DPP: Part 2 ? Digital Image Qualities: Exercise 8 – Your camera’s dynamic range
April 5, 2013
The aim of this exercise is to understand the extent of my camera’s dynamic range – the difference between the brightest and darkest tones the camera can display effectively in a single image.
The image was taken in bright sunlight, with a bright white area – and dark area in shadow. The exercise suggested using white card for the bright area, but I chose a canal side scene again, and an underbridge walkway had an element of white paint which I felt was suitable.
The image was taken at settings that avoided highlight clipping on the white point; ISO 100, f/11 and 1/160s exposure:
I measured the brightness of the white section and several darker points using spot metering mode – adjusting the camera settings so the light sensor indicated the spot would be correctly exposed. I recorded this information in a terrible sketch. I’ve transferred this to the image in the illustration below:
In Lightroom I checked the value of the white was only just below 100% on each channel, which it was – at around 99% in the R, G and B.
The next step was to zoom in to 100% and adjust the brightness/fill until details were visible in the shadows. I boosted fill to 100 (the maximum) which made detail evident right to the darkest points in the image, with no significant noise detracting from detail. Those darkest points were 4-stops below the brightest point. This indicated that the range in the image wasn’t pushing the limits of what my camera was capable of. Back to the drawing board.
The weather wasn’t on my side, but after a bit of persistence I managed to get an indoor shot that seemed to fit the bill. The setting for the below image were ISO 100, f/2.8 and 1/1600.
Again I took measures at various points (light and dark) as illustrated below:
The white measures at around 95% across each channel.
Zooming in to the dark areas shows detail isn’t really visible in the speaker when the fill light is increased:
though the silver and black branding on it is just about visible. The speaker grill measured at an exposure of 0.3seconds. One complication though is the speaker grill itself doesn’t have much texture or detail to it so it’s hard to be absolutely certain the lack of any visible detail, versus noise, is due to the dynamic range limit of my camera.
The glass on the table measured at 1/10s and appears more glass-like than noisy, so I expect the cut-off is somewhere in between.
From 0.3s to 1/1600s would be 9-stops, which is consistent with what I’ve seen forums say the 5DMk2 is capable of, and in this example I believe it’s just shy of that value – taking in to account that it would be hard to identify detail on the speaker grill anyway.