Velocity Reviews > 6 Mpix Does it For Maniac

# 6 Mpix Does it For Maniac

Roger N. Clark
Guest
Posts: n/a

 07-18-2003

Dave Martindale wrote:

> "Roger N. Clark" <(E-Mail Removed)> writes:
>
> >There are two problems with the above. 1) it assumed one
> >pixel per resolution element, when in fact it is two.

>
> No, it seems to me that he got that part correct. The eye has a
> resolutio limit of about 60 cycles per degree, or one cycle per
> arcminute. As you note, one cycle requires (at least) two pixels.
> But the maniac used a angular extent of 30 arc seconds, 1/2 an
> arcminute, in his calculations. That *is* the correct size for one
> pixel, and the resolution of 5729x3819 that he calculates is a
> reasonable value.

Repeating my post to: Mxsmanic
No. There is extensive research on this. Some is
analyzed in my book "Visual Astronomy of the Deep Sky." Look at
Figure 2.6 on the page:
http://www.clarkvision.com/visastro/omva1

The curve for resolution of the eye is the "critical visual angle"
curve which bottoms out at 0.7 arc-minute. This level is
achievable if you have 20/20 vision or correct to 20/20
(note: some people do better than 20/20). The eye effectively
needs a minimum of two resolution elements to resolve something.
It may be achieved by scanning the eye, but nevertheleess it
achieves it. The 30 arc-seconds is off, and I should have
corrected that too (that is why I used 0.75 arc-minuite in my
example). With 0.7 arc-minute resolution, one needs 44 megapixels
in the original example. But as I also noted, resolution is one
thing, and edge sharpness is another and the eye does much better,
thus you need many more pixels to show that edge as sharp.

>
> Of course, that resolution is almost 22 megapixels. He doesn't explain
> how he gets from 22 to 7 megapixels.
>
>
> >At these angles,
> >and assuming 0.45 arc-seconds resolution, one needs
> >12768 x 15360 = 196 million pixels, very close to what a 4x5
> >drum scanned Fujichrome Velvia image gives.

>
> I assume you mean 0.45 arc *minutes* per pixel. 0.45 arc seconds is
> considerably beyond the resolution capability of my best telescope, let
> alone the naked eye.

Oops, that was 0.75 arc-minutes. The calculation used 0.75 arc-minute,
I just typed it wrong.

Roger

Roger N. Clark
Guest
Posts: n/a

 07-18-2003
Mxsmanic wrote:

> Dave Martindale writes:
>
> > Of course, that resolution is almost 22 megapixels.
> > He doesn't explain how he gets from 22 to 7 megapixels.

>
> It's mainly a result of practical viewing conditions. You really need
> ideal viewing conditions to reach the 22 megapixels; try it and see. So
> it's true that you'd need 22 megapixels to _guarantee_ that you reach
> the resolution limit of the eye under all conditions, but in real life,
> you hardly ever need that. Typically linear resolution will be only
> about half that, at best, and since pixel count varies with the square
> of linear resolution, that means that a more practical number is about
> six megapixels.
>
> This is why six megapixels is a "sweet spot" of sorts. For ordinary
> prints viewed at standard distances, six megapixels provides about all
> the detail you can actually see in the vast majority of viewing
> situations. Improvements beyond that can be hard to perceive, whereas
> improvements below that limit were easy to distinguish.
>
> Beyond 6 megapixels, additional increments improve the visual experience
> only under exceptional conditions. Beyond 22 megapixels, the additional
> pixels are a waste unless you are viewing an image at less than the
> standard distance, or unless you are cropping the image (in which case
> you need enough pixels to have at least 22 Mp in the final image).

This may be a good rule for 8x10 inch prints, but is totally
off base for larger prints, getting further off base as the
print size gets larger, assuming you want tack sharp prints.
I challenge anyone to make a 30x40 inch print from a 6-megapixel
camera that even comes close to the detail in a 30x40 inch print
from a drum scanned (650 megabyte file) Fujichrome Velvia 4x5
inch image. And this does not mean that a 30x40 inch print that
has impact can't be made from a 6-megapixel digital.

Statements like the above apply to only a very restricted case
and do not in any way represent the complete parameter space
of common real situations in photography and image diaplay.

Roger

Dave Martindale
Guest
Posts: n/a

 07-18-2003
"Roger N. Clark" <(E-Mail Removed)> writes:

>No. There is extensive research on this. Some is
>analyzed in my book "Visual Astronomy of the Deep Sky." Look at
>Figure 2.6 on the page:
>http://www.clarkvision.com/visastro/omva1

>The curve for resolution of the eye is the "critical visual angle"
>curve which bottoms out at 0.7 arc-minute. This level is
>achievable if you have 20/20 vision or correct to 20/20
>(note: some people do better than 20/20). The eye effectively
>needs a minimum of two resolution elements to resolve something.
>It may be achieved by scanning the eye, but nevertheleess it
>achieves it. The 30 arc-seconds is off, and I should have
>corrected that too (that is why I used 0.75 arc-minuite in my
>example). With 0.7 arc-minute resolution, one needs 44 megapixels
>in the original example. But as I also noted, resolution is one
>thing, and edge sharpness is another and the eye does much better,
>thus you need many more pixels to show that edge as sharp.

The 0.7 arc-minute figure seems to relate to the eye's ability to see a
small object on a large background. From the graphs and text, it seems
that an object smaller than this is seen as a point, while a larger
object appears to have area. Is that correct?

The 60 cycles/degree figure I used is a measurement of the highest
frequency periodic test pattern (sine wave or alternating black/white
bars) that can be seen by the eye.

Neither of these matches the actual content of the usual photograph, but
it isn't obvious to me that the spot-on-background data is more relevant
than the periodic-pattern data. Also, you clearly need at least two
resolution elements to resolve one cycle in the periodic pattern, but it
seems to me that the 0.7 arc minute critical angle for a single point on
a background suggests that one pixel (not two) needs to be kept below
0.7 arc minute.

Anyway, there's always some room to argue. As you point out, some
people have better than 20/20 vision, particularly when corrected.

Dave

Rafe B.
Guest
Posts: n/a

 07-18-2003
On Fri, 18 Jul 2003 07:30:31 -0600, "Roger N. Clark"
<(E-Mail Removed)> wrote:

>"Rafe B." wrote:
>
>> On Fri, 18 Jul 2003 00:04:44 +0100, Pat Chaney <(E-Mail Removed)> wrote:
>>
>> >On 17/7/03 6:57 am, "Rafe B." <(E-Mail Removed)> wrote:
>> >
>> >> Found it. The post I was looking for. Note the startling
>> >> conclusion, in the final paragraph. Makes you wonder
>> >> what mxsmaniac is doing with those 80 million pixels
>> >> he gets from scanning 6x6 on his LS-8000.
>> >
>> >Digging up a variety of posts dating back to 1999 is not consistent with
>> >having a strong argument in 2003.

>>
>> Since I was the one digging, I presume you're talking about
>> me in that last sentence.
>>
>> I am confident in my own argument(s) but not at all convinced
>> that they are worth presenting to mxsmaniac.

>
>Rafe:
>Your post was really trolling. But a few of us jump in
>because we try to keep information correct.

Trolling? Whatever. I've been "dealing" with Anthony
on USENET and elsewhere for years. The purpose of
this series of posts was to show that Anthony is wildly
inconsistent in his arguments. In short, he is equally
comfortable arguing that black is white, or that white is
black. Point is, few posters are more consistently wrong,
or more sure of their correctness as they deliver mis-
information.

Plus, there's something about Anthony's unquenchable,
unshakeable ego that's intensely irritating.

>The philosophical ideas constantly being presented in this
>group often (depending on the religious camp) say digital
>is better than or equal to film versus no it's not.
>People forget that there is not one answer here. Depending
>on the target requires vastly different resolution. So
>ridiculing someone for needing 88 megapixels just shows
>bias and ignorance.

I understand. While I argue (lately) in favor of digital capture,
I am well aware that there are two sides to this story -- but
mxsmaniac only acknowledges one side.

In case you missed it, I do a lot more work with scanned
film than I do with digital capture -- at least for now. But
I expect that to change over the next few years.

>If you have ever seen a big print from large format, and tried
>to do something similar from 35mm film or DSLR, you would
>quickly realize you can't get the detail. Go to a poster store
>and look at some 24x36-inch posters (of photographic scenes)
>(if you don't have a nearby gallery of large format prints).
>I bet you can pick out the large format posters, as they will be
>much sharper than the others. And they need on the order of
>200 megapixels to achieve that sharpness.

LF hasn't really entered this discussion. Even Anthony
can only get 80 Mpixels or so scanning 6x6 on his CoolScan.
Me, I'm stuck with 55 Mpixels (645 film), but I'm not selling posters.

Still, my point is that dicams get a lot more mileage than
scanned film from a given pixel count. I can't give you
the magic ratio, but my gut says it's around 2x or more --
maybe even as much as 4x more. Obviously it depends
on lenses in both cases, and for the scanned film case, the film.

>The right tool for the right application: no current photographic
>tool can do it all for every application, and it is unlikely
>to ever occur in our lifetimes.

Who could disagree? I don't mean to be a digital-capture
evangelist, although I certainly do believe it is the wave of
the future. I don't have personal experience with an EOS-
1Ds, but I certainly have looked at some captures from that
camera and been impressd.

I get enough pixels from my 645 scans to make some really
good sized huge prints. Let's see... 8500 pixels at, say, 360
dpi still gives a print 26 inches on the long axis.

But I don't see why a print is necessary -- viewing the image
in Photoshop at 100% or 200% tells the story.

rafe b.
http://www.terrapinphoto.com

Mxsmanic
Guest
Posts: n/a

 07-18-2003
Roger N. Clark writes:

> I bet you can pick out the large format posters,
> as they will be much sharper than the others.

If you can find any! Posters made from MF originals are very rare, and
posters made from LF originals are almost nonexistent. When they do
exist, they may be extremely expensive.

Most people don't realize how fuzzy the average poster is, because
they've never seen really sharp posters.

Also, if the viewing distance is kept proportionately constant, the
extra resolution is not needed. But if the enlargement is wall-sized,
chances are that people will be viewing it at less than the standard
distance.

--
Transpose hotmail and mxsmanic in my e-mail address to reach me directly.

Mxsmanic
Guest
Posts: n/a

 07-18-2003
Roger N. Clark writes:

> I challenge anyone to make a 30x40 inch print from
> a 6-megapixel camera that even comes close to the detail
> in a 30x40 inch print from a drum scanned (650 megabyte
> file) Fujichrome Velvia 4x5 inch image.

If you are viewing it from a standard distance of 60 inches away, it
won't matter.

> Statements like the above apply to only a very
> restricted case and do not in any way represent
> the complete parameter space of common real situations
> in photography and image diaplay.

In the real world, I think that most people don't see more than a
handful of 30x40-inch prints in their entire lifetimes.

--
Transpose hotmail and mxsmanic in my e-mail address to reach me directly.

Mxsmanic
Guest
Posts: n/a

 07-19-2003
Rafe B. writes:

> Plus, there's something about Anthony's unquenchable,
> unshakeable ego that's intensely irritating.

A single sentence that reveals much.

But is it really ego that irritates you, or is it something else?

--
Transpose hotmail and mxsmanic in my e-mail address to reach me directly.

Lionel
Guest
Posts: n/a

 07-19-2003
On Sat, 19 Jul 2003 01:55:09 +0200, in
<(E-Mail Removed)>, Mxsmanic
<(E-Mail Removed)> said:

>At a stadard distance, and from a resolution standpoint, that is quite
>true. It's even true for digital originals, if the DSLR had enough
>pixels.
>
>There are other factors that might give things away, but resolution is
>not usually among them.
>
>Despite this, I'd certainly prefer LF over MF, and MF over 35mm, and
>35mm over digital, all else being equal.

This is, of course, an emotional bias on your part.

--
W
. | ,. w , "Some people are alive only because
\|/ \|/ it is illegal to kill them." Perna condita delenda est
---^----^---------------------------------------------------------------

Lionel
Guest
Posts: n/a

 07-19-2003
On Sat, 19 Jul 2003 02:33:18 +0200, in
<(E-Mail Removed)>, Mxsmanic
<(E-Mail Removed)> said:

>Rafe B. writes:
>
>> Plus, there's something about Anthony's unquenchable,
>> unshakeable ego that's intensely irritating.

>
>A single sentence that reveals much.

About you, certainly. In England & Australia, we use the word 'wanker'
to describe people like you, who pretend to be authoritative about
things they know little about, & who refuse to back off when they get
something wrong.

--
W
. | ,. w , "Some people are alive only because
\|/ \|/ it is illegal to kill them." Perna condita delenda est
---^----^---------------------------------------------------------------

Mxsmanic
Guest
Posts: n/a

 07-19-2003
Lionel writes:

> In England & Australia, we use the word 'wanker'
> to describe people like you, who pretend to be
> & who refuse to back off when they get something wrong.

More forced teaming, I see ("we" in place of "I").

As a matter of fact, I do know quite a bit about quite a few subjects.
I don't have to pretend.

And I admit when I'm wrong, but I take care not to say things that may
turn out to be wrong in the first place. Usually, when people say I'm
"wrong," they are actually attacking their incorrect assumptions about
what they think I said, instead of actually examining what I really did
say.

--
Transpose hotmail and mxsmanic in my e-mail address to reach me directly.