Log a hyperthetical question

Mike Brennan · July 18, 2007

I'm wondering if decision to use of log space to record camera output should be subject dependent.

Given that log uses up more bits in the shadows at the expense of the rest of the picture,

can we imagine a project where log recording would be counter productive to the overall look?

We are more sensative to information in the shadows, but how often is the important part of the image in the shadows?

Like a beach movie? Ski or field Sports movie?

If bright punchy images with rich blacks are required, is this a reason not to use log recording on a f23/f950 ?

I've shot on Viper filmstream and regularly use f900 user curves ect and am wondering what others think...

Mike Brennan

John Sprung · July 18, 2007

I'm wondering if decision to use of log space to record camera output should be subject dependent.
Given that log uses up more bits in the shadows at the expense of the rest of the picture, ...

It's not really at the expense of the rest of the picture. Log merely refrains from wasting bits on differences that humans can't see, and uses them to get deeper into the shadow detail.

The smallest difference in brightness that most people can see is about 1%. So, in arbitrary numbers, if you can just barely see the difference between 100 and 101, the difference between 1000 and 1010 wastes nine possible levels, and the difference between 10 and 11 looks quite large. It's sort of like the reason that we have a difference of only 0.4 between f/1.0 and f/1.4, and a difference of 5.0 between f/11 and f/16. With linear f/stops, you'd have clicks at f/11.0, f/11.4, f/11.8, etc.

-- J.S.

Mike Brennan · July 26, 2007

It's not really at the expense of the rest of the picture. Log merely refrains from wasting bits on differences that humans can't see, and uses them to get deeper into the shadow detail.

The smallest difference in brightness that most people can see is about 1%. So, in arbitrary numbers, if you can just barely see the difference between 100 and 101, the difference between 1000 and 1010 wastes nine possible levels, and the difference between 10 and 11 looks quite large. It's sort of like the reason that we have a difference of only 0.4 between f/1.0 and f/1.4, and a difference of 5.0 between f/11 and f/16. With linear f/stops, you'd have clicks at f/11.0, f/11.4, f/11.8, etc.

-- J.S.

I understand the concept that we don't see the smallest of differences in brightness compared to shadow, but we are talking about recording an image from a device that is already far less able to precieve what the eye sees.

So shouldn't we be worried about capturing a good range of highlight and shadows?

A connected question, is the eyes relative lack of ability to resolve brightness any different when we are watching TV, cinema or on a sunny day at the beach? Watching different scenes with different contrast and absolute brightness?

If the eye has difficulty with bright objects then surely we shouldn't be sacrificing tonal values at the high end where there is often information that the cinematographer needs to impart on the viewer.

Put the bits where the eye needs most help or where the eye is most sensitive?

Sorry this is couched in general terms rather than more technical:)

Mike Brennan

John Sprung · July 27, 2007

There are two separate issues here, the dynamic range of the chip, and how we assign bits to that dynamic range. Highlights that look burned/blown out/clipped are like that because the charge wells on the chip get filled up -- and may even spill over into adjacent photosites. There's nothing we can do by assigning numbers to fix that. It's like painting the number 16 on an 8 foot ladder won't help you reach a 20 foot ceiling.

It isn't that the eye has difficulty with bright objects. The problem is, the eye does a lot better with them than the camera can. The eye sees detail over a wider range from light to dark than any camera and display combination. That's one of the fundamental problems of cinematography.

-- J.S.