Where are the BEST Point and Shoot Photos ?

John Navas wrote:
On Sun, 25 Nov 2007 13:36:02 -1000, Scott W wrote
in :

John Navas wrote:

We just differ on what constitutes a sharp print. That doesn't bother
me, but for some reason it seems to really bother you -- why?

You are the one who started a thread on what ppi is needed for a sharp
print, not me.

True, but not this thread, so it's clearly something that's really
bugging you.

I am just taking issue with your numbers, if your
numbers work for you good, but you are putting them out telling all of
us that these are the right numbers.

For most people they are the right numbers, but not for all people,
including you. Fair enough?

It's not just people who differ in resolution requirements, but
subjects. For example, when people look at a portrait of a face they
usually want to view it at a distance where their eyes can comfortably
accomodate the whole face. But we often look at landscapes
differently. If, like many you like in a landscape view to be able to
roam visually around inspecting small details of interest while in the
edges of your vision getting the whole sense of a large encompassing
spatial perspective, you'll move in quite a bit closer for a wider
angle of subtended view than for a facial portrait. Hence the detail
resolution requirements will be proportionally higher.

--
Chris Malcolm DoD #205
IPAB, Informatics, JCMB, King's Buildings, Edinburgh, EH9 3JZ, UK
[http://www.dai.ed.ac.uk/homes/cam/]

John Navas writes:

And Dave's other figures are quite correct, and very well explained.
The posted link is indeed in error by a factor of more than two.

The sources used in the web page I posted are real science, and human
eyes aren't getting better.

Eyes aren't getting better. But in order for "real science" to be
applied to print resolution, the numbers need to be interpreted
correctly. There's at least one error on the web page you cited.

If you look for references on the human eye's resolution limit, you will
find values that range from 35-60 cycles per degree at least. A few
days ago I posted a simple experiment that lets someone measure the
resolution limit of their own eyes. When I do that, I get a value of 45
cycles/degree for me - that doesn't apply to anyone else, but it does
mean that the measurement method isn't wildly inaccurate either.

On the other hand, Roger's page
http://www.clarkvision.com/imagedetail/eye-resolution.html quotes a
value of 0.59 minute per line pair, which is over 100 cycles/degree.
That's above what I've seen in vision textbooks, but Roger refers to
original measurement papers for that.

Now, when you've decided what resolution limit you're going to use in
cycles/degree, you need to convert that to pixels. In theory, you need
at least 2 pixels per cycle, so at a minimum you need to double the
cycles/degree figures to get equivalent pixels/degree.

The page that John refers to describes its resolution as the spacing
between black bars in a resolution chart, which is equivalent to cycles.
But the numbers it quotes are 1 arc minute (60 cycles/degree) in normal
light and best resolution of 30 arc seconds (120 cycles/degree). This
is twice as good as any values I've seen anywhere else. And the web
page then goes on to treat these numbers as applying to pixels, not
cycles, without multiplying by 2.

All of which suggests that the numbers really apply to pixels in the
first place, so peak resolution is meant to be 30 arc seconds *per
pixel* and 60 cycles per degree. In a sense, the page has correct
numbers but the wrong explanation, or perhaps two errors that cancel
each other.

However, having said that peak resolution is one value under good
conditions, the page then uses a lower resolution of half that (and one
third of Roger's value) for the rest of the calculations. The
justification seems to be that resolution is less under "normal"
conditions - but I don't view prints under normal lighting, I view them
under good lighting, at least when looking at them critically. So I
don't agree to the factor-of-two loss of resolution (and, in fact, my
own test gives a resolution limit for my eyes that's 1.5 times better
than the page's "normal", but 3/4 of the "best".)

Then, all this applies to the case where you can actually create one
cycle with only 2 pixels. That works in test patterns and in
computer-generated images, but not images digitized by cameras.
Real cameras get 70-80% of the theoretical limit of resolution, so
realistically resolving one cycle takes 2.5-3 pixels. So when
converting from cycles/degree to pixels/degree, you need to multiply by
this larger factor.

The above depends only only a few facts, though actual values seem to
vary greatly with source:

1. The human eye resolution limit:
http://www.clarkvision.com/imagedetail/eye-resolution.html
http://en.wikipedia.org/wiki/Human_eye#Acuity

2. One cycle needs at least 2 pixels to create it (Nyquist limit of 0.5
cycles/pixel) (See any signal processing text).

3. Real digital cameras can't resolve 0.5 cycles/pixel, they achieve
0.35-0.4 cycles/pixel. (Look at any digital camera resolution test on a
review site, e.g. dpreview, and convert lines per picture height into
cycles/pixel).

Dave

On 26 Nov 2007 10:46:36 GMT, Chris Malcolm wrote
in :

John Navas wrote:

For most people they are the right numbers, but not for all people,
including you. Fair enough?

It's not just people who differ in resolution requirements, but
subjects.

So much for trying to find reasonable middle ground. [sigh]
Oh well. So be it.

For example, when people look at a portrait of a face they
usually want to view it at a distance where their eyes can comfortably
accomodate the whole face. But we often look at landscapes
differently.

Who is "we"?

If, like many you like in a landscape view to be able to
roam visually around inspecting small details of interest while in the
edges of your vision getting the whole sense of a large encompassing
spatial perspective, you'll move in quite a bit closer for a wider
angle of subtended view than for a facial portrait. Hence the detail
resolution requirements will be proportionally higher.

Sorry, but I don't think that necessarily follows.

I frankly think this thing about moving in closer is mostly a way of
justifying expensive higher resolution digital cameras. Funny how we
didn't hear about it when the best digital cameras were around 3 MP.

--
Best regards,
John Navas
Panasonic DMC-FZ8 (and several others)

John Navas writes:

Sorry, but I don't think that necessarily follows.

I frankly think this thing about moving in closer is mostly a way of
justifying expensive higher resolution digital cameras. Funny how we
didn't hear about it when the best digital cameras were around 3 MP.

10 years ago, the debate was between the people who thought that 35 mm
film provided enough resolution for landscapes, the ones who thought you
needed medium format at least, and the ones who thought anything less
than 4x5 was a waste of time. And for pretty much the same reasons, too:
you can make a 4x6 *foot* print from large format and look at its
detail from 15 inches away, and some people like that. Others think
that viewing from 1.5 times the diagonal is all you need.

The cameras involved have changed, but not the argument - because people
differ.

Dave

Isn't it interesting (and highly amusing) that this person's whole sum of
experience with the subject was found on other's web-pages. That's a glaring
clue right there.



On Mon, 26 Nov 2007 20:20:20 +0000 (UTC), (Dave Martindale)
wrote:

John Navas writes:

And Dave's other figures are quite correct, and very well explained.
The posted link is indeed in error by a factor of more than two.

The sources used in the web page I posted are real science, and human
eyes aren't getting better.

Eyes aren't getting better. But in order for "real science" to be
applied to print resolution, the numbers need to be interpreted
correctly. There's at least one error on the web page you cited.

If you look for references on the human eye's resolution limit, you will
find values that range from 35-60 cycles per degree at least. A few
days ago I posted a simple experiment that lets someone measure the
resolution limit of their own eyes. When I do that, I get a value of 45
cycles/degree for me - that doesn't apply to anyone else, but it does
mean that the measurement method isn't wildly inaccurate either.

On the other hand, Roger's page
http://www.clarkvision.com/imagedetail/eye-resolution.html quotes a
value of 0.59 minute per line pair, which is over 100 cycles/degree.
That's above what I've seen in vision textbooks, but Roger refers to
original measurement papers for that.

Now, when you've decided what resolution limit you're going to use in
cycles/degree, you need to convert that to pixels. In theory, you need
at least 2 pixels per cycle, so at a minimum you need to double the
cycles/degree figures to get equivalent pixels/degree.

The page that John refers to describes its resolution as the spacing
between black bars in a resolution chart, which is equivalent to cycles.
But the numbers it quotes are 1 arc minute (60 cycles/degree) in normal
light and best resolution of 30 arc seconds (120 cycles/degree). This
is twice as good as any values I've seen anywhere else. And the web
page then goes on to treat these numbers as applying to pixels, not
cycles, without multiplying by 2.

All of which suggests that the numbers really apply to pixels in the
first place, so peak resolution is meant to be 30 arc seconds *per
pixel* and 60 cycles per degree. In a sense, the page has correct
numbers but the wrong explanation, or perhaps two errors that cancel
each other.

However, having said that peak resolution is one value under good
conditions, the page then uses a lower resolution of half that (and one
third of Roger's value) for the rest of the calculations. The
justification seems to be that resolution is less under "normal"
conditions - but I don't view prints under normal lighting, I view them
under good lighting, at least when looking at them critically. So I
don't agree to the factor-of-two loss of resolution (and, in fact, my
own test gives a resolution limit for my eyes that's 1.5 times better
than the page's "normal", but 3/4 of the "best".)

Then, all this applies to the case where you can actually create one
cycle with only 2 pixels. That works in test patterns and in
computer-generated images, but not images digitized by cameras.
Real cameras get 70-80% of the theoretical limit of resolution, so
realistically resolving one cycle takes 2.5-3 pixels. So when
converting from cycles/degree to pixels/degree, you need to multiply by
this larger factor.

The above depends only only a few facts, though actual values seem to
vary greatly with source:

1. The human eye resolution limit:
http://www.clarkvision.com/imagedetail/eye-resolution.html
http://en.wikipedia.org/wiki/Human_eye#Acuity

2. One cycle needs at least 2 pixels to create it (Nyquist limit of 0.5
cycles/pixel) (See any signal processing text).

3. Real digital cameras can't resolve 0.5 cycles/pixel, they achieve
0.35-0.4 cycles/pixel. (Look at any digital camera resolution test on a
review site, e.g. dpreview, and convert lines per picture height into
cycles/pixel).

Dave

John Navas wrote:
On 26 Nov 2007 10:46:36 GMT, Chris Malcolm wrote
in :

John Navas wrote:

For most people they are the right numbers, but not for all people,
including you. Fair enough?

It's not just people who differ in resolution requirements, but
subjects.

So much for trying to find reasonable middle ground. [sigh]
Oh well. So be it.

For example, when people look at a portrait of a face they
usually want to view it at a distance where their eyes can comfortably
accomodate the whole face. But we often look at landscapes
differently.

Who is "we"?

If, like many you like in a landscape view to be able to
roam visually around inspecting small details of interest while in the
edges of your vision getting the whole sense of a large encompassing
spatial perspective, you'll move in quite a bit closer for a wider
angle of subtended view than for a facial portrait. Hence the detail
resolution requirements will be proportionally higher.

Sorry, but I don't think that necessarily follows.

I frankly think this thing about moving in closer is mostly a way of
justifying expensive higher resolution digital cameras. Funny how we
didn't hear about it when the best digital cameras were around 3 MP.

You didn't hear about IMAX cinema technology until cameras became
capable of the required resolution either. It's interesting to note
that the much bigger screens in IMAX cinemas haven't been used to
accomodate more viewers by having more distant view of the larger
screen. Instead they've tried to fit in as many as possible to a much
closer view than a conventional cinema screen. The reason is that
doing so supports the landscape viewing mode I mentioned previously
that "we" (human beings) can employ. If you haven't noticed the
significantly different effects on the human eye and brain of such a
wider view of a photographic image I suggest you visit an IMAX cinema.

--
Chris Malcolm DoD #205
IPAB, Informatics, JCMB, King's Buildings, Edinburgh, EH9 3JZ, UK
[http://www.dai.ed.ac.uk/homes/cam/]

John Navas wrote:
On Mon, 26 Nov 2007 03:10:55 +0000 (UTC), Peter Irwin
wrote in :

John Navas wrote:

But the PGI test is not a lax test. There is only one colour
negative film Kodak ever made that gets a perfect "less than 25"
score on the PGI test with 4x6 prints from 35mm film. That film
is Ektar/Royal Gold 25. Having seen the results posted from his
camera, I do not believe there is any way those can be printed
to yield subjective quality comparable to Ektar 25 film.

Has no bearing on the issue at hand,

It has every bearing on the issue at hand. The 14 inch distance
is part and parcel of the PGI test, but a perfect 25 score on the
PGI test is way stricter than David Littleboy, who is stricter
than me. You can't use the distance specified for that test to
justify lax standards.

Again, you're ducking my point, which is that Kodak specifies 14 inches
as the standard inspection distance.

Standards are needed for purposes of scientific comparison. There's
also a standard human stride, a standard human height, and a standard
human finger, all of which people vary quite a bit from. There's a
funny story about a chap called Procrustes who thought people ought to
adhere to a standard human height.

--
Chris Malcolm DoD #205
IPAB, Informatics, JCMB, King's Buildings, Edinburgh, EH9 3JZ, UK
[http://www.dai.ed.ac.uk/homes/cam/]

On 27 Nov 2007 10:06:29 GMT, Chris Malcolm wrote
in :

John Navas wrote:

I frankly think this thing about moving in closer is mostly a way of
justifying expensive higher resolution digital cameras. Funny how we
didn't hear about it when the best digital cameras were around 3 MP.

You didn't hear about IMAX cinema technology until cameras became
capable of the required resolution either. It's interesting to note
that the much bigger screens in IMAX cinemas haven't been used to
accomodate more viewers by having more distant view of the larger
screen. Instead they've tried to fit in as many as possible to a much
closer view than a conventional cinema screen. The reason is that
doing so supports the landscape viewing mode I mentioned previously
that "we" (human beings) can employ. If you haven't noticed the
significantly different effects on the human eye and brain of such a
wider view of a photographic image I suggest you visit an IMAX cinema.

My local IMAX has way more seats than its conventional theaters, and
seating distance is comparable.

--
Best regards,
John Navas
Panasonic DMC-FZ8 (and several others)

On Tue, 27 Nov 2007 03:00:35 -0600, Mr. Sigh...
wrote in :

Isn't it interesting (and highly amusing) that this person's whole sum of
experience with the subject was found on other's web-pages. That's a glaring
clue right there.

No citations and you complain about no citations.
Citations and you complain about citations.
Hmmm... I think I detect a pattern here.

--
Best regards,
John Navas http:/navasgroup.com

"Usenet is like a herd of performing elephants with diarrhea - massive,
difficult to redirect, awe inspiring, entertaining, and a source of mind
boggling amounts of excrement when you least expect it." --Gene Spafford

John Navas wrote:
On 27 Nov 2007 10:06:29 GMT, Chris Malcolm wrote
in :

John Navas wrote:

I frankly think this thing about moving in closer is mostly a way of
justifying expensive higher resolution digital cameras. Funny how we
didn't hear about it when the best digital cameras were around 3 MP.

You didn't hear about IMAX cinema technology until cameras became
capable of the required resolution either. It's interesting to note
that the much bigger screens in IMAX cinemas haven't been used to
accomodate more viewers by having more distant view of the larger
screen. Instead they've tried to fit in as many as possible to a much
closer view than a conventional cinema screen. The reason is that
doing so supports the landscape viewing mode I mentioned previously
that "we" (human beings) can employ. If you haven't noticed the
significantly different effects on the human eye and brain of such a
wider view of a photographic image I suggest you visit an IMAX cinema.

My local IMAX has way more seats than its conventional theaters, and
seating distance is comparable.

Sounds like someone realised you could make more profit if you ignored
quality.

--
Chris Malcolm DoD #205
IPAB, Informatics, JCMB, King's Buildings, Edinburgh, EH9 3JZ, UK
[http://www.dai.ed.ac.uk/homes/cam/]