18 megapixels on a 1.6x crop camera - Has Canon gone too far?

"Alfred Molon" wrote in message
...
In article , Neil
Harrington says...

I can think of just three kinds of people likely to find advantages in
"full
frame," and I'm not any of them.
snip
2. Someone who wants to make humungous prints and/or at very high ISOs to
put on a wall that people will walk up to and look at from an unusually
close distance. (I am not likely ever to do that.)

This guy thinks that you need 80MP for a highest quality 8x10 print:
http://tinyurl.com/c4dj5n

Very interesting article, thanks. Actually he says 80 isn't enough; you need
about 180 million pixels for highest quality (30 lp/mm) on an 8 x 10 print.

His standard for highest quality -- no visible difference at double the
resolution -- probably does make reaching such a level of quality virtually
impossible. I think human perception of sharpness is a very, very
complicated subject. I remember seeing a full-page magazine reproduction of
an 8 x 10 Kodachrome that was definitely much sharper looking than any 35mm
or even medium format shot would have been -- and that even though it was
printed in the magazine's regular halftone dots, of course.

In article , Alfred
Molon wrote:

This guy thinks that you need 80MP for a highest quality 8x10 print:
http://tinyurl.com/c4dj5n

what a load of rubbish. humans can't see 80 mp on an 8x10 print. he
even says he's guessing at the number.

80 mp is roughly 11000 x 7300 pixels (3:2 format), which would be 1100
pixels per inch on an 8x10" print. human vision can see to about 600
ppi at normal viewing distances and that's under ideal conditions. and
good luck trying to print 1100 ppi, most printers don't go that high.

he also says bayer is 'substantially less than 50% efficient' which is
totally wrong. like any decent sampling system, it is accurate to about
70-80% of nyquist.

Alfred Molon wrote:
In article , Neil
Harrington says...

I can think of just three kinds of people likely to find advantages in "full
frame," and I'm not any of them.
snip
2. Someone who wants to make humungous prints and/or at very high ISOs to
put on a wall that people will walk up to and look at from an unusually
close distance. (I am not likely ever to do that.)

This guy thinks that you need 80MP for a highest quality 8x10 print:
http://tinyurl.com/c4dj5n

Only achievable with a medium format size sensor.

As at least one other poster has said, this exceeds the resolution of
the human eye even if you were looking at the print close up with good
close in vision.

Back out of it this way: humans generally cannot tell the difference
between a 200 dpi and 300 dpi print most of the time. For B&W, _maybe_
400 dpi.

@400 dpi that would be: 400^2 * 12 * 8 = 15.4 Mpix.

That is if the original image was tack sharp in the focus plane, no
vibration, etc.

Even 600 dpi (of real detail) would be 35 Mpix.

Chuck this 80 mpix nonsense.

DRS wrote:
"Alfred Molon" wrote in message

In article ,
Giftzwerg says...
What's so *magic* about a
full-frame sensor that a body housing one can't cost $1,200?
Perhaps larger sensor being more expensive to make?

Why would a larger sensor with lower pixel density be more expensive to
make?

What Bader said.

Alan Browne wrote:
Alfred Molon wrote:
In article , Neil
Harrington says...

I can think of just three kinds of people likely to find advantages
in "full frame," and I'm not any of them.
snip
2. Someone who wants to make humungous prints and/or at very high
ISOs to put on a wall that people will walk up to and look at from an
unusually close distance. (I am not likely ever to do that.)

This guy thinks that you need 80MP for a highest quality 8x10 print:
http://tinyurl.com/c4dj5n

Only achievable with a medium format size sensor.

As at least one other poster has said, this exceeds the resolution of
the human eye even if you were looking at the print close up with good
close in vision.

Back out of it this way: humans generally cannot tell the difference
between a 200 dpi and 300 dpi print most of the time. For B&W, _maybe_
400 dpi.

@400 dpi that would be: 400^2 * 12 * 8 = 15.4 Mpix.

That is if the original image was tack sharp in the focus plane, no
vibration, etc.

Even 600 dpi (of real detail) would be 35 Mpix.

Chuck this 80 mpix nonsense.

Dammit, Alan, change your dpi above to ppi and it makes sense.
Images have pixels; printers put down dots, or droplets.
--
john mcwilliams

nospam wrote:
In article , Alfred
Molon wrote:

This guy thinks that you need 80MP for a highest quality 8x10 print:
http://tinyurl.com/c4dj5n

what a load of rubbish. humans can't see 80 mp on an 8x10 print. he
even says he's guessing at the number.


I can tell the difference between a 12 gauge top E guitar string and a
16 gauge B lying on the floor from standing. That's 4 1000ths of a inch
clearly defined from over five feet away. I can tell a 9 from a 10 at
the sort of distance a 10 x 8 would be viewed at, meaning I can see
1000th of an inch difference.

That is 80 megapixel resolution.

But although I have printers with over 1000 'droplets per inch' they
still seem to be around 300lpi in real terms, so it's all academic. But
if there was a 1000dpi printer (an actual 1000dpi, not an Epson doing
2880dpi but really forming a 240lpi image) I would be able to see the
difference.

David

On Sun, 6 Sep 2009 11:09:38 +0200, Alfred Molon
wrote:

In article , Alan Browne
says...

As at least one other poster has said, this exceeds the resolution of
the human eye even if you were looking at the print close up with good
close in vision.

Unlike many of the (anonymous) posters in this ng, this guy is a pro and
has some credibility. See what he writes:
"...if you put a 15 lp/mm print next to a 30 lp/mm print, a high
percentage of viewers will select the 30 lp/m print as being sharper..."

Yes, because the content of the photography from someone like that, the
only thing that's different from one image to the next is the "sharpness",
so that's all they'll ever notice about that quality of his photography.
Marvelously sharp garbage. If the photo itself was worth appreciating they
wouldn't be looking for any sharpness differences. Someone like that is a
tech-head hack, not any kind of photographer.


This must be based on some empirical test.

And
"How many pixels is 30 lp/mm in an 8 x 10-inch print? About 180 million.
"

Besides, don't forget that Bayer sensors have less effective resolution
than their nominal one.

All this basically means that while 10-20MP are sufficient for most
purposes, there is considerable room for image quality improvement.

David Kilpatrick wrote:


Human hair is around 1/1000th of an inch. I don't have much trouble
spotting a single stray hair at 5ft, but it's easier for me around 3ft.
Still doesn't mean I manage to spot stray hairs... Photoshop required!

Perhaps revise your maths and methodology.
Head hair might might be as thin as 2/1000 inch, but more typically may
be 4/1000 inch but up to 1/100 inch.
Your sampling method is also flawed as you see the thicker hairs at 3-5
feet, but the ones you can't see aren't included in your analysis.

Alfred Molon wrote:
In article , Alan Browne
says...

As at least one other poster has said, this exceeds the resolution of
the human eye even if you were looking at the print close up with good
close in vision.

Unlike many of the (anonymous) posters in this ng, this guy is a pro and
has some credibility. See what he writes:
"...if you put a 15 lp/mm print next to a 30 lp/mm print, a high
percentage of viewers will select the 30 lp/m print as being sharper..."

This must be based on some empirical test.

Ah, faith is a wonderful thing. Too bad it has no bearing on real life.


And
"How many pixels is 30 lp/mm in an 8 x 10-inch print? About 180 million.
"

Besides, don't forget that Bayer sensors have less effective resolution
than their nominal one.

All this basically means that while 10-20MP are sufficient for most
purposes, there is considerable room for image quality improvement.


You've latched onto a falsehood to call your own. Not really my problem
what you believe.

David Kilpatrick wrote:
nospam wrote:
In article , Alfred
Molon wrote:

This guy thinks that you need 80MP for a highest quality 8x10 print:
http://tinyurl.com/c4dj5n

what a load of rubbish. humans can't see 80 mp on an 8x10 print. he
even says he's guessing at the number.


I can tell the difference between a 12 gauge top E guitar string and a
16 gauge B lying on the floor from standing. That's 4 1000ths of a inch
clearly defined from over five feet away. I can tell a 9 from a 10 at
the sort of distance a 10 x 8 would be viewed at, meaning I can see
1000th of an inch difference.

That is 80 megapixel resolution.

But although I have printers with over 1000 'droplets per inch' they
still seem to be around 300lpi in real terms, so it's all academic. But
if there was a 1000dpi printer (an actual 1000dpi, not an Epson doing
2880dpi but really forming a 240lpi image) I would be able to see the
difference.

Not in real detail in a real image about a real subject.