[Patrick Sun]

My trusty Canon S2 IS digital camera has developed 4-6 dead pixels, and it's starting to bug me, but not to the point of retiring the camera yet, but I know that the white spots will show up in every shot I take with that camera, and more dead pixels will inevitably show up in future photos as well.

So is it more of a matter of dead pixel location within the shot, in combination with the sheer number of dead pixels that makes you move on to another camera?

[Dave Ringkor]

You can effectively "get rid" of dead, hot, or stuck pixels by using a Linux and Mac command-line utility called jpegpixi (Google it). First, you use the software to scan a test image and map the locations of all the bad pixels. Then, you run your photos through it, and it filters out the bad pixels using the map you created. I wrote a script that works on Linux and Mac that extracts the EXIF data from all the jpeg files in a folder, creates a matching text file for each one that can includes the date and time the photo was taken along with user comments about where it was taken and who's in it etc., renames the files from IMG_xxxx.JPG to christmas_xxxx.jpg (for example), renumbers them to remove gaps, and gets rid of all the hot or stuck pixels.

[Patrick Sun]

Well, no Mac, and definitely won't be rigging up a Linux box just to clean up some dead pixels (but maybe I could use VM Ware to run such a program with a Linux VM Ware build). How does the program clean up the pixels? Does it just take the adject areas around the dead pixels and blend in the average color where the dead pixels are?

[Dave Ringkor]

From the jpegpixi web site: It supports averaging adjacent pixels; linear, quadratic, and cubic interpolation in one dimension; and bilinear, biquadratic, and bicubic interpolation in two dimensions. There are sample images available that illustrate the effects. Another nice thing about the program is that it doesn't decode and re-encode the whole picture resulting in jpeg degradation. Rather, it only decodes and re-encodes the DCT blocks that contain the bad pixels, while everything else in the image is left as-is.