camera apply the log curve to that portion of the linear data? I've seen a chart like this floating around on the internet:

dynamic-range.jpg

but I'm not too sure if the concept applies to both RAW and Log images, or if the concept is flawed overall.

A lot of questions I've been getting since releasing this experiment are regarding RAW. Deviating from the native ISO begins to affect the dynamic range of the camera, but if we change the ISO of the RAW data, then does that mean we don't shift the dynamic range but disregard whatever portion of the linear RAW data doesn't fit the log curve?

Link to comment
Share on other sites

  • Premium Member

I think that's correct in terms of dynamic range recorded but if you are correcting from a Log-C file, the lowest and highest stops are compressed into a flatter gamma.  Now Arriraw is also technically a 12-bit log file -- I don't know if that means the gamma is mapped into a photographic-type log curve like Log-C. Maybe Phil Rhodes can explain.

The thing to keep in mind about dynamic range is that while you record 14-stops no matter what ISO you have, your shot content might not contain 14-stops of range. You may be underexposing a subject for example with no hot highlights so you are losing shadow detail without necessarily gaining more highlight detail.

Link to comment
Share on other sites

That actually explains a lot for why I barely noticed a difference with my highlights when I underexposed by one top vs three stops.

18 hours ago, David Mullen ASC said:

The thing to keep in mind about dynamic range is that while you record 14-stops no matter what ISO you have, your shot content might not contain 14-stops of range. You may be underexposing a subject for example with no hot highlights so you are losing shadow detail without necessarily gaining more highlight detail.

I think is a very important note that I failed to realize. Are you okay if I paraphrase this or quote you on the blog post?

Regardless, thank you!

Link to comment
Share on other sites

  • Premium Member

Sure!
You can imagine a scenario, for example, of a caucasian actor in a medium grey T-shirt against a green wall of shrubs with the sun dead overhead where the hottest highlight is perhaps the shine on his cheek.  So if you underexpose so that your 14-stop range becomes 11-stops of highlights and 3-stops of shadow information but your highlights never get beyond 8-stops above middle grey, then you've lost a few stops of dynamic range.

Link to comment
Share on other sites

Ah, that makes a lot of sense. I used this technique on a western short film that was entirely at night and what you exactly described happened. We barely noticed a change in the highlights because almost nothing in the film went over say 70% IRE!

Link to comment
Share on other sites

Pardon me if I sound like a technically challanged person. What I know ,when I underexpose digital sensor I will raise the noise floor while trying to lift the shadows. 
So I wanna ask:

Was it intentional to raise the noise to add grain to images ?(sample images look amazing needless to say and it really match film response)

However, I came across other ways to match film response, either mapping grain to different luminance range or using plug ins like Film Convert or may be using Nuke (I have no idea but Steve Yedlin showed some examples)

Do these processes really work for theatrical releases? I did test and it looked good on my consumer level monitor ,not sure about theatre screens.

Also Juan Melara have some good blogs regarding this which I don’t understand completely.

https://juanmelara.com.au/

Link to comment
Share on other sites

18 hours ago, Saikat Chattopadhyay said:

Was it intentional to raise the noise to add grain to images ?(sample images look amazing needless to say and it really match film response)

I believed that, in my approach, grain was just one piece of the puzzle. There are multiple factors that play into the look of film vs digital that the underexposure technique was hoping to replicate, chief among them was highlight response. Here's what I mean:

https://petapixel.com/2015/08/10/how-much-can-you-overexpose-negative-film-have-a-look/

When we screened our tests for The Watchman's Canoe, we watched them via DCP on a gigantic theatre screen in the local town we were shooting in (at the time, the test screening was during our tech scout). It looked great! ?

On film, Martha Marcy May Marlene and Birth are two prime examples of a theatrical release.

Edited by AJ Young
"that my underexposure technique" changed to "that the underexposure technique" because I did not create the technique, nor do I own it
  • Upvote 1
Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...