How to determine the dynamic range of a camera - my experiment

[pgriz]

I decided to test my T1i to see what its dynamic range is in actual fact, and whether my meter reading is accurate.

The testing protocol was to shoot a 18% grey card (fully occupying the field of view), varying the shutter speed by one stop for each exposure. The card was successively over-exposed and under-exposed until the grey point was either all white (255) or all black (0). The RAW frames were imported into DPP, and the values of the center point were verified against the camera reading. To make sure that the light was constant, each sequence was terminated by a shot replicating the starting point for the sequence, and the light values were compared (they were constant).

Two sequences were done, at ISO 100 and ISO 800, to determine if the dynamic range was changed by the change in ISO. To ensure that all the colour channels were reading the same values, a custom white-balance was done on the grey card at the start of each sequence.

I draw several conclusions from my experiment. When my meter is centered, the output value of the grey card is 122 (should be 127 if exactly centered, but close enough).

At ISO 100, 3-stops overexposure gives me a value of 251, and 4-stop overexposure give me a value of 255. So my upper limit is 3-stops overexposure. At ISO 800, 3-stops overexposure gives me a value of 250, and 4-stop overexposure gives me a value of 255. Therefore it appears that raising the ISO to 800 does not reduce my upper dynamic range.

At ISO 100, my readings for the center point value is as follows: -3(20),-4 (10), -5(5), -6(2), -7 (0). For all practical purposes, my lower limit for underexposure appears to be 4 stops, because a value of 10 is pretty close to black on most monitors. Repeating the exercise at ISO 800, the readings are: -3(19), -4(10), -5(4), -6(2), -7(1). Raising the ISO did not diminish the dynamic range at the lower end.

Practically, this test means that if I am practicing the “exposure to the right” technique, I must be sure that the brightest detail is no more than 3 stops above the “0” reading to ensure there’s useful detail.

I’m curious if anyone else tested their camera to determine the actual dynamic range? If so, what did the measurements reveal?

 

[tirediron]

Neat!

 

[Josh66]

Interesting...

I wonder what RGB value each zone would have in a test like that.

If Zone V is a properly exposed grey card, that should be 128,128,128 (right?). Zone IV would be about 102,102,102; Zone III would be roughly 76,76,76...

I think I will test my camera tomorrow to see how it falls out on the zone system... See if one stop is one zone...

 

[pgriz]

Interesting...

I wonder what RGB value each zone would have in a test like that.

If Zone V is a properly exposed grey card, that should be 128,128,128 (right?). Zone IV would be about 102,102,102; Zone III would be roughly 76,76,76...

I think I will test my camera tomorrow to see how it falls out on the zone system... See if one stop is one zone...

That was one of the reasons I tested for the dynamic range - whether a 10-zone system would fit within the dynamic range of my camera (in terms of 1 stop = 1 zone). Unfortunately, it doesn't, at least with my camera. However, knowing that my upper range is 3 stops, I can spot-meter the highlights and adjust my exposure accordingly.

As for your question about the different values in the RBG space, that was why I used a custom color balance, to ensure that the values were the same for each channel.

I'm going to use the data from this experiment, to develop a "characteristic curve" as described in "Light, Science and Magic, 4th edition". It is obvious from the data, that the graph is not linear, but has a short shoulder on the upper end, and a gradual shoulder on the lower end.

 

[pgriz]

The data determined by my experiment on the dynamic range of the camera allowed me to put together a characteristics curve, shown below:



As discussed in Light, Science and Magic, 4th edition, pg. 244-253, an ideal characteristics curve would be linear, with each change in aperture having a linear increase or decrease in the density of the grey. Actual sensor performance is different. I therefore was interested in how my camera would behave, and what latitude I had in terms of the dynamic range. The above graph shows the behaviour of my camera in transforming an 18% grey card to values on the resulting image.

I welcome comments from more experienced photographers as to the interpretation of the curve as seen above.

 

[2WheelPhoto]

Nice

I'm non-scientific, give me 2 cams same time of day with same lens and let me see the difference in post

 

[480sparky]

Maybe I'm missing something, but isn't dynamic range defined as the difference in the amount of light in a scene from the brightest to the darkest?

And isn't an 18% gray card ALL the same brightness?

I think in order to test the DR of a camera, you'd need subjects of different brightnesses. I'd guess your results would be exactly the same even if you used a black card or a white card.

 

[jake337]

Maybe I'm missing something, but isn't dynamic range defined as the difference in the amount of light in a scene from the brightest to the darkest?

And isn't an 18% gray card ALL the same brightness?

I think in order to test the DR of a camera, you'd need subjects of different brightnesses. I'd guess your results would be exactly the same even if you used a black card or a white card.

Maybe three cards next to each other. Black, grey, white. Expose for grey and using additional light on the white untill you lose detail and subtractive lighting on the black till it is pure black and measure it how far your camera will go in either direction while keeping the exposure on grey constant.


Lol, I have no idea.

 

[480sparky]

Maybe three cards next to each other. Black, grey, white. Expose for grey and using additional light on the white untill you lose detail and subtractice lighting on the black till it is pure black and measure it somehow.


Lol, I have no idea.

I would think you would need a set of cards, very much calibrated, each one reflecting twice the amount of light as the next darker on, in order to make such a test valid.

Let's say you could find a set of 20 such cards. Line 'em up and take a photo. Let's say 3 of the darkest ones come out totally black, and 3 of the lightest totally white. Then I would accept a result of 14-stop DR.

 

[pgriz]

Maybe I'm missing something, but isn't dynamic range defined as the difference in the amount of light in a scene from the brightest to the darkest?

And isn't an 18% gray card ALL the same brightness?

I think in order to test the DR of a camera, you'd need subjects of different brightnesses. I'd guess your results would be exactly the same even if you used a black card or a white card.

Well, there are two ways that "dynamic range" can be understood. The first, which you allude to, is the range of brightness in a scene. The second, which is what I was testing, was the ability of the camera to record that range of brightnesses in one exposure. You could, as you're suggesting, get a series of cards with different shades of grey on them, so that each shade reflects exactly 50% of the card above it in brightness, or you can simulate the same thing by using the exposure, since each exposure difference of 1 stop reduces the amount of light by 50%, which is what I have done.

The point of the exercise for me is to understand how to use the information my incident and reflected light meter readings are telling me about the scene so that I get the maximum amount of useful detail in my images. Let's say my incident light meter reading is 1/100 at f/16 at ISO 100 (the sunny-16 rule), and a reflected reading off a highlight (say the top of a snowman) gives me a reading of 1/800sec at f/16 at ISO 100 (ie, 3 stops above the ambient). Will an exposure set to the ambient light exposure cause the highlight to blow out or not? According to my test, the highlight will just within the boundary of being blown out. If the reflected highlight reading was 1/1600 sec, then definitely that highlight will be blown out. In this case, to preserve the highlight detail, I would have to under-expose my shot by 1 stop.

 

[480sparky]

..........The point of the exercise for me is to understand how to use the information my incident and reflected light meter readings are telling me about the scene so that I get the maximum amount of useful detail in my images. Let's say my incident light meter reading is 1/100 at f/16 at ISO 100 (the sunny-16 rule), and a reflected reading off a highlight (say the top of a snowman) gives me a reading of 1/800sec at f/16 at ISO 100 (ie, 3 stops above the ambient). Will an exposure set to the ambient light exposure cause the highlight to blow out or not? According to my test, the highlight will just within the boundary of being blown out. If the reflected highlight reading was 1/1600 sec, then definitely that highlight will be blown out. In this case, to preserve the highlight detail, I would have to under-expose my shot by 1 stop.

Couldn't you just turn on the 'Blinkie' function of your camera to show the blow-out highlights? That, or zoom in on the snowmans' head while watching the histogram?

 

[Helen B]

I suspect that you could do it with two or three cards, if you didn't accept the accuracy obtained by simply changing the exposure via shutter speed. Say you had one that was 18% grey and one that was 36% grey. You could use those to check successive exposures when changing by one stop of shutter speed.

In practice you couldn't get 20 stops by reflection, so transmission is normally used - a Stouffer step wedge or, if you can afford it, a DSC Xyla.

I'd be inclined to look at the raw data rather than JPEG, using RawDigger, for example.

 

[Helen B]

Maybe I'm missing something, but isn't dynamic range defined as the difference in the amount of light in a scene from the brightest to the darkest?

And isn't an 18% gray card ALL the same brightness?

I think in order to test the DR of a camera, you'd need subjects of different brightnesses. I'd guess your results would be exactly the same even if you used a black card or a white card.

Well, there are two ways that "dynamic range" can be understood. The first, which you allude to, is the range of brightness in a scene.

Isn't that better called 'scene brightness range' or 'subject brightness range' - both of which have the same initials, conveniently.

 

[pgriz]

..........The point of the exercise for me is to understand how to use the information my incident and reflected light meter readings are telling me about the scene so that I get the maximum amount of useful detail in my images. Let's say my incident light meter reading is 1/100 at f/16 at ISO 100 (the sunny-16 rule), and a reflected reading off a highlight (say the top of a snowman) gives me a reading of 1/800sec at f/16 at ISO 100 (ie, 3 stops above the ambient). Will an exposure set to the ambient light exposure cause the highlight to blow out or not? According to my test, the highlight will just within the boundary of being blown out. If the reflected highlight reading was 1/1600 sec, then definitely that highlight will be blown out. In this case, to preserve the highlight detail, I would have to under-expose my shot by 1 stop.

Couldn't you just turn on the 'Blinkie' function of your camera to show the blow-out highlights? That, or zoom in on the snowmans' head while watching the histogram?

Yes, that would be one way. But I don't always use live-view in doing the shots, so knowing where the clipping "would" happen is still useful info.

Maybe I'm missing something, but isn't dynamic range defined as the difference in the amount of light in a scene from the brightest to the darkest?

And isn't an 18% gray card ALL the same brightness?

I think in order to test the DR of a camera, you'd need subjects of different brightnesses. I'd guess your results would be exactly the same even if you used a black card or a white card.

Well, there are two ways that "dynamic range" can be understood. The first, which you allude to, is the range of brightness in a scene.

Isn't that better called 'scene brightness range' or 'subject brightness range' - both of which have the same initials, conveniently.

I agree that "scene brightness range" would be a better name, but I've seen "dynamic range" used often when describing exactly that, so I assumed this was common usage.

 

[runnah]

You lost me at math.