New Canon EIS mirrorless system - Four Thirds, but not Four Thirds!

"Doug McDonald" wrote:
On 9/20/2010 9:23 AM, Bruce wrote:

Instead, they end up spending much more money on a camera they cannot
use which also gives them a lot of problems, due to the DSLR's
comparatively very restricted shallow depth of field.

HUH??? dSLRs have adjustable diaphragms. Just increase the f-number to
suit your taste.

If you have the same angular field of view and the same same number
of megapixels you will get the exact same image as a P&S with the same
depth of field ... including the same diffraction problems and the
same noise problems.

Exactly. The extra stopping down you need to do on the larger format forces
you to use a higher ISO (to get the same shutter speed), at which point the
images are identical in all aspects.

What the dSLR allows you to do is (a) put up with longer shutter speeds to
get lower noise/higher DR at the same DoF, and (b) get worlds better
sharpness at the plane of focus in narrower DoF images, and (c) get truly
shallow DoF images.

But that assumes the same pixel count. Normally, you split the difference
and go for a higher pixel count. At which point (if you want the sharpness
implicit in the higher pixel count) you do lose DoF.

In general, larger format cameras are more awkward, harder to use, and a
pain in the butt, but the IQ is worth it.

--
David J. Littleboy
Tokyo, Japan

In article , Bruce
wrote:

For a camera user who wants to improve their photography, one of the
best investments they could make is to buy some tuition.

Instead, they end up spending much more money on a camera they cannot
use which also gives them a lot of problems, due to the DSLR's
comparatively very restricted shallow depth of field.

Sheesh. You need to follow your own advice about buying some
tuition.....

--
Charles

On 20/09/2010 17:03, SMS wrote:
On 9/20/2010 5:41 AM, DanP wrote:
On Sep 14, 12:46 am, wrote:

Probably every D-SLR owner has at least a couple of P&S digital cameras
as well, so they're well aware of the advantages of the D-SLR.

Even more, there are DSLR owners who had a P&S first and felt a need
for something better.
Me included.

Very true. There are very specific areas where a D-SLR is indispensable:

-Low light, where a larger sensor with larger pixels are required.

-High ISO, where a larger sensor with larger pixels are required for
lower noise.

-Action shots where fast AF is required

-Shots where long telephotos or extreme wide angle is required (those
adapter lenses for point and shoot cameras range from abysmal to mediocre).


I'll add one to this:

- Good quality pictures


--
Bertrand

David J. Littleboy wrote:

But that assumes the same pixel count. Normally, you split the difference
and go for a higher pixel count. At which point (if you want the sharpness
implicit in the higher pixel count) you do lose DoF.

Pixel count is quite irrelevant, interesting is only the final
result. Unless the final result is always a 100% view of the
image, regardless of the megapixel size and regardless that one
will see only a tiny bit of the whole image. Hence the final
result is usually:
- a print of a given size (be it 4x6 inch or "fills the whole
wall")
- an image on the web of a given size (usually small, not more
than 2 MPix)
- an image on the monitor (usually small, usually not more than
2 or 3 MPix)

It's easy to show that
a) a 20MPix sensor of a given type and state of the art has
more per pixel noise than a 5MPix sensor of the same type
and state of the art
b) a print from said 20MPix sensor and 5MPix sensor is
quite indistinguishable[1], as long as the print size
doesn't cause pixelation in the 5MPix sensor's case.
That assumes no overly drastic 'noise removal' in the
20MPix sensor's case.

Hence the inherent higher possible sharpness only matters when
we come to prints where less MPixels pixelize already.

In general, larger format cameras are more awkward, harder to use, and a
pain in the butt, but the IQ is worth it.

Try using a P&S for fast sports, then complain about the DSLR
being harder to use. (OK, for even larger formats than 35mm you
are probably right about the harder to use.)

-Wolfgang

[1] There's a bit more read noise in the 20MPix version, but
except for extreme circumstances that won't matter
visibly, as read noise rarely becomes visible.

"Wolfgang Weisselberg" wrote in message
...
David J. Littleboy wrote:

But that assumes the same pixel count. Normally, you split the difference
and go for a higher pixel count. At which point (if you want the
sharpness
implicit in the higher pixel count) you do lose DoF.

Pixel count is quite irrelevant, interesting is only the final
result. Unless the final result is always a 100% view of the
image, regardless of the megapixel size and regardless that one
will see only a tiny bit of the whole image.

In ivory-tower theory, yes. But in real life, when you have a 21MP sensor,
you require 1.3 times the resolution than you do when you have a 12MP
sensor. If you didn't want larger/better prints, you wouldn't have upgraded.

You forget that some of us actually make prints.

And the DoF goes down due to the smaller CoC you now require.

Hence the final
result is usually:
- a print of a given size (be it 4x6 inch or "fills the whole
wall")
- an image on the web of a given size (usually small, not more
than 2 MPix)
- an image on the monitor (usually small, usually not more than
2 or 3 MPix)

It's easy to show that
a) a 20MPix sensor of a given type and state of the art has
more per pixel noise than a 5MPix sensor of the same type
and state of the art

But the "same type and state of the art is rarely true". The 12 and 21MP
cameras here are three years apart in technology, and have very similar
pixel noise.

Also, at base ISO, for most dSLRs, it's not the shot noise but the analog
circuit noise that's the limiting factor, so dSLR noise at ISO 100 is pretty
much the same (10 or 11 stops of DR in RAW) for nearly every dSLR currently
made. And some P&S cameras get surprisingly close at ISO 50.

b) a print from said 20MPix sensor and 5MPix sensor is
quite indistinguishable[1], as long as the print size
doesn't cause pixelation in the 5MPix sensor's case.
That assumes no overly drastic 'noise removal' in the
20MPix sensor's case.

At the standard print size here, 12x18, 21MP is significantly better than
12MP. For landscape/cityscape work, the difference is appreciated.

Hence the inherent higher possible sharpness only matters when
we come to prints where less MPixels pixelize already.

It's not pixelization, it's improvement in detail and texture rendition.
"Pixelation" is only a problem with insanely stupid upsampling. Bicubic
smoother in Photoshop, or any of the even better methods in Qimage, to say
nothing of Genuine Fractals, all produce pixelation-free images at any size.

In general, larger format cameras are more awkward, harder to use, and a
pain in the butt, but the IQ is worth it.

Try using a P&S for fast sports, then complain about the DSLR
being harder to use. (OK, for even larger formats than 35mm you
are probably right about the harder to use.)

That's because P&S cameras are slower than molasses; if someone bothered to
make a P&S superzoom that had decent AF, it'd get used for sports and news
reporting, where the final image is 200x300 pixels or smaller.

--
David J. Littleboy
Tokyo, Japan

David J. Littleboy wrote:
"Wolfgang Weisselberg" wrote in message
David J. Littleboy wrote:

But that assumes the same pixel count. Normally, you split the difference
and go for a higher pixel count. At which point (if you want the
sharpness
implicit in the higher pixel count) you do lose DoF.

Pixel count is quite irrelevant, interesting is only the final
result. Unless the final result is always a 100% view of the
image, regardless of the megapixel size and regardless that one
will see only a tiny bit of the whole image.

In ivory-tower theory, yes.

In real life the final result is interesting, not only in theory.In
real life the final result is interesting, not only in theory.

But in real life, when you have a 21MP sensor,
you require 1.3 times the resolution than you do when you have a 12MP
sensor.

Not if you print with essentially 12 MPix.

Additionally, the resolution is determined by the sensor and the
lens, with better sensors improving resolution in all cases (OK,
not usefully with coke glass bottle bottoms).

If you didn't want larger/better prints, you wouldn't have upgraded.

I've got a 20D and I would like to upgrade to a 5DII --- not
for the added pixels, but for the increased ISO capabilities.
Thus your statement is incomplete.

You forget that some of us actually make prints.

If that's the final result, that matters.

And the DoF goes down due to the smaller CoC you now require.

Nope. CoC is defined by the enlargement and the viewing distance.
If you insist to print larger but keep the viewing distance
the same, you get smaller CoCs, but they are not inherent to
more pixels.

Hence the final
result is usually:
- a print of a given size (be it 4x6 inch or "fills the whole
wall")
- an image on the web of a given size (usually small, not more
than 2 MPix)
- an image on the monitor (usually small, usually not more than
2 or 3 MPix)

It's easy to show that
a) a 20MPix sensor of a given type and state of the art has
more per pixel noise than a 5MPix sensor of the same type
and state of the art

But the "same type and state of the art is rarely true". The 12 and 21MP
cameras here are three years apart in technology, and have very similar
pixel noise.

That, however, is another problem.

Also, at base ISO, for most dSLRs, it's not the shot noise but the analog
circuit noise that's the limiting factor, so dSLR noise at ISO 100 is pretty
much the same (10 or 11 stops of DR in RAW) for nearly every dSLR currently
made. And some P&S cameras get surprisingly close at ISO 50.

Pixel noise or whole image noise? And yes, under ideal
circumstances a P&S can do impressive things, but we don't buy
DSLRs because we have ideal P&S circumstances in which we shoot.

b) a print from said 20MPix sensor and 5MPix sensor is
quite indistinguishable[1], as long as the print size
doesn't cause pixelation in the 5MPix sensor's case.
That assumes no overly drastic 'noise removal' in the
20MPix sensor's case.

At the standard print size here, 12x18, 21MP is significantly better than
12MP. For landscape/cityscape work, the difference is appreciated.

That's fine, then use 21MPix. Or switch to 60MPix :-)

12x18x300²=19.4M
You'd probably gain something up to 27.5MPix (thank bayer)
but nothing above at 300ppi.

Hence the inherent higher possible sharpness only matters when
we come to prints where less MPixels pixelize already.

It's not pixelization, it's improvement in detail and texture rendition.
"Pixelation" is only a problem with insanely stupid upsampling. Bicubic
smoother in Photoshop, or any of the even better methods in Qimage, to say
nothing of Genuine Fractals, all produce pixelation-free images at any size.

Maybe pixelation wasn't the best word. You nailed what I
meant.

In general, larger format cameras are more awkward, harder to use, and a
pain in the butt, but the IQ is worth it.

Try using a P&S for fast sports, then complain about the DSLR
being harder to use. (OK, for even larger formats than 35mm you
are probably right about the harder to use.)

That's because P&S cameras are slower than molasses; if someone bothered to
make a P&S superzoom that had decent AF,

If pigs had wings ...

it'd get used for sports and news
reporting, where the final image is 200x300 pixels or smaller.

.... they'd fly.

-Wolfgang