We aren't saying it can't be done, just that there are so many areas where errors could occur that it doesn't seem worth the effort. Especially for the individual just testing the odd camera. Digital camera information websites that regularly test new cameras as part of their remit are in a slightly different position, and the setting up of a test like this might prove worthwhile, if accuracy in all areas can be guaranteed and maintained, however they are analyzed, by sight or analytical software

But the fact that one digital camera review website using this testing method and the analytical software to analyze the results then queries them as not really seeming to correspond with what it thought about the cameras performance in real tests, seems to confirm our views.

Undertaking the Grey card test

This is a very simple test that is accomplished by under and over exposing in steps from a base exposure value. By doing this you are testing the camera's exposure latitude if you like. Seeing how many steps it takes until all tones are exposed as white, and how many steps until all tones become black. You can then sample the images that result, take the tonal values into a spreadsheet and generate the results on a graph. Or compile a visible tonal step wedge from them and count how many visible steps exist between the two extremes, the workable, practical dynamic range. We normally do both, and view the tonal step wedge as the best indication. Compiling a tonal graph is okay for seeing what the tonal curve is, but isn't really any good for deciding what the visible DR range is.

Making a grey card

The first requirement for this test is a grey card. All a grey card is is a piece of card that is mid-grey in colour which you can used to take an 'average' reflected light reading with which to set your camera metering. They are nothing special in these days of the inkjet printer as they can easily be printed out on one. So if you don't have one then this is the first thing to do. It needs to be printed on matt paper, not gloss, and the heavier weight paper used the better. It means it will be stiffer and stay flat better and the tonal value obtained won't vary too much across the image frame. And trust us, it will. 'Proper' grey cards are matt, and either 10"x 8" or A4 which is the size you need. The colour should be around mid grey, and if you can manage the correct tone so much the better. It should be 128/255 on the 0-255 8bit tonal range. To get this tone all you really need to do is pick the 50% grey out of the default colour swatches in Photoshop or the image editor you prefer to use. Set this as the colour for your A4 size image file, at a resolution of 300dpi, and then just print it.

You may read that a grey card has a reflectance of 18%, indeed it is often referred to as a 18% grey card. Don't let this confuse you. This is quite correct and is all to do with the human perception of light which is logarithmic. Pure black is the absence of any reflected light , 0%, and pure white the result of the reflectance of all light, 100% - in theory if not in practice - but as we see light levels on a logarithmic value, the tone mid-way between the two, whilst it may be mid-grey to us, only has a reflectance of 18%. 

Setting it up

The next step is now to photograph the grey card. All this needs is just a little bit of care and commonsense. First it needs to be done somewhere the light is even, there's no shadows or highlights across the grey card, and the light level is not going to change while you take the shots, so the tonal value remains fairly even and constant. It's also easiest if done during daylight hours to overcome/reduce colour cast problems, and to make the Ev values reasonably high across the range, as those needed to produce a pure black result will be quite long anyway. So, indoors during daylight hours, and away from direct window light unless it's even and constant is what is required.

To make life easier it's best to set up the camera being tested on a tripod. Framing of the grey card will then be consistent, and you'll probably need one for the longer shutter speeds anyway. So if you can set the grey card up somewhere in the vertical or near vertical position then so much the better.

The camera being tested must be set in the manual exposure mode, and the lens used - (and it doesn't really matter what the type or focal length, although a longer focal length, say around 50mm in 35mm terms is handy) - must be set so that the grey card fills the frame and is square on to the camera/lens. Manual focus should be set because otherwise all the autofocus will do is hunt around as it's unlikely to find a point on the card to lock on to. In any case focus doesn't matter here, it's not being tested, and roughly focusing the lens is all you need to do. 

Now set the camera so that the Ev value is in 0.3Ev steps. If your camera won't allow you to do this and 0.5Ev steps are all that are used don't worry, it's not the end of the world, it's just that any results you get will be less accurate. The difference between the tonal steps captured are just bigger. It doesn't figure in the middle of the tonal range but it will at the ends, which is what the test is all about, in that the move from visible steps to those that are not discernable will be more severe and the accuracy of the apparent practical dynamic range less so than when 0.3Ev steps are used. 

The widest dynamic range a camera can deliver is at it's lowest ISO value. There is a general trend that is common to both film and digital use that dynamic range reduces as ISO speed rises. So set the ISO rating of the camera to it's lowest or default setting. Don't set auto ISO, that would be no good at all.

Now we come to white balance. You can leave this to auto if you like. If the daylight levels are good and your camera has a good white balance performance this is okay. But to reduce the effect of colour casts, which don't alter the overall dynamic range but can give colour sampling problems later when compiling the spreadsheet graph, the best course of action is to set a custom white balance. If you can't do this, again don't worry, we will deal with the results that might arise during the tonal step wedge image compilation, but if you can, so much the better. Many will tell you just to set this using the grey card, we often see this advice. You can if you like, but all this will do is set the colour saturation to a higher level than normal, because your not doing a white balance, but a grey one, and this will set the tonal range off when you meter the card before you start taking the shots. So use a white reference point, a white sheet of paper. Even if you do this strange hues may still result with exposures that are far removed from the correctly metered base value. For example our *ist-D produces images with a marked blue cast when the shutter speed gets very long, whatever the white balance setting, auto or custom. 

Now we come to the issue of file type settings. Just use best quality Jpeg at the maximum resolution of your camera. You can use Raw if you want. But it won't make any difference to the end result. Many say it does, that dynamic range results using Raw indicate a wider range. But this is not what we have found when doing grey card tests. All Raw file capture means is another layer of processing before image sampling can be carried out.

Now we come to the last setting, the base exposure. To do this set manual exposure, choose a fairly large aperture, and change the shutter speed value until the camera's metering tells you that the correct combination has been reached. With this set, and if it seems slow open up the aperture until it's a better value, you are ready - finally!

As far as any other camera settings go we suggest you use whatever you normally would. Because what your basically trying to find out is what it produces under normal conditions. This is particularly so as far as brightness contrast and saturation level settings go, so use the default values. A lower contrast setting may well result in a slight increase in DR, and a higher level a reduced one. If you want to try different settings later to see if it makes a difference that's fine. After all you can do as many tests as you like. Or as many as you have the time and patience for. And use the colour profile you normally do. Some cameras have the choice between sRGB and Adobe sRGB or other options. Our advise is just to use the standard sRGB profile.

Taking the grey card shots

The task now is a fairly simple one. Keeping the aperture value constant, take a series of shots both above and below the shutter speed 'base' value. You should be able to do this simply by 'flicking' the shutter dial one step at a time. This will shift the value in those 0.3/0.5Ev steps you set the camera up to move in.

First take one at the correct 'base' value, then a series going up the scale to white, longer exposures. Then go back to the base value, and do the same in reverse, a series going down the scale to black, shorter exposures. Don't rely on looking at the images on the camera's LCD screen to decide when you've taken enough. When the images have either gone pure white or pure black. As most will know, they aren't that accurate unfortunately. So take 15 step shots from the base value up to white. And then 20 shots below base value to black. This is if you are using 0.3Ev steps. You can if you like go as far as your shutter speed range exists, both up and down the scale, there is nothing to stop you. But it will normally be a waste of time. 35 shots in 0.3Ev steps gives a possible DR of 11.6Ev. You won't need more than this, despite what figures you might have read elsewhere, electronic sensors just can't deliver this kind of range.

When you have done this transfer the images files taken to your computer for the next stage.

Generating a grey card step wedge scale image

With the shots you have taken saved to a folder, you can open up your image editor and begin the compilation of the tonal step wedge image. As all you want is a tonal scale representation just assembling all the images together isn't really the best way to go about it for a number of reasons,  the file size would just be unmanageable for one, but mainly because of the tonal shift across the frame.

What you have to appreciate is that light levels entering a lens are not constant. The brightest values are always at the centre, and reduce towards the edge. This is why corner shading occurs. This is always at it's worst when a lens is used with it's aperture at the maximum size and the difference reduces as the aperture is closed down. As a result tonal values towards the outer edge of an image can be somewhat different from those at the centre. And this is before any tonal change as a result of differing light levels hitting different parts of the grey card. Even if it is small, and not really noticeable by sight, sampling will reveal different values.

So begin by opening a new file, 7000 pixels wide x 400 deep, at 300dpi and set the mode as greyscale. Doing this at this stage will eliminate any colour casts that have arisen in the shots. Go to the rectangular selection tool and set a fixed size 200 pixels wide and 400 deep. This is for a folder with 35 shots. 15 to white - including the base shot -  and 20 to black. 

Now go to the image file folder. In thumbnail view you should see that you have a range of images. The first you took should be the base mid-grey exposure, the 'correct' one. Following this should be a series of 14 that get lighter until they become just pure white, followed by a series of 20 that get darker until they become pure black. If you took more than 35 shots, say you ran up and down the shutter speed range, then you need to cut them down, because those at both ends of the scale will contain no useful information.

When you have got to the stage that you have 35 shots you can begin. You may well find that there are several at both ends that are either pure white or pure black. Don't discard these, leave the folder at 35 images.

Find the last shot taken in the series that was under-exposed, the first that records pure black, number 20 from the base exposure. This is the shot to start with. You open and sample this, and then all the under-exposed one's in turn, backwards, until the least under-exposed is reached. Then the base image, and following this all those until the first pure white one is reached. So you sample all the images taken in turn, from the very darkest, black, to the very lightest, white. You may well find that when you start sampling the variation in readings between the black frames doesn't change much for a while. You might even find some that have the same readings. Sample and record them all. We'll explain the reason why this has happened later on when we come to interpreting the results.

Here is a folder view of one of the sets we took. We've indicated where to start, where the yellow spot is, the last of the -Ev step images and how to progress. The red spot is the base image and the start of the + Ev step images. In order to show the progression we has to reduce the number of files in the folder, otherwise they didn't all show on screen together. There is just 29 here. 

Okay, open the first image and with the colour picker/sampler take a reading from the centre of the image. This will set the foreground colour. It's important that you do this for all images. Don't vary it. If you run the picker over the image, left to right, up and down you will probably see that the reading values alter. Oh sorry, forgot to mention. If your 'info' tab is nested in the toolbar well, Photoshop/Elements etc, drag it out so it's visible at all times and you can see the readings. These will be in R,G,B values, 0-255. On the image step wedge you are creating the readings will be different because it's not in RGB mode but greyscale. Ignore this. The values will be correct later when we do another change. In greyscale mode all that is shown is a single figure in percentage terms, 0%-100%. If the RGB values are not the same when you sample, and this will happen if there is a colour cast, then don't worry. It can be a problem when you come to transfer them to a spreadsheet, you don't know which figure to use, so that's why we generate the wedge image in greyscale. 

Sample the image and then close it. Now with the selection tool click on the greyscale image and a selection box the size you have specified will appear. Drag this to the left to the end of the blank image. Change to the paint bucket and use this to set the colour for the selection box. This colour will be that which you sampled from the first image. Now do this for every image in turn in the folder that you took. When you select another selection box and move it to the left, move it so it just overlaps the previous one. If you don't do this you will end up with a series of tonal values with white lines between, even if you think you've butt joined them.

When you have sampled all the images a tonal step wedge image will result. Before you do anything else, go to the image mode menu and change it from greyscale to RGB. Then save the file. Changing the file back to RGB at this stage now gives constant RGB values without variation due to colour shift. Using RGB values gives a more accurate reflection of the tonal steps, and is better when you take the values into a spreadsheet and make a chart.

Eventually you will end up with an image like the one below.  This is for the Canon 350D. Now the task is to examine it to determine the result. 

* * Whatever you do please don't alter the tonal step wedge files tonal values in any way. We have seen descriptions where the file is altered by using levels or a highlight/shadow tool to bring tonal values into view. This defeats the whole object of the exercise, which is to see what tonal values the camera can produce at the settings used. Making alterations such as these is another reason why some dynamic range results are so inconsistent and inaccurate and aren't a true reflection of the abilities of the cameras sensor, or the processing involved.

Interpreting the tonal step wedge result

We suggest you do this with the image magnification set at 100%. What is called actual pixels. First it's handy to mark the tonal step that was the base exposure value. So re-open this file and check the reading with the colour picker tool, find this on the wedge and mark it. We use a red star. Now shift along to one end or the other to find where the tonal steps end, where they disappear and you can't make them out anymore. So that we can see them when the file is at smaller magnifications, as when we are counting the number of steps from one end of the range to the other, we drawn a green line where this occurs. You might find that you have to move the image sideways one way and then the other a few times so your eyes get used to the tones and you can see where the cut off step happens. We have generally found that tonal readings above 250 are not visible, nor tonal readings below about 10. They are often there, but not as visible steps. We'll return to this. With both ends of the visible tonal range marked you can now count the number of steps in between. Remember to divide the count either by 3 if using 0.3Ev steps or 2 if using 0.5Ev steps. It's something easily overlooked.

Here is the tonal step wedge for the Canon 350D marked as above. As you can see, if you look closely, the result suggests that it has a useable/practicable visible DR of about 7.0Ev. 19 0.3Ev visible steps + black and white. So 21 all told. The green lines are where in our opinion, the visible tonal step difference ends. So a step each of white and black at the ends past these marks. These are still tonal steps and valid to be included.

 

 

Canon 350D DR tonal step wedge.  visible range 7.0Ev

 

Although the chart past the green line to the left looks black it isn't pure black but has been recorded as tonal values. Having said that they are to all intents and purposes not visible. It is these tones that some DR tests record as part of the DR range. Our opinion is that this is wrong. These tonal values go down, eventually, to either 0, or a figure slightly above. Where we have drawn the line the tonal value drops from 16 to 11 and is at the juncture of a tonal step. The same can be said of the white end. Where we have drawn the line the tonal value becomes invisible. Above 250.

There is another factor to be born in mind in conjunction with this. Tonal print values. As a general rule ink printed images cannot depict tonal values above 250, or below about 10. This is because paper is not, as we have said before, really pure white, and even when black ink is used black is never pure black, but more a very, very dark grey. Around 10 on the scale. It just looks black to our eyes, just as the 250 tonal level looks white. These are the practical limits, and the two go hand in hand. So in our minds setting the tonal range limits to a maximum of 10-250 - if they can be observed - is the correct method to use. Especially as that seems to correspond with what we can observe with our eyes.

Here is a combined tonal step result comparing the Canon 350D and the Pentax *ist-D. The result suggests that the 350D has about 0.3Ev more range than the *ist-D, which is the rough conclusion we had come to before we undertook these tests through using the two cameras in real life picture taking for some time. We had thought it was somewhere around the 0.3Ev - 0.5Ev region. 

This comparison also raises some interesting points. The first thing to note though, is that the ISO speeds used are different, the lowest in each case, and this should be born in mind. However It leads us to the conclusion that in the main, carrying out a grey card DR test, in the manner we have detailed, would appear to be a reasonably accurate way of getting a rough idea of the DR range of a digital camera.

Looking at the two results you can see that the advantage the 350D has is all in the highlight end, which is where it is most effective as it is here the problems with digital capture usually lie. Another point of note is the different base exposures. The *ist-D's being darker. We have aligned the step wedges so that they are in step with tonal values. Some of this may be ISO speed difference. It is probably also as a result of the processing algorithms the camera has been given. But also perhaps the way the metering has been aligned with the needs of the sensor output. The *ist-D has a habit of losing highlight detail and we now leave  - 0.5Ev compensation dialed in for most of the time. This brings it's highlight range virtually in line with the 350D's. The 350's last visible tone in the highlight range is 246. The next is 251. By contrast the *ist-D's is 231.The next is 252. In the tests we have carried out, a fair number, the *ist-D is not always this bad and another tonal step, around 245 is often present. But it always remains a 0.3Ev step behind the 350D.

Now lets look at that large area of dark steps. At the extreme left the Pentax's does go down to 0, but the Canon's never does, remaining at 1. This would indicate that this is generated solely through noise. The Pentax has noise reduction enabled which kicks in at longer exposures so this is obviously removing the noise from the general tonal values, well some of it. There are also many steps in each that do not alter much in value and this is an indication that the exposures are too short to capture any useful amount of light and thus tonal value.

If we were just to sample the images, take the readings and put them on a spreadsheet, and count the tonal steps between 0 and 255 that the cameras have recorded, visible or not, we could easily say that their DR ranges are much wider than those we have indicated above.

So lets do that.

A Spreadsheet DR tonal chart

The spreadsheet application we use is Microsoft Excel. There are others just as good in many ways, Open Office etc. All you need is one that can generate charts from the inserted data.

Here is the chart that results from the tonal step wedge images 

 

 

This is quite revealing in it's own way, although this is mainly because of the information already gleaned from looking at the tonal step wedge images themselves and the further inferences that can be drawn. It's quite obvious that the processing written for the 350D is far superior to that of the *ist-D's. It gives a consistent line with even exposure steps. By contrast the *ist-D's is erratic and all over the place at times. Both have a quite gentle tonal curve at the dark end, but then the differences begin to tell. And at the highlight end the *ist-D just hits the buffers, almost head on, terrible. Whilst the 350D has a much more refined curve, making the most out of what it can of highlight detail.  What is known as a more gentle tonal curve roll off. Perhaps not as good as film, but getting there. What is needed is a tonal curve that is a mirror image of the dark end in reverse. If this were ever to happen then digital cameras and their sensors would not just be the equal of film, but much, much better, and the DR's would then be very wide. 

We have indicated by yellow dots where the in our opinion, based on the wedge step images the visible DR range ends. How you would ascertain this just by looking at the chart we have no idea. And if you just go on noise levels well, you could claim all sorts of ranges exist, which is what it seems some are doing. There are those that use 2 as this figure. Which would add at least another 3Ev onto the 350D results and 2.3Ev onto the *ist-D's. So then the 350D would have a range of 10Ev, and the *ist-D 9.6Ev. But results such as these are just pipe dreams at present and not reality no matter how nice it would be if this were actually the case. And won't alter at all the rate at which highlight detail is lost anyway.

The examples we have used here are just some of the many grey card tests we have carried out using these two cameras. Just to make sure the tests returned consistent figures. Those used are a good reflection of the consistent results achieved. Which have a possible error factor of around +/- 0.3 Ev. The 350D was between 6.6Ev and 7Ev and the *ist-D between 6.3Ev and 6.6Ev. So we have used the best results.

So there you have it. 

Go test - if you have a mind to - but remember, finding out what your cameras DR range is isn't going to change how it works, or the levels of DR it returns one iota. All it might do is reveal why it gives you problems in certain lighting conditions, although the answer to the problems won't change either.  

For this you should visit our page on  Image Exposure