|This page contains notes about my experiences with my Kodak DC290 Zoom digital camera, as they relate to exposure, zones, and latitude. I'm putting this here not just to organize my thoughts, but so that others may benifit. Some will be specific to this camera, or similar models, but others will be general.|
Back to Digital Photography and Scanning page.
Outdoors in bright sunlight, I found that I had no detail in both the shadows and the highlights.
Upon further inspection, using the eyedropper tool in PhotoShop, I found that the white areas were indeed saturated (solid 255's everywhere), but the black areas showed some variation. That means I can selectivly lighten the black areas and bring out some color and detail. In the red squares I've "dodged" more brightness in, and the figure went from a black shiloette to showing pink skin and purple cloth. The tree on the left also shows some more detail.
So these apparently too-high-contrast pictures are not as bad as I originally feared. However, I still have the question about contrast control. I can adjust the exposure up or down, but I don't beleive there is any way to adjust the dynamic range lower or higher. Negatives have a lot more range, and I can control this when making a print by choosing the paper (or using variable-contrast paper and changing my filter, if doing B&W). With an affordable digicam, this is a handicap I have to deal with.
Lesson: watch your "latitude". Expose for the highlights, and add fill light for the shadows.
I can find out just what the range is using a light meter, in the near future. Then I can do the same in the field and not have to check the results in PhotoShop to see if I'm not exceeding my gamut.
OK, so how bright, and how dark, can something be and still appear in the picture? In an effort to find out, I shot some gray cards.
The first picture was exposed correctly. The position of the light meter indicates F8, and that is what I shot it at. The second picture was under exposed by two stops, and the last picture was over exposed by two stops. Also note the position of the light meter in the bright photo: this shady position read F2.8, or 3 stops darker than the position indicated in the first photo.
Various positions in the photos are center (correct exposure on the gray card), -1, -2, -3, +1, and +2 stops different from this. The measured pixel values were consistant among different ways of generating the same image density: putting an object in the shade (light meter reads darker), or under-exposing the same number of stops.
Interestingly (and unsurprizingly), the medium value is about 50% gray pixel value. Darker values are higher, lighter values are lower. It appears to be roughly linear between -3 and +2, as my sketch shows. (Well, maybe not quite up to -3.) The larger vertical size of some of the spots indicates measurment uncertainty.
Another interesting observation is that on the dark side, -5 is still approaching the 100% value but not quite reaching it. There is still some variation from pixel to pixel in that area. However, the +4 side is most defintily over exposed, being pegged at 0 pixel values. I don't know where the cutoff is, as these test images didn't yeild data between +2 and +4. Knowing the ballpark, I can do more tests later that explore the toe and sholder of the curve.
The usable range of the Kodak DC290 is from at least three stops below medium grey through two stops above. At the bright end, the curve dives into the axis and quickly produces white-out areas of over exposure. On the dark side, the curve approaches the 100% line asemtopicly, though the step between one pixel value and the next is a good fraction of a stop. The over exposure behavior is different from the under exposure behavior.
As shown earlier, some shadow detail can be brought out even when it appears black on the screen, as the eye can't distinguish values above 80% black or so on the display. However, solid white areas really are solid 0 pixels. Small differences in pixel values near 0 are visible shade differences. Furthermore, the range to the left of middle gray is larger than the range to the right of it.
So, when in doubt, under expose rather than over expose. To refine the advise given earlier on exposing for the highlights, the brightest value that will show detail is 2 stops brighter than the "average" medium gray mid tone.
These pictures were taken one stop apart. The grey card in the 3rd one reads 48 or 49%, so is the correct exposure for middle grey. The others show me what pixel values represent darker and lighter shades, in one-stop increments.
The other objects in the scene let me further extend the range. I don't know what the absolute brightness of the CD jewl box is supposed to be, but if I can reference the pixel value with the known chart prepared in the previous step, I can deduce that it's about two and a half stops darker than middle grey. The corresponding values in the lighter pictures also track my known chart, and the corresponding values in the lighter pictures allow me to extend my chart.
Likewise for the zipper pull on the dark camera bag, one stop darker yet. And for a specific spot of carpet, which is 3 stops brigher than middle grey.
The carpet in the 3rd photo still shows good detail, has a pixel value of 5%, and represents a brightness of a little over +3 stops. One stop brighter is zero'd out and useless. This is the brightest I can capture.
The camera bag shows some shadow detail up to a pixel value of about 90%. Although it keeps getting darker and darker, still not quite reaching solid 100% at -6.5 stops, it is pretty useless below 4 stops. This represents the darkest object that shows any kind of shadow detail.
The Kodak DC290 can show good image data from values between -4 and +3, inclusive. This is the dynamic range of the system.
The image on the little LCD on the camera isn't very good, so it's hard to judge things with it. However, here's an idea: keep a correctly exposed picture of a gray card in the camera. Then you can reference that to refresh you memory of what it's supposed to look like on the LCD, and compare that with a test shot.
Further advancement of that theme: stick a bright and a dark target in the corners of a grey card. These would be designed to show a detail at the extremes; say, black writing on almost as black background (likewise for the light end). If you can read them both, you're perfectly centered in your range. However, I don't know if you should zero in on medium grey like that, in contrast to conventional wisdom. I'll explain more in the next topic...
Consider what happens if I use a fully-automatic exposure on a mostly light scene. This is a classic example found in any textbook and camera brocure, of when to adjust what a reflection meter tells you.