View Single Post
  #30  
Old 31-01-2008, 10:25 PM
citivolus's Avatar
citivolus (Ric)
Refracted

citivolus is offline
 
Join Date: Sep 2007
Location: Carindale
Posts: 1,178
Quote:
Originally Posted by chunky View Post
Regarding ISO:
There is a school of thought that because of the 14-bit light collection (never, never use jpg that only gives 8-bits of data and with artefacts), the Canon reaches 'gain-unity' at ISO 400, so from a astrophotography point of view you never need to go above this. For a 20/30D I think the figures are around ISO 1000 (only possible on the 30D - 800 or 1600 on 20D etc).
Thanks Clinton, you inspired me to do some reading, and after having a look at http://www.clarkvision.com/imagedeta...mance.summary/ I now understand what you are saying. My understanding is that this has the most impact if you are stacking images, as a single photon per frame will be visible in data shot at ISO 400, while taking it up to ISO 1600 would just give quantization error and gaps in the histogram, even if pulling the data slightly more above the read noise. By the time you have stacked 8 frames or so at ISO 400, read noise should average out and be subtractable, and single photons should be apparent, leaving you with just the signal and the thermal noise. You then have a larger well to work with and can shoot a longer exposure without saturating.

Amusingly as an extension of this, artefacts aside, unity gain for JPEG shooting would be somewhere around ISO 20,800

Regards,
Eric
Reply With Quote