Olympus C-8080 dark frame noise

• see also:
  • home page
  • Olympus C-8080 camera
  • astrophotography
  • compare these with:
    • my tests with the Olympus E330 digital SLR - now this is fantastic - only sparse noise at 5min
    • the remarkably low noise levels in the Canon 1D Mark II digital SLR (although much more expensive) http://www.pbase.com/image/30343120 without noise reduction.
    • Pentax *istD - http://homepage.mac.com/warenes/FileSharing9.html

To demonstrate the dark frame noise inherent in my camera, here are photos with lens cap on at approx. 16degC:

Details	Full image resized to 10%	Cropped image of 100% original size, cropped from right bottom corner region
400 ISO, 300sec, no NR
400 ISO, 60sec, no NR
50 ISO, 180sec, no NR
400 ISO, 60sec with NR

As you can see, the bottom left corner is severely affected by by dark frame noise to the extent that the pixels in the bright region are almost fully saturated in a 300 sec exposure and thus any object imaged in the area of this bright region would become invisible in the final dark frame subtracted image.

• the pixels in the brightest region have RGB values of:
  • 400 ISO 300sec: 255R, 186G, 241B ie. fully saturated in red and almost so in blue
  • 400 ISO 60sec: 192R, 107G, 148B ie. almost fully saturated in red
• the pixels in the bottom right region have RGB values averaging:
  • 400 ISO 300sec: 157R, 71G, 116B
  • 400 ISO 60sec: 108R, 47G, 63B

When one adds the effects of light pollution, it does not take much to saturate the left bottom corner and the top edge of the frame, and no amount of subtraction of the dark frame and subtraction of contribution of the light pollution will render an object visible once a pixel is saturated by the combination of noise and light pollution.

The bright area in bottom left seen above is just visible in dark frames taken at 4sec exposures at 400 ISO but is not visible in exposures shorter than 1sec at 400 ASA.

Here is the test repeated but at 9.5deg C which shows a considerable improvement, although there seemed to be more hot pixels:

Details	Full image resized to 10%	Cropped image of 100% original size, cropped from right bottom corner region
400 ISO, 480sec, no NR
400 ISO, 180sec, no NR
400 ISO, 60sec, no NR

Possible options to minimise the dark frame noise saturation problem:

• ensure the object of interest is is the darkest region - in this case, the right upper quarter, avoiding the top edge.
• minimise light pollution - dark site and short exposure duration
• although the maximum exposure duration for the camera is 8 minutes, avoid dark frame saturation by minimising exposure duration to less than 120secs unless there is little light pollution and object can be confined to the darker regions of the frame
• try colder temperatures to reduce thermal component of dark frame noise

In addition, to further reduce dark frame noise in the final image, take RAW images and either:

• take dark frame photos with NR of the same duration at the time of your session, perhaps at the start and at the end, then create a master dark frame and subtract this from your image taken with NR on, or,
• take dark frame photos without NR at the same exposure duration at the start and end of your session, then create a master dark frame from these and subtract this from your original image taken with NR off.

Contribution of light pollution using Olympus C-8080:

• outer suburban Melbourne at zenith:
  • 400 ISO, 60sec, f/3.5: average background pixel = 93R, 100G, 108B
  • 400 ISO, 40sec, f/3.5: average background pixel =  55R, 75G, 73B
  • 400 ISO, 15sec, f/3.5: average background pixel = 36R, 41G, 44B
• 30km from Melbourne with mild light pollution 30 deg. above horizon, taken at ~5degC:
  • 400 ISO, 240sec, f/3.5: average background pixel = 111R, 75G, 75B
  • 400 ISO, 180sec, f/3.5: average background pixel = 86R, 73G, 54B
  • 400 ISO, 120sec, f/3.5: average background pixel = 47R, 35G, 26B
  • 400 ISO, 50sec, f/3.5 at zenith: average background pixel = 17R, 19G, 18B

NB. the above average pixels are only eyeballed averages not true calculated averages over an area, and are provided to give a guide only.