Thanks for the reply!
How would I find or figure out the offset of the actual data?
From the Strip Offsets tag in the DNG header. Or you can subtract the raw size (width * height * bpp /8) from the file size, since in CHDK the data is always at the end (note this is only valid for CHDK DNGs, not DNG in general.)
Alternately, if you just want raw data without any exif, you could use CHDK raw instead. This is just a framebuffer dump. However, it's in a packed little-endian format that very few programs are likely to understand (In DNG data is always big-endian for bit depths that aren't an integer number of bytes).
FWIW, I was able to open a CHDK DNG using the "file -> import -> image" option in the "fiji" imagej package
http://fiji.sc/Fiji It looks quite distorted, but I don't know how to use ImageJ.
Is there a way to take pictures possibly in a higher bit? Or is that a limitation of our camera?
It's defined by the hardware.
Its just a few pixels in an otherwise correct image. Its just throwing off the results when measuring the section.
This sounds like normal bad pixels. The camera sets the pixels it knows are bad to 0, and "fixes" them when creating the jpeg. When only one bayer elment is bad, these will show up as spots of color. If you use DNG 1.1 on the camera, they will be interpolated over. If you use DNG 1.3, your DNG processing software is supposed to take care of them, but most programs don't support this feature, so you see unfixed badpixels instead.
If you are doing some scientific work, you might want to just ignore these pixels values instead.
You will probably also see some hot pixels, especially if you are using long exposures.
Do you suggest to upload the files to dropbox to see if they are openable after converting, or what is the purpose of uploading to one of those hosting sites?
I was suggesting you could post a sample that we could look at, because if there's a bug in the CHDK port for your camera, it could output broken DNG files.
