An LR view of the Stouffer wedge

Monday, 22 August 2011 00:11 administrator

Photography is not about technicalities, but understanding the physics and math behind photography makes it easier to maximally utilize the tools that we have.

Also, putting the tools that we have to best advantage may enable us to achieve results that we hope we could attain with more expensive equipment.


That said, optimization has it's limits. We can make best use of the equipment we have, but the same principles could also be applied to the "better equipment" we feel we cannot really afford.


I recently acquired a 41 step Stouffer Transmission wedge. The intention was to use it to compare dynamic range of negative film with DSLRs, but just studying how it works with Lightroom and DSLRs is quite fun, and may gleen some insights on how LR handles a subject with wide dynamic range.


Original DMG images are here: Sony Alpha 900 and Sony Alpha 55


The equipment I used is what I have. The cameras represent two generation of Sony Exmoor sensors.  The Alpha 900 is three years old and uses a sensor very much like the one in the Nikon D3X, although Nikon seems to make better use of it. The Sony Alpha 55 shares sensor design with the Nikon D7000 and the Pentax K5. I have only written about the Alpha 900 image, but may possibly fill in with info on the A55 later on.

About the wedge

The wedge I have is a 41 step transmissive target. It's Dmax is 4.1 meaning and is stepped in 0.1 density steps. We can translate this into photographics stops directly. A stop means a factor of two, pretty exactly 0.3 change in density. So our D-max of 4.1 translates into 13.7 stops.  Another way to put it is a contrast range 1:12500.


Looking at ETTR

Michael Reichmann, the man behind "Luminous Landscape", introduced the term ETTR (Expose To The Right). The idea is to achieve maximum exposure without clipping nonspecular highlights. Maximum exposure is a good thing. The problem is that we base ETTR on in camera histograms and they are a bit conservative. One of the objectives with the recent experiment was to find out how conservative?
What the wedge look like? Cannot be shown easily because of it's excessive dynamic range!

Exposing the wedge

I have chosen to base my exposure on step number 10 of the wedge. This was exposed so that "blinking highlights" indication was not shown. The idea was to find out what headroom was available. To achieve this the upper nine steps of the wedge were covered, and exposured increased until "blinking highlighst" were shown on step 10, after that exposure was reduced 1/3 step.


Evaluating the Wedge in LR

Using a "raw converter" to evaluate a raw images is not really optimal, I'm trying to develop a tool for analyzing the wedge in linear (gamma =  1) space. I did some preliminary studies using LR.
Lets look at shadow detail:

This shows the lower part of the wedge. Exposure was increased 2EV in LR to make the steps more visible. It is not easy to say which steps are clearly visible but is seem reasonable to assume that step 35 or 36 is the first really discernible step. The assumption is based on alternatives views with higher exposures and extensive fill light.


It's quite interesting to study the histogram:

What we can see here is noise and signal. Noise is the continuum (grey mass) on the left side of the diagram. Going to the right, that is with increasing exposure, we start to see peaking. The peaks correspond to individual steps on the wedge. We can see that on the left side we have much noise and little signal while on the right side we only see signal (that is peaks).

The higlight screendump below was made with -1.5 Ev exposure correction in Lightroom.

In this image we can see step 5 as clearly visible. This really indicates that we have about 1.5 stops of headroom.

There seems to be a color shift in the recovered highlight area. I'm not sure it is for real. The Stouffer wedge may not be neutral gray. In the higlights we may see the color of the light table used for exposure but on the higher density areas we may see the color of the wedge.


Dynamic range

It is hard to exactly tell the dynamic range from this test. it seems that the highest discernible field is 36 and the lowest 5. This would give us 36-5 = 31 step. The wedge has 0.1 density steps, corresponding to 1/3 stop. So the dynamic range is about 10 stops.


Imatest evaluations

Imatest is a commercial program developed by Norman Koren. The program has an option for analysing step charts. It can read raw files using dcraw, but also files converted by other tools.
The image below was converted by LR with:

This image was converted internally by "dcraw"

It is interesting to note that the image converted by Lightroom seems to have a wider dynamic range than the one converted by "dcraw". I have no explanations for this.


Finally, I also made a conversion in LR "luminance noise reduction" set to 30. This did not affect DR in Imatests but had a significant effect on usable dynamic range.

Alpha 900 Imatest evaluations



Last Updated on Thursday, 01 September 2011 05:34