« Home | Panning with a slow shutter speed » | Lens Contrast (Part 2) » | Better to increase the ISO » | Wildlife Photography » | Lens Contrast (Part 3) » | Lens Contrast (Part 4 : Color Creates Contrast) » | Lens Contrast (Part 5 : MTF Chart) » | Lens Contrast (Part 6 : MTF Chart) » | Lens Contrast (Part 7 : MTF Chart) » | Lens Contrast (Part 8 : MTF Chart) »

High ISO is better than underexposure


_DSC4086, originally uploaded by shutterhack.

Taken with a Nikon D50 and AF Zoom-Nikkor 70-300mm f/4-5.6G lens

In general dSLR imaging sensors have about 1/10th of a stop latitutde for overexposure but a bit more for underexposure, depending on your skills at image processing.

You can easily check this yourself on your own camera with a well chosen subject and bracketed exposures. You may find that for your uses there is more or less than 1/10th latittude at the bright end. Hence the rule that dSLRs should be used like slide film: expose for the highlights. There are some plug-ins that claim to be able to retrieve data from mildly blown highlights by using algorithms to derive detail from the color information that may not have clipped, e.g. the green channel. This is analagous to Kodak's recent announced modification of the Beyer filter. This can only be done with raw data.

Particularly for amateur uses, shooting raw, reasonable quality images can generally be brought up from areas that are as much as 1 stop underexposed. This is why raw is the most flexible way to use a dSLR: you don't really increase the limited latitude of the sensor but you can create the appearance of an image that has greater latitude than in-camera jpeg processing can achieve.

Obviously there are noisy and less noisy ways to bring up the detail from underexposed areas.Ultimately you have to understand the limits of your gear and software in your hands (not what you read in a magazine or in sources of misinformation), sort of a modified Zone approach, so that you don't spend too much time fiddling in Photoshop trying to bring back the dead.

Labels: , ,