In short the correct exposure is what ever captures the range of scene brightness that you are interested in onto your capture media (film or sensor or even datafile). Given this premise the rest of the article explains how to get that on your film and how to use digital camera to help you.
Basically I'm intending this article to be useful to the following groups
of people:
New Tools that you already know
Today in photography it seems that there is always something 'new' to learn
which will make your life easy. Perhaps it may, but its always hard to decide
before hand what will help and what will just waste your time before you commit
the time to learning it
That's where this article is different as what I'm talking about here is applying
what you (may probably) already know, in a way you might not have thought
of. For the last 100 or so years photographers have pursued understandings of how
photographic materials respond to light. Among the tools thought of as standard
today in commercial photography and printing, are
"what's a densitometer??" you ask, yes, I know that I just promised you don't need to learn any new tools here and you don't, but knowing they're used is a key step.
Now, probably not many people use a hand held light meter anymore,but for those who still use and prefer them (or even those who've seen and heard so much and are tempted to learn about them), I'll remind you that you can't just point it and use the number it gives without expecting some hit and miss. This is even more true with spot-meters.
To effectively use a hand held light meter you need to have some basic understanding of "what is a good exposure and why". Basically this means that you need to know what the brightest and darkest thing you can record (with your media negative / slide / digital) in any given scene. The light meter gives you an EV number for a scene or if its more sophisticated may give you an EV number and a range for the points you've measured. But if you have a digital camera, why would you bother with a light meter when your digital camera has this inside and then even displays the selected EV the picture with a histogram to allow you to decide if the exposure decision it made was what you wanted.
A digital camera is effectively also a sophisticated metering system which more or less instantly stores the results of reflectance metering of a scene as millions of point (or spot metered) samples. This gives the photographer not only a set of light intensity values, but also:
The purpose of this exploration there for is to understand how the information stored by the digital camera can be used to determine exposure for each of the major film types:
I've used the Epson 3200 scanner for gathering my data from the film, as you'll
see even though these scanners may not be ideal for their resolution when working
with smaller format films (like 35mm, thus needing more enlargement to get a print)
they are not challenged with respect to the density range that must be penetrated
with respect to negative film. I'm not sure that the same will be true for positive
or slide films.
Note 1: I've found that using the scanner to scan the
negative working as a 'positive' the, input / output controls seem to operate
more logically. I don't think there are any great advantages in working this way
but at least it affords better and simpler measurements.
Note 2: it is important to set the grammar to linear (or 1) on the
software as well, otherwise there are 2 sets of readings from the eyedropper tool
Note 3: I thought it might be usefull to link to my test of scanner linearity
which I did with this particular Epson here
This backs up that the machine is not really up to getting down into the shadows
of E-6 films, but certainly well within limits of the densest I can find
on any negatives I've used.
So setting the controls as:
| input | output |
|---|---|
| 0 | 255 |
| 0 | 255 |
| black sheet | clear sheet |
|---|---|
| |
|
You can see that the blackest parts of the sheet just about hit the edges of the scan, this means that the scanner can penetrate the densest parts of the negative. Also, you can see that there is some space at the top end of the range where there is something appearing, yet there is no image data in the cleared sheet. This essentially is what digital folks would count as noise. So really anything in this area is also unreliable as well. So then the effectively useful area of picture information from the scan lies between the very dense and the very thin. Combining this graphically makes it clearer how this works out
The significance of this is that we can say that if cut off anything below the edges of the darkest parts and cut it off above the thinnest parts of the negative (which was unexposed anyway) we will have the useful range of the film.
This is essentially what setting the black and white clipping points is about when we are scanning. By removing the areas where we know we will have trouble discerning the difference between our exposure and the film itself we can get rid of what is essentially just noise.
So, if our exposure of the film keeps the faintest bits of the negative and the blackest part of the negative within these boundaries we will have a good data set to work with!
To make my life easy, I set my camera to 100ISO because the film is 100ISO. I then developed this film according to the manufactures recommended N development. Below is the result of that:
Now, this is not a contrasty scene and as you'll see the exposure was well within the range of the camera
and so I'd expect to see that it doesn't challenge the film's ability.
Note: at this stage, development is strictly by the book, meaning that I developed
exactly according to the film makers specifications.
| Koala exposure | digital metering histogram |
|---|---|
 |
 |
As you can see in the histogram above (beside the image) there is nothing in there which pushed the limits of the image, and the spectral highlights in the Kola's eyes are still not blown (although its close). This is what digital folk would call good "scene referred data". From this, because there is no clipping (and in fact its very close to but not clipped) application of curves and correctly setting black and white points would yield as good a print as you could get from this scene with any camera.
Now, lets look at the results from the scan of the film, now remember that this a negative (meaning that white areas expose the film and appear black) and I'm scanning the negative in positive mode, which keeps it just as film looks. So the histogram must be 'flipped' for us to see it in the same 'Spence' as the histogram of the digital camera (which is not a negative).
 |
 |
Very close aren't they! Below I've flipped the histograms of both to show you that there is no trick here
| |
 |
| |
|
There seems to be more room for brighter exposure on both. As you can see, as the negative gets blacker as it gets more light, perhaps things could have been pushed much brighter in the film exposure as there is some space left to the left there (in the black area of the negative). Lets look at how the histogram of the film image of our Koala fit within the range we had above for the film.
Great, the detail I wanted from the darker parts of the picture (the shadows) fit neatly in and there is definitely more room there on the left and so we could squeeze more exposure into the film too! This would give us the ability to would bring up the amount of detail in the shadows (and perhapsgive us a greater tonal range or simply handle greater spectral highlights than existed in this image.
Either way, its clear that there is a strong relationship between the recorded exposure of the digital camera (as shown in the histogram) and the exposure and resulting scan histogram found in the film.
A useful finding, but upon reflection should be hardly surprising because we have come to digital from using film. So the cameras surely would have been designed to fit within the same sorts of limits to be useful as tools to existing photographers.
So, where to from here?
Next I thought I'd explore what might be close to HDR. Certainly I've found that digital cameras seem to be able to recover amazing amounts from the shadows, but what happens when we push the spectral highlights and tonal range further up into the white when we already have an extreme situation? Accordingly I took an image of the a high contrast, a lamp illuminating the room.
In this image I pushed the exposure of the film up a little higher (2 EV) than the digital recommendation to see how well it would be able to get the background and also to try to blow out the exposure on the brightest thing.
Now, when you look at the histograms in the top row, remember that unlike (say) a graph of exposure, clipping is not reflected by anything in the X axis (the hight of the graph). Recall that we're looking at is a count (on the X axis) with the "step" from black to white (left to right). You can see that my negative does not have anything in the very black area (I clipped off the 'noise' in this area but left the remaining graph intact) thus keeping away from the floor noise in the base fog of the film . (remember this is alreay inverted from the negative, and what we see as white here is black on the film).
Despite the initial difference in shaps, there is some similarity in the peaks even though they occur at different locations. Note that the digital graph is histogram is obscured a little by the different locations of colours (it being RGB) while the black and white is a monochrome just panchromatic rendering.
| Raw Film scan clipping black and white points off | 10D RAW conversion (2EV lower than film) |
|---|---|
 |
 |
 |
 |
The difference between these images to me is that I only need to adjust black points to the make the shadows darker in the negative source image, while in the digital I may need to bring them up (assuming I want them) but the highlights are lost very easilly.
Significantly, the negative continues to hold more detail even in the face of the high brightness around the bulb. Texture remains and there is no ugly blow outs. Note: the depth of field differences between the 180mm lens used on the LF camera and the 50mm lens uses on the 10D mean that the lampshade behind the "tassels" is rendered as a blur on the 4x5 while its just blurry on the 10D.
So, I didn't use the same exposure as the camera obtained. This is where understanding your medium comes in. I know that digital is linear in its reaction to light (like a good light meter should be) but this means that I'll hit hard and unforgiving limits on highlights (unlike film). So to get something usable from the digital I actually exposure compensated the image to adjust for the limit on the highlights (and I also know that I can bring shadows up more effectively in the digital than I can with a film). This might move the mid tones up or down a little, but in this sort of contrasty image who's sure of exactly where the mids are exactly anyway?
If I had kept my exposure to be exactly the same (not going up 2 EV) then the shadow details would be darker on the negative (meaning that as they would be thin on the negative I could not get much usable from them) and perhaps the mid tones of lamp would have been in a slightly different location. All very easy to shift and print easilly. Rember, this is capture, not making a final print. Our goal here is making a good (read printable) exposure onto film.
So in this high contrast example (black and white negative) I think there is still a very high degree of transferability between digital camera exposure values and film exposure results. Certainly any differences in media response can be accomodated with slight curves alteration in a photo editor or in traditional printing with paper exposure and contrast controls there. To illustrate, I've just altered the gamma of each of these (albeit in different and opposite ways) to show how close they come to each other. Eg:
| ADOX | Digital |
|---|---|
 |
 |
The key to remember here is that while the responce of the digital sensor to light is linear (and the films is not) your light meter should also be linear. Its the readings you'd take from your light meter to then figure out what you need to set a final aperture and shutter to get a negative you know will be usable. Using the digital you can see where your printable shadows are, and how much highlight you're going to get and the relationship between them. Isn't that what you want to know with your spotmeter?
Did you ever rough sketch your image and write in the EV's of the spotmeter as part of your procedure? By using your digicam you get a visual rendering of that relationship and a final single EV that you used for the exposure.
Just because the image on the digicam screen does not match your final print desire does not invalidate the information.
So, what's the takeout? For me its use the digital cameras preview to look for something being present in the shadow's and looking at the historgram see what blows out on the digital and keep in mind you'll have more lee way in the film exposure. Just as you wouldn't use a negative directly (unless you know what your looking at) to judge what a print will look like, don't use the display on your digicam to determine what a print will look like either. Where as in digital you'd abide by the 'expose to the right' principle, when your using it to evaluate light readings you're not so worried about clipping (cos negative film won't clip as easilly if at all) and interested in adjusting more for the shadows (remember the 'expose for the shadows' adage). I over exposed this negative by two stops (compared to my digital) and we found:
The image below is a montage of the digital (JPG) image and the results of the 3 exposures scanned together on my flatbed scanner (all in the same pass, so no individual optimisation yet).
Straight away its pretty darn clear that the exposure blew the clouds out on the digital, but even on the 2 stops over exposed image there remains some texture in the clouds. The colours of each are quite different, but then that can be adjusted for both.
C-41 negative film type has an orange mask, this will result in altering the basic contrasts of the negative. It makes the clearest part of the negative darker than on the ADOX. This means that the least exposed part of the negative will be at a darker overall 'density' than with the ADOX. Here is the result of a scan of the negative. Again I scanned this as "positive" because the densitometer reading tool operates better in this mode.
| scanner range (from above) | C-41 scan |
|---|---|
| |
 |
| Looking again at the ADOX film scan above there is a much greater spread of the scene brightness across the scanner captured data. This means we'll need to strech it to make it "fit" without our desired intention. To avoid getting problems with posterisation we need 16 bit scans. | |
This shows that also the densest part of all of the negatives being scanned here is very close to the limits of the scanner. This means we'll expect some noise. IMPORTANTLY, because this is a negative this noise will turn out to be in the light areas (not the shadow areas as with scanning slides).
So, lets look at how much we have in each of the Red Green and Blue parts
Its not much is it. Just like a bayer array There is not equal amounts of signal for each of the colours.