Wednesday, December 24, 2014

The Great Dynamic Range Mystery

Backdrop
I have been a photographer for over thirty five years, and during that time have come up against many problems and hurdles whilst practising my profession.
Recently I met and spent a social hour with an old photographer friend who showed me his latest Nikon (a lovely camera) and during the evening the conversation touched the subject of the range of tones that modern digital cameras could capture and how this related to the films we used to use for wedding and portraits.


There seem to be a consensus that digital  can record a wider range of tones than  negative films, which seems strange as that is far from the experience of most photographers I know who still use both mediums.

My friend pointed me to a site often used to back-up this assertion:

A test of film vs digital dynamic range

The linked web page from Mr R Clark states:
"Digital cameras, like the Canon 1D Mark II, show a huge dynamic range compared to either print or slide film, at least for the films compared"

A huge dynamic range? This is not what most people I know have observed!

The evidence put forward was a series of photographs showing a black box with test charts photographed on each medium the digital showing great shadow detail and good highlights the negative film showed OK highlights and non existent shadow.
(see website linked)


My first thoughts were how very strange quite the opposite to what I have observed-what could have happened?
The clue is early on in the text, he compares a digital camera, a slide film and a colour negative film and states:

"the three data sets are registered at the bright end. This is why the curves in Figures 8 and 10 overlay each other at the high end"

Now it begins to make sense why the author only sees the film has a range of seven stops–he used the same exposure for all three mediums so he has exposed for the highlights.

Let me explain; both digital and slide film need to be exposed so you don't clip highlight information and just end up with either clear film or saturated sensor wells, so you effectively expose for the highlights
Negative film on the other hand needs exposing for the shadow info and let the highlights fall on the upper part of the curve.


Expose negative film so the shadow falls in the toe
The correct way for Mr Clark to evaluate the DR of each medium would need very different different exposures for each medium; the one for the negative should place the darkest shadow in the toe of the curve so the shadow will be perfectly reproduced and because of the compression effect of the negative films shoulder the highlights will still be recorded.

My experience (and Kodak's figures) has shown that most negative films have a range of about 12-14 stops (albeit the top of the range is compressed). Negative films have a shoulder that means the highlight roll-off is very gradual  on the other hand digital has better shadow recovery so more detail given the same exposure.

So what Mr Clark has done here is exposed to give good highlight retention on the digital and used the same exposure on the negative film; thus showcasing the shadow recovery of digital and showing up the poor shadow of film when underexposing (the exposure values for each are different)

He has either deliberately or by accident truncated the film curve (used only the left part of the curve and placed the highlight on the 'straight line' of the curve this caused his underexposure and that is why he gets only seven stops of useable data rather than the fourteen Kodak cite

You could conversely do this 'experiment' again exposing the film correctly (expose for shadows and stop down 2 stops) and then expose the digital with the same exposure values thus blowing the highlights; this would give us a very different conclusion from Mr Clark's test.

I could verify Mr Clark's results by making my own 'black box' (that might be a future post) and placing the negative exposure on the correct part of the curve for that medium.

Here though are a couple of tests that show the effect:

Film and digital compared over a wider range than Mr Clark's test

And
A test showing film vs digital over the complete range (mentions Clark in the conclusion)