Matching Pixel Size and Telephoto
Hughes
My experience is mostly on astronomical telescopes
and I now want to view terrestrial daytime subject
like sceneries or trees/nature.
I have a 4" 1000mm F/10 Telephoto. I want to get
a digicam that can produce the optimum image
quality. In Astrophography, what we do is get
the resolving power of the scope and get a CCD
with pixel scale that is at least 1/2 of it.
Calculating:
Resolving power of 4" telephoto is 1.16 arcsec
Airy disc linear size of F/10 is 13 micron
that subtends 2.68 arcseconds of the target
Questions.
What do you do in terrestrial photography? Do you
also get pixel scale that is 1/2 the resolving
power of the telephoto? Or do you just match
the pixel size to the airy disc size (in my
case, 13 micron which is the linear diameter
of the airy disc)??
In angular term, one gets the pixel scale
which is 206265 x pixel size/focal length
or 2.68 arcsec/pixel. But since the resolving power
of the telephoto 4" aperture is 1.16 arcsecond...
then 2.68 arcsec/pixel is undersampled. For optimum
sampling, the pixel scale should be at least 0.58
arcsec/pixel or half of the resolving power, at
least in astrophotography. How about in terrestrial
daytime photography? What's the pixel scale?
Hughes
Me
> Hi,
>
> My experience is mostly on astronomical telescopes
> and I now want to view terrestrial daytime subject
> like sceneries or trees/nature.
>
> I have a 4" 1000mm F/10 Telephoto. I want to get
> a digicam that can produce the optimum image
> quality. In Astrophography, what we do is get
> the resolving power of the scope and get a CCD
> with pixel scale that is at least 1/2 of it.
> Calculating:
>
> Resolving power of 4" telephoto is 1.16 arcsec
> Airy disc linear size of F/10 is 13 micron
> that subtends 2.68 arcseconds of the target
>
> Questions.
>
> What do you do in terrestrial photography? Do you
> also get pixel scale that is 1/2 the resolving
> power of the telephoto? Or do you just match
> the pixel size to the airy disc size (in my
> case, 13 micron which is the linear diameter
> of the airy disc)??
>
pitch that the manufacturers happen to churn out. I never heard of
anyone checking available lens resolution figures before buying more
megapixels in the latest camera body - but you do hear the occasional
whine that existing lenses can't match sensor resolution, which is
probably true in some cases.
At normal working apertures with dslrs, sensel size is still well below
double (average wavelength) airy disk size. That's around f11 for APS-c
dslrs with 12mp (or >24mp with 35mm sensor format dslrs with 24mp).
There's generally a small but gradual fall-off in resolution from the
point where average wavelength airy disk is already about sensel size,
but a more prominent resolution loss once 2x sensel size is reached and
exceeded (possibly due to RGBG bayer filter interpolation and
demosaicing methods), hence the notion that many lenses are sharpest at
say f5.6 or f8 though resolution is reduced by inevitable diffraction at
smaller apertures, not by some optical design "fault".
At present pixel densities and with current formats, most reasonable
quality optics can at least resolve (in the centre of the frame) close
to and around theoretical limits at current pixel densities, exceptions
being typical inexpensive long zoom lenses at maximum zoom, and wide
angle lenses - some of which aren't very good toward the edges of the
frame, even stopped down to just below the point where diffraction
really robs resolution. In practice, technique (focus and elimination
of camera shake) are more significant as expectation of final resolution
has increased with higher pixel density cameras.
Small sensor high megapixel "Point and Shoot" cameras are already at or
well beyond the "diffraction limit" at normal working apertures -
particularly at the long end of the zoom range. But the manufacturers
keep churning out models with more and more pixels - because that's
apparently what many people want.
So at f10 (maximum fixed aperture), arguably there wouldn't be much
point going above about 25 megapixels on a 36x24mm sensor for "normal"
photography. But if there are lenses good enough at wider apertures
than f10 (and there most probably are, and will be in future), then
there may be a point in the "megapixel race" going well beyond 50
megapixels for that format.
N
news:73e6a9a3-3c81-4758...@c18g2000prh.googlegroups.com...
I tend to consider the subject matter, composition, lighting and then what I
might do with the resulting picture. I might print it, frame it and hang it
on the wall or give it to someone to keep or just keep it on file.
Until you get a camera and take pictures all the theory in the world is
worth nothing.
Hughes
>
> - Show quoted text -
I plan to buy a Canon 1000D. Calculating the pixel
scale of the 1000mm telephoto f/10:
pixel scale = 206265 x 5.7 micron/1000
pixel scale = 1.175 arcsec/pixel
resolving power of telephoto is 1.16 arcsec
airy disc subtended is 2.68 arcsec
In terrestrial photography, do you admit that we
shouldn't aim to match the resolving power of
the telephoto but the airy disc size itself or
its average diffraction rings. If so, then the 1000D
is oversampling already?
If I get a used Canon 300D, the pixel scale is
206265 x 0.0074/1000= 1.5 arcsec/pixel, which is
about optimally sampled?
But here's a complication. My telephoto is a mirror
or mak, central light of airy disc is pumped to the
diffraction rings. Add to it is possible wavefront
error of 1/3 wave, then my airy disc is twice the
size so it's about 26 micron or subtending 5.3
arcsecond. So I think I must get a camera
with pixel size of at least 11 micron? Resolution
there can be lower than 5 megapixel (this
means if my telephoto, increasing the megapixel
wouldn't produce better image, but just magnifying
all the defects).
I'm calculating what is the theoretical pixel pitch
I should get for my 1000mm f/10 fixed aperture telephoto where
decreasing its size wouldn't
produce improvement in quality. Note this is for
theoretical exploration and to better understand
the optical concept of telephoto and sensor sizes
and their optimum matching depending on target
structures.
Hughes
Me
I'd expect a newer Canon dslr sensor to have lower read noise and higher
quantum efficiency than an older Canon DSLR. I'd also expect a higher
end Canon dslr (ie 40d or 50d) to have lower read noise than a digital
rebel model with the same pixel count and density, as that's the way it
seems to have worked in the past when raw data has been analysed. Canon
seem to cheapen out the sensors and support electronics on the lower
models - for normal photography there's actually not a lot of
difference, but there might be for you.
It might be ideal if Canon made a new 6 megapixel APS-c sensor dslr, but
they haven't done so for years now, and sensor technology has moved on
quite a lot over that time - since the 300d. Some terrestrial
photographers would be happy if they did make some lower pixel count
models, but that opens up an argument that's a can of worms.
But I suggest you ask further questions on specific astrophotography
sites. Sorry I can't offer links, but I've seen sites where dslr
performance for astrophotography is tested very thoroughly by
enthusiasts who are very quick to check out new models as soon as they
can get their hands on them.
Hughes
>
> - Show quoted text -
I won't be using it on astrophotography. I'd use it only
terrestrially to image flowers, bees, trees, etc.
So astrophotographers can't answer my questions
because their only concern are the stars and
nothing more. My main question is simply this.
First resolving power is not the same as airy
disc size. Resolving power is smaller. For example,
in my 4" aperture f/10 telephoto. Resolving power
is 1.16 arcsecond while the airy disc is 2.68
arcsecond. Now should my pixel be based
on resolving power or airy disc size?
What's the norm or rule-of-thumb in photography?
Should it be 1/2 the size of the resolving power
or 1/2 the size of the airy disc?? This is simply what
I wanted to know. Astrophotographers mostly haven't viewed
terrestrially so can't answer the question. Now
you photographers and optical specialists might. Thanks.
Hughes
Me
Okay - I don't "get" what you're saying with your airy disk definition.
At f10 with any normal camera lens, the central peak is (only) about
11 or 12 microns, and it seems that this central peak diameter is what
matters most. At 6 um sensel "pixel" size (about 10mp on APS-c) airy
disk diameter of 11 microns on the verge of being diffraction limited,
you'll be fine with a Canon 1000d pixel density, but going over this
only means you get bigger files with little or no resolution gain. Same
as if your effective airy disk diameter is larger, then there mightn't
be any gain from 10 mp over 5, but you don't lose anything except extra
file size. Overall noise performance has improved along with and
despite higher pixel counts so far and in most cases.
You can't get a (new) low cost dslr with close to 11um pixels.
Nikon's 12mp D700 (or Canon's original 5d Mk 1) has the largest pixels,
but 36x24mm sensor, and they are not at the budget end.
Sigma's Foveon sensor dslrs have large pixels (about 5mp and APS-c
size), but crappy (TM) high ISO performance, and that's what you're
probably going to need.
At 1000mm holding it steady enough won't be easy. At longer distances,
atmospheric effects will likely degrade resolution. Forget "optimally
sampled" and go for a model with good low noise high ISO performance and
other features you need. Make sure the model you get has mirror lock-up.
Some basic dslrs don't have this, and at 1000mm mirror slap will be a
real issue.
Martin Brown
>>>> I plan to buy a Canon 1000D. Calculating the pixel
A small amount of oversampling upto about 1.5x makes the image look more
natural. Hard sampling below the lens resolution limit will produce the
photographic equivalent of jaggies in artificially rendered images.
If you expect to be routinely trying to photograph and separate equal
brightness double stars at the diffraction limit then you would need 3
pixels in the 1.16" arc to match the ultimate resolution and see the dip
in brightness between the two peaks. This would almost certainly be
wasteful since mirror lenses tend to be a bit soft and hard to focus.
In practice atmospheric turbulence will prevent you getting even close
unless you go down the webcam and lucky exposure stacking route. Air
turbulence is generally much worse for horizontal long distance
telephoto shots with this type of lens. It will dominate any theoretical
considerations of what the lens might be capable of doing.
>> What's the norm or rule-of-thumb in photography?
>> Should it be 1/2 the size of the resolving power
>> or 1/2 the size of the airy disc?? This is simply what
>> I wanted to know. Astrophotographers mostly haven't viewed
>> terrestrially so can't answer the question. Now
>> you photographers and optical specialists might. Thanks.
The sensors are typically matched to what a realistic lens design can
manage whilst illuminating the entire sensor. It is always a compromise.
Fast lenses are almost never diffraction limited at full aperture and
sharpness improves to an optimum around f8-f11 and then degrades again
as diffraction effects start to limit performance. That is what the MTF
graphs for a lens shows. A quick search gave this site which appears to
explain things reasonably well:
http://www.normankoren.com/Tutorials/MTF1A.html
> Okay - I don't "get" what you're saying with your airy disk definition.
> At f10 with any normal camera lens, the central peak is (only) about 11
> or 12 microns, and it seems that this central peak diameter is what
> matters most. At 6 um sensel "pixel" size (about 10mp on APS-c) airy
> disk diameter of 11 microns on the verge of being diffraction limited,
> you'll be fine with a Canon 1000d pixel density, but going over this
> only means you get bigger files with little or no resolution gain. Same
A modest amount of oversampling is a good thing. Otherwise you can get
Moire fringe effects on periodic structures from beats between the
sensor pitch and the pattern. A small amount of oversampling as a form
of anti-aliassing is useful.
> as if your effective airy disk diameter is larger, then there mightn't
> be any gain from 10 mp over 5, but you don't lose anything except extra
> file size. Overall noise performance has improved along with and
> despite higher pixel counts so far and in most cases.
> You can't get a (new) low cost dslr with close to 11um pixels.
> Nikon's 12mp D700 (or Canon's original 5d Mk 1) has the largest pixels,
> but 36x24mm sensor, and they are not at the budget end.
> Sigma's Foveon sensor dslrs have large pixels (about 5mp and APS-c
> size), but crappy (TM) high ISO performance, and that's what you're
> probably going to need.
> At 1000mm holding it steady enough won't be easy. At longer distances,
> atmospheric effects will likely degrade resolution. Forget "optimally
> sampled" and go for a model with good low noise high ISO performance and
> other features you need. Make sure the model you get has mirror lock-up.
> Some basic dslrs don't have this, and at 1000mm mirror slap will be a
> real issue.
A 1000mm lens is only usable on a very sturdy tripod, and a cable
release or remote operate device helps. Mirror lock up is also useful.
There is someone asking almost the same question again and again on the
astro groups and they never seem to understand the answer.
Regards,
Martin Brown
Savageduck
OK! OK! Slow down.
You are over thinking this and you might hurt yourself. You have stated
your intentions for "terrestrial" photography. You don't have to assume
a pseudo intellectual "astrophotography" position to do what you intend
to do with something long settled.
You obviously have an intellectual curiosity which is going to stifle
your ability to actually enjoy what is relatively simple today. You
are, to put it crudely, going to "mind fuck" yourself out of any
"terrestrial" photographic opportunity.
All you want to do is "image flowers, bees trees, etc." This has been
done successfully since photography went beyond portraiture. Most of
the issues have been dealt with
http://en.wikipedia.org/wiki/Macro_photography
Stop thinking in astronomy terms if all you want to do is something
others have done successfully for years. There are any number of macro
lenses available to give you 1:1 results. There are many publications
to refer to.
That deals with the flowers & bees, as far as the trees and etc. part
of your intended subjects, pick your distance and select a lens. There
are plenty to choose from.
--
Regards,
Savageduck
Hughes
wrote:
How about if I want to see the most details in
the flowers or bees. Should I also use 3 pixels in
the 1.16" arc???
Hughes
> Martin Brown- Hide quoted text -
>
> - Show quoted text -- Hide quoted text -
>
> - Show quoted text -- Hide quoted text -
Me
balance overall sensor efficiency against pixel density. As I've said
elsewhere, the dslr manufacturers have consistently increased overall
sensor performance despite increasing pixel density. Buy something that
they currently make - they know what they're doing, even if there's a
lot of "puffery" involved in marketing dslrs with ever increasing
megapixel counts.
The final objective for a terrestrial photograph is for a print of a
certain size, or an image file to be displayed at a certain size on
screen. Even if "pixel level" sharpness is reduced at higher pixel
densities, after resampling to final display size, the result should be
better. At low pixel densities, there will be a limit to the maximum
size that the image (or a crop from it) can be displayed without
pixelisation (and stair stepping/ aliasing) being visible to the naked
eye. There's no point where increasing pixel size (thus reducing pixel
count) fails to deliver better signal to noise ratio based on "shot
noise", so that fact leads to an inevitable but totally absurd
conclusion that a "one pixel" sensor must be best. Thus the counter
argument, that there's something to be gained by increasing pixel
density (so long as overall quantum efficiency is maintained) appears to
be the correct one. (the only other possibility is that we're currently
at the perfect pixel pitch, and because pixel pitch has decreased over
time, you can happily eliminate that unlikely argument from consideration).
Hughes
Well. 3 pixels for the resolving power of 1.16" arcsecond for the 4"
telephoto is already optimally sampling because to
get 5 pixels or more inside the 1.16" arcsecond is literally
wasting pixels. This is true for astrophotography.
Now when imaging terrestrially where I'm only interested
in the optical principles, does one still need to get 3 pixels
in a 1.16 arcsecond resolving power, or can it be 1 pixel
for 1.16 arcsecond?
Well. You may be confusing the Airy Disc size and resolving
power which is not the same, the resolving power is 2X or even 4X
smaller than the Airy Disc, therefore in Martin
message... he is like saying there are 6-8 pixels in the
airy disc because the 1.16 arcsecond being the resolving
power is 2-3X smaller than the airy disc.
Ultimately. My sole purpose of imaging bees and flowers is
not to print it or display it but just solely to have experimental and
conceptual demonstration of optical
principles. It's like i'm studying optics and just want to
understand the imaging of extended object terrestrially
rather than the moon, planets or the stars.
Hughes
> elsewhere, the dslr manufacturers have consistently increased overall
> sensor performance despite increasing pixel density. Buy something that
> they currently make - they know what they're doing, even if there's a
> lot of "puffery" involved in marketing dslrs with ever increasing
> megapixel counts.
> The final objective for a terrestrial photograph is for a print of a
> certain size, or an image file to be displayed at a certain size on
> screen. Even if "pixel level" sharpness is reduced at higher pixel
> densities, after resampling to final display size, the result should be
> better. At low pixel densities, there will be a limit to the maximum
> size that the image (or a crop from it) can be displayed without
> pixelisation (and stair stepping/ aliasing) being visible to the naked
> eye. There's no point where increasing pixel size (thus reducing pixel
> count) fails to deliver better signal to noise ratio based on "shot
> noise", so that fact leads to an inevitable but totally absurd
> conclusion that a "one pixel" sensor must be best. Thus the counter
> argument, that there's something to be gained by increasing pixel
> density (so long as overall quantum efficiency is maintained) appears to
> be the correct one. (the only other possibility is that we're currently
> at the perfect pixel pitch, and because pixel pitch has decreased over
> time, you can happily eliminate that unlikely argument from consideration).- Hide quoted text -
Chris H
s.com>, Hughes <eugen...@gmail.com> writes
>
>I won't be using it on astrophotography. I'd use it only
>terrestrially to image flowers, bees, trees, etc.
Then you are very much over complicating things I think.
--
\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\
\/\/\/\/\ Chris Hills Staffs England /\/\/\/\/
\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/
Martin Brown
> Hughes wrote:
>> On Apr 23, 11:41 pm, Martin Brown <|||newspam...@nezumi.demon.co.uk>
>> wrote:
>>> Me wrote:
>>>> Hughes wrote:
>>>>> First resolving power is not the same as airy
>>>>> disc size. Resolving power is smaller. For example,
>>>>> in my 4" aperture f/10 telephoto. Resolving power
>>>>> is 1.16 arcsecond while the airy disc is 2.68
>>>>> arcsecond. Now should my pixel be based
>>>>> on resolving power or airy disc size?
>>> If you expect to be routinely trying to photograph and separate equal
>>> brightness double stars at the diffraction limit then you would need 3
>>> pixels in the 1.16" arc to match the ultimate resolution and see the dip
>>
>> How about if I want to see the most details in
>> the flowers or bees. Should I also use 3 pixels in
>> the 1.16" arc???
You probably do not want to be using a 1000mm mirror lens for this.
Something between 100-200mm with a dedicated macro mode would be more
appropriate.
Chances are your lens and photographic technique are nowhere near good
enough to get close to even the 2.6" arc criterion with an f10 mirror
lens. Depth of field issues will drive you nuts.
>>
> No - you should "oversample" and use as many pixels as possible, but
> balance overall sensor efficiency against pixel density. As I've said
Oversampling by more than a factor of 2 has no benefits at all, and
could be deleterious in the event that signal to noise is compromised by
the smaller sensor areas with more silicon lost to the gaps between
cells. Astronomers tend to think about this a lot more and there are
plenty of FAQs the OP could read. The thing he needs to remember about
camera lenses is that they are only close to diffraction limited at
around f8 or slower.
The trend towards ever more insane pixel numbers has run its course.
There are already P&S cameras with 12Mpixel sensors and lenses that
cannot provide the required detail to use them.
At least with DSLRs you can buy top grade lenses that can use all the
available pixels to store real image data.
> elsewhere, the dslr manufacturers have consistently increased overall
> sensor performance despite increasing pixel density. Buy something that
> they currently make - they know what they're doing, even if there's a
> lot of "puffery" involved in marketing dslrs with ever increasing
> megapixel counts.
That is still good advice.
Although I would look very carefully at the astroimaging reviews if use
for deep sky astrophotography is also intended. Not all digital cameras
have long exposure friendly electronics. The ones with warm corners
suffer extra dark current there and in some cases weak IR emission from
the readout electronics.
> The final objective for a terrestrial photograph is for a print of a
> certain size, or an image file to be displayed at a certain size on
> screen. Even if "pixel level" sharpness is reduced at higher pixel
> densities, after resampling to final display size, the result should be
> better. At low pixel densities, there will be a limit to the maximum
> size that the image (or a crop from it) can be displayed without
> pixelisation (and stair stepping/ aliasing) being visible to the naked
> eye. There's no point where increasing pixel size (thus reducing pixel
> count) fails to deliver better signal to noise ratio based on "shot
> noise", so that fact leads to an inevitable but totally absurd
> conclusion that a "one pixel" sensor must be best.
No it doesn't lead to that conclusion at all. It leads to the conclusion
that one pixel for each idependently sampled point across the image is
the optimum in terms of minimum file size against maximum information
content. And you can derive this specification from the theory of
optics. Once you go beyond that sampling theorem limit you are not
gaining any new knowledge by increasing the number of pixels.
But as I said before a practical sweet spot is about 1.5x the
theoretical limit which allows some linear dependence in adjacent
pixels. Some cameras have internal anti-alias screens on the CCD.
A circular aperture cannot measure frequency content in the image finer
than the diffraction limit of 1.22 lambda/D. Anything at higher
frequency will be lost and/or aliased into the measured band.
Some oversampling upto about 1.5x is good because it helps make the
adjacent pixels linearly related and so smooths the edges of sharp
transitions rather than a jagged staircase effect if undersampled.
Thus the counter
> argument, that there's something to be gained by increasing pixel
> density (so long as overall quantum efficiency is maintained) appears to
> be the correct one. (the only other possibility is that we're currently
> at the perfect pixel pitch, and because pixel pitch has decreased over
> time, you can happily eliminate that unlikely argument from consideration).
The only reason there is an advantage to increasing the pixel density is
that marketting men can say "Hey buy our new N+1 *mega* pixel camera".
Regards,
Martin Brown
Hughes
wrote:
> Me wrote:
> > Hughes wrote:
> >> On Apr 23, 11:41 pm, Martin Brown <|||newspam...@nezumi.demon.co.uk>
> >> wrote:
> >>> Me wrote:
> >>>> Hughes wrote:
> >>>>> First resolving power is not the same as airy
> >>>>> disc size. Resolving power is smaller. For example,
> >>>>> in my 4" aperture f/10 telephoto. Resolving power
> >>>>> is 1.16 arcsecond while the airy disc is 2.68
> >>>>> arcsecond. Now should my pixel be based
> >>>>> on resolving power or airy disc size?
> >>> If you expect to be routinely trying to photograph and separate equal
> >>> brightness double stars at the diffraction limit then you would need 3
> >>> pixels in the 1.16" arc to match the ultimate resolution and see the dip
>
> >> How about if I want to see the most details in
> >> the flowers or bees. Should I also use 3 pixels in
> >> the 1.16" arc???
>
> You probably do not want to be using a 1000mm mirror lens for this.
> Something between 100-200mm with a dedicated macro mode would be more
> appropriate.
Well. I'm just studying the theoretical principles involved
and want to know what is the case. Also you can assume
a situation where macro is not good like imaging the
face of a cobra in a zoo. So you think that in terrestrial
imaging using telescopes and CCDs, it is still correct
to use 2 pixels for the resolving power or resolution
in arcsecond of the scope (even if we are not imaging double stars)? I
just want to hear a Yes to confirm my
understanding that the 2 pixels per resolving arcsecond
is a rule for both the sky and terrestrial targets.
>
> Chances are your lens and photographic technique are nowhere near good
> enough to get close to even the 2.6" arc criterion with an f10 mirror
> lens. Depth of field issues will drive you nuts.
I know. It's just to understand optics. What do you mean
depth of field issues can drive one nuts. At f/10,
depth of field is big so you don't have to worry about
focusing.
>
>
> > No - you should "oversample" and use as many pixels as possible, but
> > balance overall sensor efficiency against pixel density. As I've said
>
> Oversampling by more than a factor of 2 has no benefits at all, and
> could be deleterious in the event that signal to noise is compromised by
He meant by more than 2 pixels is the airy disc which
can even have 5 pixels while only 2 pixels is optimum
for the resolving power or resolution in arcsecond.
> Martin Brown- Hide quoted text -
>
> - Show quoted text -
What do you think is the best Webcam in the world
about 5 micron pixel that can be at least 1/2, 3/4
as good as a DSLR like a canon 1000d?? I'm only
after the pixel scale and not field of view (noting
that a webcam and dlsr with similar 5 micron
pixel would have the same pixel scale or resolution
in the target).
Also I saw many astrophotographers using Webcam
when they can use just a cheap dslr like 1000d.
Maybe there is some situation that webcam can
even produce better colors than the dslr??
Hu
Me
as microlenses re-focus light back to the photodiode. At very high
pixel densities (P&S cameras etc) backlit sensors are coming.
terrestrial photography, as (within limits), the closer you look at a
scene, the more you'll see. That might be different for
astrophotography, where sample points are effectively finite (because I
guess that even though faint stars exist between the stars you intend to
image, the signal is below the threshold where you can detect them over
noise, so then they're effectively ignored?). So (forgetting optical
capabilities, diffraction etc) you're limited by availability of
sufficient signal, which may not be the case for terrestrial photography
back on planet earth.
Sorry if that sounds dumb - but astrophotography is foreign to me.
>
> But as I said before a practical sweet spot is about 1.5x the
> theoretical limit which allows some linear dependence in adjacent
> pixels. Some cameras have internal anti-alias screens on the CCD.
>
> A circular aperture cannot measure frequency content in the image finer
> than the diffraction limit of 1.22 lambda/D. Anything at higher
> frequency will be lost and/or aliased into the measured band.
>
> Some oversampling upto about 1.5x is good because it helps make the
> adjacent pixels linearly related and so smooths the edges of sharp
> transitions rather than a jagged staircase effect if undersampled.
>
> Thus the counter
>> argument, that there's something to be gained by increasing pixel
>> density (so long as overall quantum efficiency is maintained) appears
>> to be the correct one. (the only other possibility is that we're
>> currently at the perfect pixel pitch, and because pixel pitch has
>> decreased over time, you can happily eliminate that unlikely argument
>> from consideration).
>
> The only reason there is an advantage to increasing the pixel density is
> that marketting men can say "Hey buy our new N+1 *mega* pixel camera".
>
Many people expect way too much in increased "resolution" than they will
ever get from a few more pixels. Here's a reasonable comparison between
12mp on APS-c format, and 24mp on 35mp format:
http://www.astroweb.no/a900/a700-vs-a900.html
With the conclusion that there's really not very much difference, even
between about a theoretical ~40% increase in linear resolution (OTOH I
understand that about 25% increase is about the minimum of what can be
seen clearly by the naked eye as a difference, unless you're looking at
images taken of converging lines etc designed to highlight resolution
differences. So just a little bit above the minimum that can be seen
seems to correlate nicely with what the writer of the article above
observed.) It also correlates exactly with tests I did with a 6mp and a
12mp dslr.
Now proponents for the extra pixels and larger format would say that
it's worth it because of the small incremental gain. But another way of
looking at it (yes - perhaps a cheapskates viewpoint - because larger
formats cost much more as a system) is to say that if 12mp truly isn't
enough, then it's fairly unlikely that 24mp will be enough either.
> Regards,
> Martin Brown
Rich
ebfde7...@c18g2000prh.googlegroups.com:
> Hi,
>
> My experience is mostly on astronomical telescopes
> and I now want to view terrestrial daytime subject
> like sceneries or trees/nature.
>
> I have a 4" 1000mm F/10 Telephoto. I want to get
> a digicam that can produce the optimum image
> quality.
If that's one of those old white Meade SCT's you've got (2045's), the
optics aren't good enough to be worrying about the camera. Likely 6
megapixels would be sufficient.