Megapixels and the cost of geometric progression

There is a nice graph buried in the D200 review at dpreview (may show
elsewhere, I don't know). It shows the relative output size of an image
(all other things being equal) at a variety of megapixel counts from 8
up to 16.7. It clearly demonstrates the dwindling returns of pixel
growth to output quality. eg: to double the size of a print in both
dimensions, you need 4x the pixels. Stated otherwise: from the 8.2 Mpix
of the 20D to the 16.7mpix of the 1Ds Mk II, you only get 1.4x larger
prints. 1.4x at 5X the price (20D to 1Ds Mk II).

http://www.dpreview.com/articles/Nik...mpcompared.gif

This has been stated here often, but the graph illustrates it nicely.

Cheers,
Alan.

--
-- r.p.e.35mm user resource: http://www.aliasimages.com/rpe35mmur.htm
-- r.p.d.slr-systems: http://www.aliasimages.com/rpdslrsysur.htm
-- [SI] gallery & rulz: http://www.pbase.com/shootin
-- e-meil: Remove FreeLunch.

Alan Browne wrote:
There is a nice graph buried in the D200 review at dpreview (may show
elsewhere, I don't know). It shows the relative output size of an image
(all other things being equal) at a variety of megapixel counts from 8
up to 16.7. It clearly demonstrates the dwindling returns of pixel
growth to output quality. eg: to double the size of a print in both
dimensions, you need 4x the pixels. Stated otherwise: from the 8.2 Mpix
of the 20D to the 16.7mpix of the 1Ds Mk II, you only get 1.4x larger

That is why I think you really want something like 80MP.

Scott

This is why the upgrade from a 6mp to an 8mp dSLR makes absolutely no sense
for Canon owners and the uprade from a D70 to its successor, presuming it
will have the 10mp sony sensor in the D200 makes marginal sense.
If one factors in the rarity with which prints larger than letter size are
ever made by most users absolute pixel counts do not necessarily constitute
major sensor improvements once you arrive at the 6mp level.
However the more pixel information in the image the more options there are
for image manipulation and cropping and this is the major, practical reason
why increasing pixel counts and increasing sensor sizes matter.

bmoag wrote:


However the more pixel information in the image the
more options there are for image manipulation and cropping and this
is the major, practical reason why increasing pixel counts and
increasing sensor sizes matter.

Agreed. However most people compose for most of the viewfinder to be
printed. Not to make crops.


--
-- r.p.e.35mm user resource: http://www.aliasimages.com/rpe35mmur.htm
-- r.p.d.slr-systems: http://www.aliasimages.com/rpdslrsysur.htm
-- [SI] gallery & rulz: http://www.pbase.com/shootin
-- e-meil: Remove FreeLunch.

Alan Browne wrote:
bmoag wrote:

However the more pixel information in the image the
more options there are for image manipulation and cropping and this
is the major, practical reason why increasing pixel counts and
increasing sensor sizes matter.

Agreed. However most people compose for most of the viewfinder to be
printed. Not to make crops.

You'r probably right, but it depends heavily on subject. Bird photogs
rarely have the luxury of composing in the viewfinder - in particular
if they specialise in shy passerines.

And it occurs at least to me from time to time that I realise that the
Really Interesting Image was something that I didn't compose for, and
which occurs in only a part of the frame.

Jan Böhme

On Mon, 02 Jan 2006 16:48:01 -0500, Alan Browne
wrote:

bmoag wrote:


However the more pixel information in the image the
more options there are for image manipulation and cropping and this
is the major, practical reason why increasing pixel counts and
increasing sensor sizes matter.

Agreed. However most people compose for most of the viewfinder to be
printed. Not to make crops.

Not because they know how to compose, because they don't know
enough to crop the junk.
-Rich

"bmoag" wrote in message
. com...
However the more pixel information in the image the more options there are
for image manipulation and cropping and this is the major, practical
reason why increasing pixel counts and increasing sensor sizes matter.

Nah its cos punters will pay for more megapixels. Its also why £50 with 6mp
on the box sell. Interpolation is also used.

As for Dslrs lower latencies, lower noise, faster frame rates and larger
buffers are the improvements to watch for. Better construction, smaller,
lighter and longer battery life will be other advances used to tempt buyers
and upgraders. You will find as battery efficiency and camera efficiency
improves rather than longer lasting the batteries will get smaller. So you
will always need to buy a 2nd battery rather than wind up with a large
battery that lasts for thousands of shots. The claim will be to save weight
and to make the total body smaller.

Canon execs last year said they could make a 100mp camera tomorrow but
nobody has affordable computing power to process and then there is the cost
of storage.

ian lincoln wrote:
"bmoag" wrote in message
. com...
However the more pixel information in the image the more options there are
for image manipulation and cropping and this is the major, practical
reason why increasing pixel counts and increasing sensor sizes matter.

Nah its cos punters will pay for more megapixels. Its also why £50 with 6mp
on the box sell. Interpolation is also used.

As for Dslrs lower latencies, lower noise, faster frame rates and larger
buffers are the improvements to watch for. Better construction, smaller,
lighter and longer battery life will be other advances used to tempt buyers
and upgraders. You will find as battery efficiency and camera efficiency
improves rather than longer lasting the batteries will get smaller. So you
will always need to buy a 2nd battery rather than wind up with a large
battery that lasts for thousands of shots. The claim will be to save weight
and to make the total body smaller.

Canon execs last year said they could make a 100mp camera tomorrow but
nobody has affordable computing power to process and then there is the cost
of storage.

I wrok with 100+ MP images all the time, it does not take all that much
computer power to deal with and the storage is not all that bad either.
figure a 100 MP photo takes 300 MB to store, at about $0.50 / GB this
works out to around $0.15 to store the photo, not all that bad.

The real problem with a 100 MP camera is the lens to make full use of
it is going to be pretty costly.

Scott

In message . com,
"Scott W" wrote:

The real problem with a 100 MP camera is the lens to make full use of
it is going to be pretty costly.

I don't know about that. My Tamron 90mm Di macro, and my Canon 300mm
f4L IS can both deliver standard pixel-sharpness with my 20D, with a 2x
TC.

8.2 * 1.6 * 1.6 * 2 * 2 = 84MP, full-frame.
--


John P Sheehy

ian lincoln wrote:

Canon execs last year said they could make a 100mp camera tomorrow but
nobody has affordable computing power to process and then there is the cost
of storage.

They could make a 100MP sensor, and stick it in a camera, but (in 35mm
format) it wouldn't be of any benefit.

Megapixels aren't everything. We've already surpassed the quality of
35mm film; I really think it's time to stop relying on those silly
numbers and start looking at things like dynamic range.

--
Jeremy |