The following are Duncan Munro analyses of D80 suitability for long exposure astronomy and comparison to D50. This write up was originally emailed to Andrew and is posted on Andrew's website with authors permission.
Here's some IRIS stats for a 15min DF (mode 3) at room temp. IRIS is able to
read the RAW files directly and output the actual 12bit (0-4095) data values.
(IRIS is a freeware astronomical image processing package and can be obtained
here: http://www.astrosurf.com/buil/us/iris/iris.htm
)
=== Image: c:\img2006-09-17\80d1.pic ===
Format: 3904x2616
(0,0)-(0,0) - Binning: 1x1
17/09/2006 - 09:57:08 - IT: 902.00
bayered (mono) image:Mean: 7.3 Median: 0
Sigma: 68.5
Maxi.: 4095.0 Mini.: 0.0
2nd 15min DF subtracted from the first showing residual noise:
Mean: -0.1 Median: 0
Sigma: 3.9
Maxi.: 814.0 Mini.: -622.0
The good news is that the dark current seems to be extremely low.
There are some random hot pixels but they seem to be mostly removed by the DF.
The bad news is that amp glow is worse than on the D50. Here's the stats for a
portion of the most intense amp glow after 15min:
Mean = 214.36 - Median = 210
Noise (deviation) = 54.51
Maxi = 4085 - Mini = 128
still not bad, and easily removed via a DF.
Here's the stats for the equivalent area from a D50 dark frame:
Mean = 50.09 - Median = 47
Noise (deviation) = 18.95
Maxi = 280 - Mini = 9
Given that the D50 is likely to have less Gain, this really shows how quiet the
D50 amp is. Moreover, the D80 uses a two channel amp so there are two amp
induced glow patterns rather than one (and a bit), on the D50. Speaking about
dark frames, I discovered that the D80/D200, in camera long exposure noise
reduction, only generates a DF of 1/2 the length of the light frame. As a result
it will not subtract all the noise and amp glow. External software subtracted
DFs are mandatory.
these are the colour, debayered, image stats for the centre 1/4 of the D80
frame:
Mean = 2.83 - Median = 0
Noise (deviation) = 68.74
Maxi = 4095 - Mini = 0
this is quite good and compares well with the 30min D50 dark frame:
Mean = 4.52 - Median = 2
Noise (deviation) = 23.32
Maxi = 4026 - Mini = 0
For the whole frame:
*** Red layer ***
Mean: 6.6 Median: 0
Sigma: 69.7
Maxi.: 4095.0 Mini.: 0.0
*** Green layer ***
Mean: 5.1 Median: 0
Sigma: 51.6
Maxi.: 4055.0 Mini.: 0.0
*** Blue layer ***
Mean: 11.8 Median: 0
Sigma: 37.7
Maxi.: 4095.0 Mini.: 0.0
stats for a 30minute ISO D50 frame (mono bayered):
=== Image: c:\copyofnef\irglob\ir200d.nef ===
Format: 3039x2014
(0,0)-(0,0) - Binning: 1x1
26/05/2006 - 17:59:01 - IT: 1799.60
Mean: 7.2 Median: 0
Sigma: 30.6
Maxi.: 4095.0 Mini.: 0.0
So the noise levels are very similar, except that the Gain on the D80 shots
is likely to be 8-10 versus 7.45 on the D50. So a D80 actually generates about
80-100 units of dark current (using a 16bit scale and doubling the D80 data to
compensate for exposure time) against about 52 for the D50. Gain is the amount
that the CCDs electron well is divided by during the A/D
conversion to get a 12bit (0-4095) data unit. As photons are converted to
electrons by the CCD they are stored in each pixel's electron well, until read
out at the end of the exposure to generate the image. The D50 electron well can
hold about 30,500 electrons at ISO200 and this value must then be divided by
7.45 to produce the 0-4095 12bit RAW file, so the Gain (multiplier factor) is
7.45 for the D50 at ISO200. I would guesstimate that the D80 has an electron
well capacity of 33,000 to 41,000 electrons at ISO100 for a gain factor of about
8 to 10, but this is only a crude estimate, based upon the pixel area and
reviews of D80 performance versus the D200.
IRIS can also estimate the noise in the dark frames using the BGNOISE command,
and for the D50 it is:
RMS noise: 17.995
and for the D80
RMS noise: 6.578
which indicates that there is about the same noise in the D80 image (remember
15mins for the D80 and 30min for the D50) which is a good thing.
I would guess that much of the D80 dark current difference vis a vis the D50 is
generated from the amp glow. Overall, the D80 would appear to be
very usable for long exposure astronomy, but given the increase in amp glow I
would recommend shorter exposures times than a D50, except at ISO 100 or 400.
Using a separately generated dark frame, to remove the amp glow and dark
current, will be mandatory with the D80.
Author:
Since Roger Clark made many of calculations based upon
Duncan Munro's D50 data, additional credits to:
http://www.clarkvision.com/imagedetail/evaluation-nikon-d50/index.html
and
http://www.clarkvision.com/imagedetail/evaluation-nikon-d200/index.html
© Duncan Munro 18 September 2006
With questions regarding this article, please email Duncan directly. Just remove the XXX and spaces from the following address: dmunro XXX @ XXX sfu.ca
- Support Andrew's efforts to document Canada's beauty - purchase the maps and software that you need from: www.CanadianMaps.ca or license / purchase photographs from this site - contact Andrew@TopoCanada.com