CCD Dynamic Range
James
Does anyone know figures on the dynamic range of DCs ( and how to relatively
interpret them ). I can accept and love most of the characteristics of
digital but I find the low dynamic range the most objectionable. If
pictures are taken during cloudy, low light or overcast days, they appear
pretty good. However there does not seem enough of the life-like range of
brightness and darkness of the reproduction during the bright days. I saw a
recent Kodak 260 picture that was a daylight snow scene and the snow and sky
lacked the brightness and whiteness produced by other methods. I see some
DCs are touting the new chips have greater dynamic range and it appears true
but I would sure like to see some honest figures.
Thanks
m...@ij.net
Gerry K8EF
James <jgas...@tampabay.rr.com> wrote in article
<wK6j1.1802$WD5.3...@newse2.tampabay.rr.com>...
> Does anyone know figures on the dynamic range of DCs ( and how to
relatively
> interpret them ). I can accept and love most of the characteristics of
> digital but I find the low dynamic range the most objectionable.
Judging from the performance of silicon photodiodes, it should be possible
to make ccd arrays with very large dynamic range. On the other hand, the
charges generated in the ccd's are converted by analog to digital converter
techniques, so there is a very wide variety of outcomes which can result
depending upon the decisions of the designer of the device.
The number of brightness steps which will result from digitization and
encoding is also under the control of the designer. There seem to be few
arrays which use more than 8-bits per color (24 bits for 3 colors), which
in the absence of any sophisticated encoding, like JPEG, would give rise to
a maximum of 256 brightness levels for each color.
The practical dynamic range of a number of digital still cameras is pretty
good, and the number of brightness steps available in the camera output
often allows some good post-processing to achieve particular aesthetic
effects.
Gerry
http://home.columbus.rr.com/gfoley
Rich Scarlet
The "pro" digital cameras give 10 or 12 bits (30 or 36 bit color),
just like high quality scanners. That's the dynamic range you need,
but they cost $5,000 and (way) up.
All the consumer cameras ($1000 and less) have 24 bit color.
Technically they're probably limited more by the analog-to-digital
electronics and the data storage and processing than by the CCD
itself. Different cameras do better or worse at making the best
possible use of that 8 bits of range. But from the test pictures, it
doesn't seem to me that the brands are notably different in this
regard.
Net result is that any of them will be limited in sunlit or similar
scenes, as you say. I picked the Olympus D-600L, and really like it,
but no question that in a sunlit scene it can't hold both highlights
and shadow areas. It does very well with autoexposure, and exposure
bias is available, but the dynamic range is just *less* than even
slide film, let alone print film. The pictures generally look good,
but if you're accustomed to looking critically at good film prints the
difference is obvious. I don't think the other brands are different.
And, as you say, in cloudy scenes (or with flash) it's not a problem.
I've seen one note - unfortunately I can't find it again - that one of
the new cameras uses 10 bit depth to optimize its exposure. But that
just means it's analyzing more information to pick the exposure
setting before truncating to 8 bits per color. Again, it's up against
the fact that data storage is limited and most simple image processing
programs just aren't set up to handle more than 8 bit color. I wish
one of them used 10 bits and then actually compressed the range to get
to 8 bits instead of truncating.
Rich
P.S. Technically, CCD's themselves are easily capable of signal to
noise ratios of 10,000:1 - that's about 13 bits - and 50,000:1 and
perhaps a bit more is achievable. But the electronics to keep that
range is difficult.
On Sun, 21 Jun 1998 12:10:04 GMT, "James" <jgas...@tampabay.rr.com>
wrote:
>Does anyone know figures on the dynamic range of DCs ( and how to relatively
>interpret them ). I can accept and love most of the characteristics of
Anthony
Good CCDs have a range of 60 dB or more (ten to twelve stops). I don't know
what grade of CCDs is used for consumer digital cameras, though--I doubt it
if is the best of the best.
--
Anthony
James wrote in message ...
Anthony
I also neglected to mention that irrespective of the CCD, with 24-bit color
you cannot get more than eight stops of dynamic range.
Don Stauffer
> > Does anyone know figures on the dynamic range of DCs ( and how to
> relatively
> > interpret them ). I can accept and love most of the characteristics of
> > digital but I find the low dynamic range the most objectionable.
>
> to make ccd arrays with very large dynamic range. On the other hand, the
> charges generated in the ccd's are converted by analog to digital converter
> techniques, so there is a very wide variety of outcomes which can result
> depending upon the decisions of the designer of the device.
>
in transferring the charge in the "bucket brigade." The capacitors
(wells) have only a certain capacitance, which limits the maximum number
of photo-electrons that can be stored. There are a certain number of
noise photons that get transferred, even from dark areas, and always
some variations in the exact number of photo-electrons that get
transferred from well to well. This limits the dynamic range of the
camera. There ARE photodiode arrays that do not use CCD readout, but
these are generally more expensive.
--
Don Stauffer in Minneapolis
home web site- http://home1.gte.net/stauffer/
home email- stau...@gte.net
work email- stau...@htc.honeywell.com
Don Stauffer
James wrote:
>
> Does anyone know figures on the dynamic range of DCs ( and how to relatively
> interpret them ). I can accept and love most of the characteristics of
> pictures are taken during cloudy, low light or overcast days, they appear
> pretty good. However there does not seem enough of the life-like range of
> brightness and darkness of the reproduction during the bright days. I saw a
> recent Kodak 260 picture that was a daylight snow scene and the snow and sky
> lacked the brightness and whiteness produced by other methods. I see some
> DCs are touting the new chips have greater dynamic range and it appears true
> but I would sure like to see some honest figures.
>
> Thanks
> m...@ij.net
Keep in mind that most output media are also very strapped for dynamic
range. The need for good dynamic range in the input device (scanner or
camera) is to allow some latitude for incorrect exposure and for
doctoring of digital file (contrast, brightness adjustments, etc.).
If you can control exposure really well, this helps offset the need for
wider dynamic range in camera.
Ewald R. de Wit
Anthony wrote:
>I also neglected to mention that irrespective of the CCD, with 24-bit color
>you cannot get more than eight stops of dynamic range.
Sure you can.. with gamma encoding for example.
--
-- Ewald
Dave Martindale
stau...@htc.honeywell.com writes:
>A bigger problem comes in between the photodiode and the A/D. That is
>in transferring the charge in the "bucket brigade." The capacitors
>(wells) have only a certain capacitance, which limits the maximum number
>of photo-electrons that can be stored. There are a certain number of
>noise photons that get transferred, even from dark areas, and always
>some variations in the exact number of photo-electrons that get
>transferred from well to well. This limits the dynamic range of the
>camera.
The dynamic range can be improved by making each pixel physically larger.
For the same light level during exposure, each pixel will accumulate more
charge - but it can hold more charge without saturating because of its
larger area. When this charge is shifted to the output amplifier, it
gives a stronger signal. Increasing pixel area might also increase
dark current, but cooling can eliminate that. So if you are willing
to use relatively large CCD cells and cooling, you can get very good
dynamic range out of a CCD, and that's one reason CCDs have pretty
much replaced film in astronomy.
The problem is that a larger chip is much more expensive than a smaller
one with the same number of pixels, and cooling is expensive too.
Plus the larger CCD means a longer focal length lens to keep the same
field of view, and a heavier lens to maintain the same f/number at
the greater focal length.
So most digital cameras use a small CCD and small lens, ending up
with small size and weight, but some compromise in image quality.
A larger sensor would give better quality, even with the same number
of pixels, but you would start approaching SLR size and weight. This
is, in fact, what some people want anyway, but not everyone.
Dave
Anthony
Ewald R. de Wit wrote in message ...
>Sure you can.. with gamma encoding for example.
If you don't care about any detail in-between, yes.
--
Anthony
Michael McGuire
0]>
: I also neglected to mention that irrespective of the CCD, with 24-bit color
: you cannot get more than eight stops of dynamic range.
: --
: Anthony
To add to this, the maximum dynamic range you can actually get at the image
plane, be it film or ccd, is about 10 stops due to flare. Some flare is due
to light that reflects off the image plane onto the interior of the camera and
then back to the image plane. Some more of it is due to the fact that every
points on the object is imaged as a spread of light concentrated about a
point in the image but spreading over the whole plane and further adding to a
base exposure that the whole image gets. It is image dependent, being greater
for mostly light images than for mostly dark.
Mike
--
Michael McGuire Hewlett Packard Laboratories
email:xmcg...@xhpl.xhp.com P.0. Box 10490 (1501 Page Mill Rd.)
(remove x's from email if not Palo Alto, CA 94303-0971
a spammer)
Phone: (650)-857-5491
************BE SURE TO DOUBLE CLUTCH WHEN YOU PARADIGM SHIFT.**********
Don Stauffer
Good point, Mike. Unless the lens is a VERY good one, flare and ghosts
will indeed limit the input dynamic range. This problem is so bad in
many telephoto lenses that I fear we have almost forgotten what wide
range photos look like. With most 35mm cameras today with a zoom lens,
you will never come up with Ansel Adams quality prints.
On the plus side for digital cameras, a general rule of thumb is that
the more surfaces in the lens, the worst the flare. Since many of the
digital cameras have pretty simple lenses (few elements) their flare
performance may actually be better than many current zoom lenses for
35s.
Anthony
Don Stauffer wrote in message <3590FB...@htc.honeywell.com>...
>Good point, Mike. Unless the lens is a VERY good one, flare and ghosts
>will indeed limit the input dynamic range.
I have yet to see anything even remotely approaching a good lens in a
consumer digital camera. In fact, I think that much of the poor quality
attributed to the digital format is in fact a consequence of a cheap lens.
Video buffs know that the same consumer camcorder will give astonishingly
sharp and clear results if you replace the cheapo lens with a good lens.
Unfortunately, good lenses cost so much that they are usually passed upon by
manufacturers of consumer gear.
--
Anthony