Nikon D90 - Dark Analysis

Before I begin, I will introduce a piece of notation.
Log Entry ####: - Specifies a log entry
Summary Entry #####: - Specifies a summary

This blog is mostly a logbook but every once in a while I will try to summarize a few interesting things discovered along the way that may be beneficial to others. The following entry is a log entry.

Nikon D90 Dark Analysis
Log Entry Friday July 4th, 2014 8:14pm  


I have ended up interfacing Dave Coffin's dcraw C code into yorick. It took a bit of time to figure out which pieces were essential and which not. I will not post it here just to avoid any copyright issues, but I'll post a summary of the steps later. It is open source but he does request the full source code to be available. If you would like to see it, please comment here or contact me and I will reply.

What I can say is that his code is wonderful and efficient (although obfuscated). If starting to read any raw image, I recommend using his code and interfacing it into your matlab or yorick environment as a starting point. The main piece you need is the call to "load_raw", just look for that in the main function.

Before plotting any images, I first look at the dark images. In this example, I turn the RAW image into a black and white image. Basically, I take the RGB information and just sum them. Here they are:

400 ISO dark near 0 degrees Celsius for a 10 second exposure. This is an average of 6 darks. The inset is the plot of the intensity along the red line (in ADU).

800 ISO dark near 0 degrees Celsius, for a 10 second exposure. This is an average of 5 darks. The inset is the plot of the intensity along the red line (in ADU).

What I would like to point out is how nice these dark images are. The average ADU levels are very close to zero (the dynamic range of the RAW images is 0 to 4096 ADU per pixel). Keep in mind this is a 10 second exposure as well. From now on, I will ignore the dark images since it turns out that the images themselves will have a much higher signal.