Noisy sensors -myth explored

"doug" wrote in message
...
Darrell wrote:
"C Wright" wrote in message
. ..

On 7/6/05 6:03 PM, in article , "Ryadia"
wrote:


So... You all think the 20D is a low noise camera, do you?
I suppose the Panasonic and Olympus sensors are pretty bad at high ISO
too, eh? Well maybe for a tech-head the specs might say so but what
about a photographer who takes photographs? would he/she know the
difference? For that matter, would he/she actually give hoot about the
deceptions we all refer to as "product specifications" that we must base
our purchase decisions on?

How too, do you handle the situation when you suddenly discover the
serious limitations of your digital SLR masterpiece, when you start to
use it for traditional, highly creative photography where lighting and
shadow become the picture's prime ingredient but the digital
masterpiece's sensor has some serious short comings when capturing the
two extremes which no one told you about?

Maybe I've lost the plot here but I'd have thought a 20D (read 1D II as
well) would be better at recording detail under adverse lighting
conditions than a lowly P&S camera like the Olympus C760 or Panasonic
FZ20. Surprise, surprise! http://www.technoaussie.com/hmm-detail.htm

How is it that the so named "reviewers" at Pbase and the like never
mentioned the strong points of the Panasonic or the weak points of the
20D? The fact that digital cameras cannot record the contrast range of
film, is the reason these examples of extreme contrast have failed.

Douglas

The sky is blown out in the P&S camera shot, and the P&S shot was made
with
several stops more light. What again is this supposed to prove?
I am not either defending Canon or trying to trash P&S cameras but the
conclusions on this page are not supported by the examples.
Chuck W.


I dislike stupid, flawed tests and sloppy methodology...




I'll spell it out for you Darell.
The myth about Canon DLSRs is that when you shoot RAW, you get a couple of
stops of exposure latitude you can apply during decoding the data to an
image.

Actually that is the myth everyone thinks of RAW. A grossly overexposed or
grossly underexposed RAW image is still bad. Tou are also comparing a 1/2.5"
(10.2mm) CCD with a 15x22.5mm CMOS. The exposure was different, regardless
of RAW. It would have been more accurate to compare a Canon S2 IS to a
Panasonic FZ-20

Any exposure difference between these two cameras is a non event if that
fable is true... Which it is not because the sensor didn't capture enough
data for it to be true.

The myth was false, but spread by pseudo-experts who don't have a clue.

The Panasonic on the other hand has certainly blown the highlights but it
has also recorded data in a area of the picture which is the same density
as the Canon deep shadow area.

The Panasonic and Canon also have vastly different processors Venus vs.
Digic2.

If this were film, the negatives would read both the same density on a
densometer but one would have detail and the other not. What is so
different about digital capture in your mind? The page is about
photogrpahy, not digital specifications. The question that page poses is
why can the Panasonic capture detail in an area of equal density to one
which the Canon cannot?

Film exposed at those vastly different exposures certainly would not have
the same density. We are also comparing sub-110 format to APS format in
sizes.

doug wrote in :

I dislike stupid, flawed tests and sloppy methodology...

I'll spell it out for you Darell.
The myth about Canon DLSRs is that when you shoot RAW, you get a
couple of stops of exposure latitude you can apply during decoding the
data to an image.

Well, if you misunderstand the advantages of RAW, then maybe you could
think that.

Any exposure difference between these two cameras is a non event if
that fable is true... Which it is not because the sensor didn't
capture enough data for it to be true.

I have never heard it said that it makes no difference if you under expose
by a stop or two when you shoot RAW, where the hell did you get that idea
from?

The Panasonic on the other hand has certainly blown the highlights but
it has also recorded data in a area of the picture which is the same
density as the Canon deep shadow area.

Of course it did, the camera had 3 EV more to work with.

If this were film, the negatives would read both the same density on a
densometer but one would have detail and the other not. What is so
different about digital capture in your mind? The page is about
photogrpahy, not digital specifications. The question that page poses
is why can the Panasonic capture detail in an area of equal density to
one which the Canon cannot?

Jeez, what a tool! Do you deliberately come out with lies or are you a
little ignorant?

How do you know that the Canon cannot capture the detail? We certainly
don't see an equivalent picture from the Canon to compare to the Panasonic.

This is one of the stupidest tests I have seen and proves nothing beyond
the fact that exposing a shot correctly makes a difference regardless of
which camera you use.


--
Mark Heyes (New Zealand)
See my pics at www.gigatech.co.nz (last updated 25-June-05)
"There are 10 types of people, those that
understand binary and those that don't"

In article , doug wrote:
Darrell wrote:
"C Wright" wrote in message
. ..

[ ... ]

The sky is blown out in the P&S camera shot, and the P&S shot was made
with
several stops more light. What again is this supposed to prove?
I am not either defending Canon or trying to trash P&S cameras but the
conclusions on this page are not supported by the examples.
Chuck W.


I dislike stupid, flawed tests and sloppy methodology...

[ ... ]

I'll spell it out for you Darell.
The myth about Canon DLSRs is that when you shoot RAW, you get a couple
of stops of exposure latitude you can apply during decoding the data to
an image.

I don't see any signs that the person who set up that page even
*tried* to use RAW format, let alone knew *how* to use it properly. If
you let the camera write as .jpg directly, there is no chance to recover
information beyond the eight bits per color which JPEG can handle.

Any exposure difference between these two cameras is a non event if that
fable is true... Which it is not because the sensor didn't capture
enough data for it to be true.

The Cannon apparently tried to minimize blown highlights, thus
forcing the darker areas into cutoff (for JPEG), but allowing it to
recover more detail if RAW had been used. The P&S didn't care about
preventing blown highlights, so it used a much higher exposure,
preserving detail in the sky, and allowing the preserving and recovery
of the detail at the low end -- *only* if RAW were being used, and
properly processed. (Taking the defaults in whatever program would
probably give something like what we saw.) RAW has to be used
intelligently. It is not at all clear that *either* camera was used
intelligently, and each made different choices about what to preserve.

The Panasonic on the other hand has certainly blown the highlights but
it has also recorded data in a area of the picture which is the same
density as the Canon deep shadow area.

If this were film, the negatives would read both the same density on a
densometer but one would have detail and the other not.

*Where* would they have read the same, given the different
exposures? I doubt whether as much as 5% of the area of the negatives
would have read nearly the same. Certainly neither the highlights nor
the shadow areas would have read the same on both negatives.

What is so
different about digital capture in your mind? The page is about
photogrpahy, not digital specifications. The question that page poses is
why can the Panasonic capture detail in an area of equal density to one
which the Canon cannot?

By sacrificing detail in another area.

I'm not a Cannon user, but *any* camera has to be used with an
understanding of what it does and how. A *lot* of that photo was sky,
and the Cannon (lacking instructions to use narrower metering) tried to
average the whole scene, and as a result exposed for the sky (the
largest part of the image). The P&S probably had its AE sensors more
tuned towards the lower central area, and ignored the bright sky.

I believe that the Cannon could be told to use a particular area
for its auto exposure calculations. Certainly my Nikon D70 can be told
to do that.

The original photographer probably did not bother looking at the
histogram in either camera (presuming that the P&S even offered one).
If you're going to be shooting images for calendar pages, you *should*
know how to use your tools, and how to interpret what they tell you.

Enjoy,
DoN.

--
Email: | Voice (all times): (703) 938-4564
(too) near Washington D.C. | http://www.d-and-d.com/dnichols/DoN.html
--- Black Holes are where God is dividing by zero ---

TAFKAB wrote:

Holy Crap! This is some of the funniest stuff I've seen in a long while! A
camera comparison that provides a direct comparison using different ISOs,
different exposure values, and different image capture variables.
Surprisingly, the results from each camera are, well, different.

Funny, funny ****.

There are similar examples of such stupidity in almost every debate on
any subject you can think of. Medical studies are famous for such crap.
I.e., there was one medical study that concluded that caffeine caused
miscarriages, then when on to state "the women who had (miscarriages)
were significantly older than the control subjects, were more likely to
have been born outside the Nordic countries...and were more likely to
have had previous pregnancies and previous (miscarriages)." These four
huge differences made the entire study completely bogus, but the authors
tried to include the differences in the control groups as a mere footnote.

Mercifully, the page with the stupidity about noisy sensors has been
taken down. I guess that the author was pretty embarassed once the
errors were pointed out.

What kind of dumbass test is this? A DSLR versus a non DSLR. It's like
comparing gas milage of a moped versus a SUV. Lies, damned lies and
statistics. Most photographers will expose correctly.

Scott W wrote:

That is why it is a pretty good idea to have
reviewer who tests under controlled conditions, so you can get an
apples to apple comparision.



Don't you mean a reviewer who is paid by Canon so your brand gets a shining
review so you can feel good about your camera purchase?

--

Stacey

MarkH wrote:


I have never heard it said that it makes no difference if you under expose
by a stop or two when you shoot RAW, where the hell did you get that idea
from?


When you have Canon fans like Scott W posting things like there is no
quality lost shooting at ISO 1600 rather than ISO 100 you wonder how BS
like this gets around? He was talking about shooting 7 stops under and
pulling an image out of it the otherday! Where were you when that sort of
BS was being posted?
--

Stacey

http://www.technoaussie.com/hmm-detail.htm
"Placeholder page"
(email here)

frederick wrote:
http://www.technoaussie.com/hmm-detail.htm
"Placeholder page"
(email here)

It'll be back soon enough Fredrick.
I just need to make the Canon pictures look a little worse and the
Panasonic ones look a little better so the deciples of EOS will have
something worth ranting about!!

Seriously, the pic which was posted from the Canon is the top one of a
pair of 2. Top one correct for sky, bottom one correct for ground, blend
together in Photoshop and get the tonally correct image. This one is for
the calendar and therefore not to be on the 'net.

I posted the wrong pic, Oh, slap my wrist! I just need to find the time
to go through the disc and get the right one and the page will be back
for all to see. Maybe sometime late tonight or early tomorrow. (GMT +10)

Those who claim they could somehow get the exposure correct in one shot
might consider their statements before posting again. Any suggestion of
reading the histogram at 4:00 am in almost total darkness on a winter
morning with the camera mounted on a seven feet high tripod could only
have come from someone who has never gotten out of bed at 2:30 AM to go
and take some photographs - maybe 99.76% of the group's regular posters?
Certainly the ones with "photographic and digital camera advice"!!

Patience is a virtue often found in women, seldom found in men... So
goes the rhyme...

Douglas

The sensors in DSLRs do not have the capacity to gradually and smoothly
cope with highlight information. In the real world this means that the
brightest areas of the scene can all too easily retain no information
at all - no image data is present, just pure white. The sensor
abruptly blows it.

This means that DSLR photographers, especially those who shoot
landscapes, for example, will need to do all they can to accommodate
the highlights, referring constantly to the histogram. Digital
techniques and/or a grey graduated filter is often essential to balance
the brightness of the sky with the land below. If you are new to DSLR
photography, try experimenting with the RAW format so that you can
exercise more control over highlight detail. When the image is
converted, you can fine-tune the image by using levels and curves (the
latter especially useful and powerful) in image-editing software.

Generally speaking, the DSLR photographer should take care not to blow
the brightest parts of the image, although smaller blown highlights do
not take away from the overall appearance of the shot, and are
sometimes unavoidable. Contrast Masking* and Digital Blending
techniques can be used when there are highlight areas within the scene
that need to be balanced with shadow detail to extend the dynamic range
of the print. Digital blending - a very flexible method - can be
used with two or more images. The conscientious digital photographer
will need to learn how to use good image-editing software. (Highlights
are less of a concern when using negative film. Try to hold detail in
the shadows and if necessary use a graduated filter. Always meter off a
mid-tone, and bear in mind that it can be useful to overexpose negative
film slightly.)

"Meter a medium-toned area as medium and the whites will be white,
and the blacks, black" is not particularly good advice for the DSLR
photographer learning to cope with the demands of electronic sensors.
Also, as often as possible it's helpful to push the exposure to the
point where the histogram shows more information to the right of the
display, without the severe abrupt peaks that indicate blown
highlights.

DSLR sensors share slide film's highlight responses but will get more
from the shadows. Shadow retention will be particularly good if the
exposure is routinely pushed just short of blown and unprocessed and
uncompressed data (RAW) is captured rather than JPEG. If this isn't
possible and the subject being photographed allows for it, two or more
images can be used to substantially expand the range, as mentioned
above. Sensor pixels, or light receptors, wrestle with bright light
because their response to it is not gradual. Instead they peak quite
quickly, totally losing highlight data. Digital camera manufacturers
are working on this problem but the application of their technological
advances has been less than ideal. It's certainly an irritating
problem that makes some digital exposures quite tricky, like shooting
slide film, and it's likely to be an integral part of DSLR technology
for quite some time to come.


* In software, a desaturated and inverted (negative) black and white
copy of an image can be blended with the original to expand the tonal
range. The increased density of the b&w inverted copy will be added to
the original. To learn more about this method and other techniques that
control the contrast range, search the Web for Digital Blending and
Contrast Masking (originally a conventional darkroom technique).
Mastering image-editing software is essential for the dedicated
enthusiast who wants to get the very best from his or her DSLR, or
scanned film.

http://www.theimageplane.net/slrs.htm#lose