Re: Nikon D80/200 - Canon 30d

newsmb@xxxxxxxxx <newsmb@xxxxxxxxx> wrote:
On Jun 6, 8:12 am, Wolfgang Weisselberg <ozcvgt...@xxxxxxxxxxxxxx>
wrote:

So how about D200 versus 30D WITHOUT (Noise Ninja or NeatImage)?

And -- even more interesting --- D200 images + (Noise Ninja or
NeatImage) versus 30D + (Noise Ninja or NeatImage)?

Do whatever you will. I DON'T CARE. None of this makes one iota of
difference in real photography.

I see. Would you care to define "real photography"?

Or is that your way of saying "you were right, but I won't
admit it"?

Which is why dpreview writes:
The dpreview guys are so far out to lunch, especially when it comes
to noise, that their opinions don't matter to me one iota.

Ah.
So what makes your oppinions more valuable than dpreview's
oppinions?

Apparently, a basic understanding of the state of sensor technology.

You have a basic understanding?

Then you will surely be able to follow the factual data given
below to their conclusions.

You surely have also read and understood
http://www.canon.co.jp/Imaging/cmos/technology-e/noise_reduction.html
and friends then.

After all, dpreview has been doing that stuff for
years and years with well over a hundred cameras. (Which doesn't
make them right, but at least increases the chance they know what
they talk about.)
What are your references?

I used to host the now-defunct www.digital-sucks.com

Used to host? As in "provide the internet infrastructure for"?
Or did you provide content? If so, what type of content?

And when would that site have existed?

What are yours?

I happen to have 2 eyes that happen to function well enough to
provide my working brain with facts. I also happen to use
digital camera on quite insane effective ISO settings.

It couldn't be that Canon uses a CMOS sensor --- which is a
completely different technique to the CCD sensor of the D200 ---
and it couldn't be that a completely different technique just
_might_ have a different type and strength of noise!

The CCD vs. CMOS argument does not exactly work in your favour.
See:
http://www.shortcourses.com/how/sensors/sensors.htm
http://www.dalsa.com/markets/ccd_vs_cmos.asp
http://electronics.howstuffworks.com/question362.htm
http://www.dalsa.com/shared/content/Photonics_Spectra_CCDvsCMOS_Litwi...

Very nice.
You have now conclusively proven that Canons CMOS-sensors can
not exist.

I have pointed to some major electronics websites that cast serious
doubt on your contention that the Canon has a less-noisy sensor. given
that as a rule CMOS sensors are *more* noisy than CCDs.

Don't be so meek. If I take everything these websites say as
a fact (and ignore that they even disagree among themselves in
parts), then Canons cameras cannot have a CMOS sensor.
Given the fact that they produce the low amount of noise they do.

At this point I have 2 choices:
a) believe the experiments, in spite of the theory --- this is
what scientist do, though they might take a very close look
at the experiments to check their validity.
b) believe the theory, in spite of the experiments --- this is a
popular choice of fanatics and extremists and creationists
everywhere, though even the Roman Catholic Church by now
agrees that Gallileos theses were not heretical and all that.

Well, then ...
.... let's try what happens if I present facts (which
obviously have zero influence on "real photography", as "real
photography" can even be done with a pinhole camera and ISO 5
chemical film, which is much worse than any DSLR can manage.)

D200:
ISO Gain Read Noise
(electron/ (electrons)
12-bit DN)
100 7.98 10.0
200 4.0 8.1
400 2.0 7.7
800 1.0 7.4
1600 0.5 7.4

Full Well capacity at ISO100: 32,680 electrons
Source:
http://www.clarkvision.com/imagedetail/evaluation-nikon-d200/index.html
That means:
ISO 100 200 400 800 1600
RN in DN 1.25 2.03 3.85 7.4 14.8
(DN is the Data Number, the value between 0 and 4095, the camera
reports. RN in e- divided by gain is thus RN in DN.)
That site attests the D200 to be operating at "near perfect levels"
for the sensor, and "for high signals, noise is dominated by
photon statistics." (and hence the well capacity), but the read
noise should be lower.

The data also indicates that the maximum usable ISO for RAW is 800,
since at 1600 you are just multiplying the signal by 2 (compared to
800) and thus cutting away a full stop for highlights. (Of course,
if you want/need in-camera JPEGs, that's a different story.)


The 20D has the following gains and read noises:
ISO Gain Read Noise
100 12.4 25.3
200 6.2 13.5
400 3.1 7.5
800 1.5 4.8
1600 0.8 3.6
3200 0.4 (none given)

Full Well capacity @ ISO100: 51,400
Source:
http://www.clarkvision.com/imagedetail/digital.signal.to.noise/
That means:
ISO 100 200 400 800 1600
D200 RN in DN 1.25 2.03 3.85 7.4 14.8
20D RN in DN 2.04 2.18 2.42 3.2 4.5

Advantage D200 at ISO 100 and 200.
Advantage 20D at ISO 400, 800 and especially 1600.

Additionally, the data shows that, as with ISO 1600 on the D200,
ISO 3200 on the 20D is contraindicated for RAW.

Looking, however, at the full well capacity, we see a slight
advantage for the 20D. Noise at bright parts is photon noise.
Phothon noise is the square root of the photons collected.

Hence (at ISO 100) D200 20D
Full Well (electrons): 32,680 51,400
Noise (electrons): 181 227
Noise (DN) : 22.7 18.3
Signal/Noise: 180.5 226.4

Advantage 20D: 25% better signal/noise ratio.

Let's assume the sensors are 100% linear (the D200 is not in
the last 10%, but that's not a real problem). Let's look at
the noise for ISO 100-1600 over 10 stops of light difference.
(Sources:
http://www.clarkvision.com/imagedetail/evaluation-nikon-d200/index.html
http://www.clarkvision.com/imagedetail/digital.signal.to.noise/
(full well for D20 estimated to be half for each ISO step)
Calculate as follows:
- Take the full well value for 0 stops
- halve the value for every stop below 0
- take the photon noise as SQRT of the sensor
- add the read noise (for the ISO setting) to the
photon noise, for the total noise.
- divide the total noise through the electrons in
the sensor well.
(- round to 2 digits after the decimal point for
displaying)
)

Stops below full sensor
ISO 0 -1 -2 -3 -4 -5
100 200D 173.13 120.32 83.06 56.82 38.41 25.57
D20 203.96 138.46 92.67 60.93 39.18 24.57

200 200D 121.88 84.13 57.55 38.90 25.89 16.91
D20 147.86 101.29 68.60 45.77 29.98 19.19

400 200D 84.44 57.85 39.18 26.14 17.11 10.94
D20 106.32 73.30 50.05 33.76 22.41 14.57

800 200D 58.09 39.40 26.33 17.28 11.08 6.91
D20 75.63 52.25 35.79 24.23 16.15 10.56

1600 200D 39.45 26.38 17.32 11.11 6.95 4.20
D20 53.29 36.77 25.14 16.97 11.27 7.35

ISO -6 -7 -8 -9 -10
100 200D 16.69 10.65 6.60 3.96 2.29
D20 14.97 8.85 5.07 2.83 1.54

200 200D 10.79 6.70 4.02 2.33 1.32
D20 11.96 7.24 4.26 2.43 1.35

400 200D 6.81 4.12 2.40 1.37 0.76
D20 9.24 5.71 3.43 2.00 1.09

800 200D 4.20 2.45 1.40 0.78 0.43
D20 6.76 4.21 2.55 1.45 0.83

1600 200D 2.45 1.40 0.78 0.43 0.23
D20 4.69 2.91 1.70 0.99 0.56

As you can see, even the D20 out-performs the 200D in signal
to noise in almost every situation, except at ISO 100 in
darker image areas (-5 stops below full), and they are about
even at very dark areas (-8 stops below full) at ISO 200.

*If* we *assume* a head room of 2 stops, then -5 stops is -3
stops below "medium gray" (i.e. in the shadows), and -8 stops
is -6 stops below "medium gray" (deep deep shadows).

Also have a look at
http://www.clarkvision.com/imagedetail/digital.sensor.performance.summary/low-light-sensitivity-factor-v1.gif
(Figure 7 of
http://www.clarkvision.com/imagedetail/digital.sensor.performance.summary/
), which "describes the camera performance in shadows or
darkest parts of an image at high ISO.". The D20 is about
as good as the 10D (which has larger sensor sites), but a
factor 3(!) below the 20D. Even the D50 and D70 outperform
the D200 there, by almost 100% and 70%, respectively.


I have also pointed to Ken Rockwell's tests, which pretty much confirm
that any differnces in noise are attributable to stronger aliasing by
the Canon.

Well, if you look above, the only data used is read noise and
sensor well capacity --- none of which have anything to do
with aliasing, lenses, blurring, etc.

That nicely deconstructs that theory.

Oh, regarding aliasing, look at
http://astrosurf.com/buil/d70v10d/eval
and look for MODE 1 and MODE 3 for what Nikon secretly does to
RAW files ...

I have nothing further to say on the subject.

I have to add that you should also have al look at the 20Da,
which has even less read noise behaviour than the 20D ... and
basically the same sensor.

-Wolfgang

--
In science it often happens that scientists say, 'You know that's a really
good argument; my position is mistaken,' and then they actually change
their minds and you never hear that old view from them again. They really
do it. It doesn't happen as often as it should, because scientists are
human and change is sometimes painful. But it happens every day. I cannot
recall the last time something like that happened in politics or religion.
-- Carl Sagan, 1987 CSICOP keynote address
.

Relevant Pages

  • Re: How important is Olys dust reduction system?
    ... Of course there are the limiting effects of noise, and if an E-500 or E-330 begins to show them above ISO 400, a compact camera like the Sony 828 begins to shows them at even lower ISO numbers, according to the even smaller size of their sensor. ... As I said, printed to a reasonable noise / "grain" limiting size, the E-500 can't print as cleanly nor as crisply as the 20D for that reason. ...
    (rec.photo.digital.slr-systems)
  • Re: How important is Olys dust reduction system?
    ... It is absolutely no problem to enlarge 7.5 or 8 megapixel FourThirds images to excellent, clean and sharp 16"x12" prints - virtually no difference to 8 megapixel cameras of other makers, slightly if not significantly better than 6 megapixel cameras of other makers. ... E-330 begins to show them above ISO 400, a compact camera like the Sony 828 begins to shows them at even lower ISO numbers, according to the even smaller size of their sensor. ... I used to try to stay below ISO 160 with that camera, which delivered about as much noise there as I was inclined to tolerate. ...
    (rec.photo.digital.slr-systems)
  • Re: Noise levels as a function of pixel size
    ... small sensor data I can find. ... the 1D MarkII has an ISO 50 setting. ... reached chip saturation at their lowest ISO in the other astronomer ... and the total noise at that same level. ...
    (rec.photo.digital)
  • Re: Digital Camera Sensor: 14bit sucks?
    ... the read amplifier, ... which is a fixed gain (does not change with ISO). ... source of the sensor read noise. ... Sensor read noise comes from the first amplification ...
    (rec.photo.digital)
  • Re: ISO 200000 ?
    ... Doesn't suggest that you can achieve 3e read noise in a sensor fitted to a dSLR. ... That way you could select the ISO after the event, and improve the effects that you are showing in your shadow assessment page. ... Compare that to your device saturation of 20 photons, which can have a maximum dynamic range of 20. ...
    (rec.photo.digital)