Popping the Myth of the Focal Length Multiplier I

keithallenlaw · Apr 11, 2007

Two DSLR's, and a chit load of L glass. I'm gonna get me a
12x zoom p/s camera and enjoy the same quality as all this
gear in one compact camera. I have finally seen the light...

--
Logic is superior to emotion
BUT
Emotion is much more fun!

Watch out for the exposure police.
They will tell you that its wrong even
though they weren't there to witness it.

Come visit me at:
http://www.pbase.com/keithallenlaw
See my profile for what i shoot with.

Chato · Apr 11, 2007

keithallenlaw said:
Two DSLR's, and a chit load of L glass. I'm gonna get me a
12x zoom p/s camera and enjoy the same quality as all this
gear in one compact camera. I have finally seen the light...

My phoney argument apparently went right by you...

'

Dave -

BTW, I'm going to do the same thing....

keithallenlaw said:
--
Logic is superior to emotion
BUT
Emotion is much more fun!

Watch out for the exposure police.
They will tell you that its wrong even
though they weren't there to witness it.

Come visit me at:
http://www.pbase.com/keithallenlaw
See my profile for what i shoot with.

Petteri Sulonen · Apr 11, 2007

I really don't see what it is about this that (a) you find so hard to understand and (b) bothers you so much that you keep harping on it.

It's very, very simple: if the lens you're using resolves more lp/mm than your sensor, for purposes of resolution, the physical size of the lens and sensor do not matter.

In other words, a 45 mm lens on a small sensor resolves exactly as much as a 450 mm lens on a sensor ten times larger with the same pixel count, if the 45 mm lens resolves ten times more lp/mm. This is usually the case with small-sensor cameras and their lenses: the lenses project very small image circles but with very high resolutions.

You only hit a limit when diffraction comes into play, which it doesn't with current digital cameras at their open apertures.

So, Chato my friend, that 45 mm on the little sensor is the equivalent of the 450 mm lens on the big sensor, for purposes of resolving power. No marketing hype about it.

What's more, bringing "crop factors" into it is just plain silly. There's nothing magical about the 24 x 36 mm frame size that makes telephoto lenses for it "real" telephotos.

Petteri
--
http://www.prime-junta.net/
http://www.flickr.com/photos/primejunta/

Petteri Sulonen · Apr 11, 2007

Chato wrote:
[snip]

Chato said:
Nontheless the sensor cannot resolve detail that is not being
presented to it.

True.

Chato said:
There is no Magic to the science of optics.

True.

Chato said:
You
cannot turn a 45mm lens into a 450mm lens by reciting the formula
or more pixels...

True.

However, what you can do, is make a higher-resolution 45 mm lens. This is actually rather easy, if you're allowed to relax one important constraint: the size of the image circle. This allows you to use a telephoto design on the lens, which makes for higher resolution. Say, you could take an existing 450 mm lens and just scale down the design by a factor of ten: you'll get a 45 mm lens with an image circle 1/10 the size, and ten times higher resolution (until you hit the diffraction limit, naturally). (It'll also be drastically cheaper to make, since it's way smaller.)

Do that, and you're feeding the sensor with enough detail to max it out. In this case, your 45 mm lens on the small sensor will resolve as much as the 450 mm lens on the big sensor with the same pixel count.

Petteri
--
http://www.prime-junta.net/
http://www.flickr.com/photos/primejunta/

Chato · Apr 11, 2007

nickleback said:
nickleback said:

nickleback said:

Both seem to resolve detail quite well. The 400D did a bit better,
but it's shooting through a prime lens. If I believed your story,
the Pany numbers should be about 1/3 the Canon numbers.

Click to expand...

I did not see in the test of the digicams what focal length was
chosen. While the sSLR's were shot with primes.

Click to expand...

Click on the image and check the EXIF data, it is about 20mm IIRC.
The Canon was shot with a 50mm lens. The distance is about the
same. Yet, somehow, both resolve the same detail. According to
you, this is not possible. Perhaps you should rethink your
theories.

How do I check the EXIF data? I downloaded both images and can't find it in Photoshop...

However it appears that the settings used certainly varied between the two tests. I doubt if "all things were equal."

Here is a side by side, interpolated up to twice normal, but no other processing

They sure don't look like identical conditions...

nickleback said:
nickleback said:

Clearly comparing a prime with a prime will yield similar results
in resolving power.

Click to expand...

The comparison is a prime on the DSLR and a zoom on a digicam with
a sensor 1/3 the size. The digicam is handicapped. Regardless,
both end up with quite similar resolution at similar "35mm
effective" focal lengths (i.e. around 80mm).

It doesn't "look" to me that we have a flat out equal comparison here.

But I can do a simple test FOR you. I have a Nikon D1x, 5 megs, and a whole batch of primes and zooms. And I have a Canon 540 P&S, 6Megs and 35 to 140 "equivalent" lens.

So tell me what paramters you wish, and I will present you with the results...

Dave

nickleback said:
Seen in a fortune cookie:
Fear is the darkroom where negatives are developed

Dave

PS. At the moment there appears to be a problem with my server. Hopefully the Yahoo folks will fix it soon. As far as I know, my hanbd written I.O.U. didn't bounce...

Andrew dB · Apr 11, 2007

nickleback said:
Andrew dB said:

This
explains the high resolutions achievable with large format systems
despite the low (image) resolution lenses used.

Click to expand...

Yep.

nickleback said:

Perhaps the better formula would be something like 250,000λ d
where λ is the wavelength in metres and d is the aperture,
also in meters. This formula gives the resolving power of the lens
in seconds of arc so clearly, larger physical apertures can achieve
higher resolutions, irrespective of focal ratio.

Click to expand...

"can" achieve is different from "does" achieve.

It's a nice theoretical argument, but it doesn't pan out for the
cameras in question. Undoubtably if you had a very fine 450mm
f/2.8 in front of a 35mm sensor with 2 micron pixels, and a very
fine 90mm f/2.8 in front a 1/1.8" sensor also with 2 micron pixels,
the 35mm sensor will absolutely smoke the 1/1.8" sensor for detail.

Very true. The formulas work just fine when dealing with telescopes and microscopes but when ordinary cameras shooting normal photos are being discussed, there are a lot of caveats.

Of course, the way to get very high res results, even if the 216MP sensor doesn't exist is to use either the small sensor or boost the image magnifications and use the large pixel 35mm sensor and then take multiple shots to build up a high res mosaic. Great for astronomy and utterly useless for the majority of photographs!

nickleback · Apr 11, 2007

keithallenlaw said:
Two DSLR's, and a chit load of L glass.

I have but one DSLR and two L lenses. But I'm not going to sell.

keithallenlaw said:
I'm gonna get me a
12x zoom p/s camera and enjoy the same quality as all this
gear in one compact camera. I have finally seen the light...

Chato was talking resolution. He's wrong. Current digicams and DSLRs resolve about the same detail.

There are other qualities of DSLRs that I'm not willing to give up.

--
Seen in a fortune cookie:
Fear is the darkroom where negatives are developed

Chato · Apr 11, 2007

BorisK1 said:
(I do not recommend reading this old subthread):
http://forums.dpreview.com/forums/read.asp?forum=1018&message=21847767

It's a thread about people who want to do professional photography without having to invest in all that USELESS equipment. They can save a heck of a lot of money by buying a super zoom digicam...

BorisK1 said:
Would you please, please, stop quoting microscope-related articles?
We're talking about digital cameras here. Not about microscopes.
Things that you read about microscopes do not necessarily apply to
digital cameras. Even though, as you point out, they are written
by (gasp) scientists.

The science of optics is the science of optics. Please read at least some of the links I posted in the original two posts.

Dave

avidday · Apr 11, 2007

Chato wrote:
[snip]

Chato said:
Just because you have x number of MP’s on the sensor, that has a ten
times cropping factor only means you have the field of view of a telephoto
and NOT the resolving power of a telephoto.

[snip]

Unfortunately that ignores the very definition of resolving power. Lord Rayliegh (1906 Nobel prize winner for physics) observed in 1896 that the spatial resolving power of a lens was directy proportional to the aperture ratio and the wavelength of the light being focussed. His relationship

theoretical maximum spatial resolution = 1.22 * light wavelength * focal length / diameter

is one of the cornerstones of modern optical physics. A short focal length lens has the same theoretical resolving power as a longer focal length lens if they share the numerical aperture ratio.

Andrew dB · Apr 11, 2007

avidday said:
avidday said:

Just because you have x number of MP’s on the sensor, that has a ten
times cropping factor only means you have the field of view of a telephoto
and NOT the resolving power of a telephoto.

Click to expand...

[snip]

Unfortunately that ignores the very definition of resolving power.
Lord Rayliegh (1906 Nobel prize winner for physics) observed in
1896 that the spatial resolving power of a lens was directy
proportional to the aperture ratio and the wavelength of the light
being focussed. His relationship

theoretical maximum spatial resolution = 1.22 * light wavelength *
focal length / diameter

is one of the cornerstones of modern optical physics. A short focal
length lens has the same theoretical resolving power as a longer
focal length lens if they share the numerical aperture ratio.

But that is the spatial resolution of the image. By using a larger physical aperture, you can resolve finer angular resolution and hence detail in the object.

Take the example of two lenses: a 100mm f4 and a 200mm f4. Both will have the same spatial resolution in the image ( ~~2.6 microns) but the angular resolving power of the longer lens will be twice that of the shorter one (~~ 2.8 arcseconds versus 5.6 arcseconds) due to the larger physical aperture so it can resolve details twice as small in the object being photographed.

BorisK1 · Apr 11, 2007

Chato said:
It's a thread about people who want to do professional photography
without having to invest in all that USELESS equipment. They can
save a heck of a lot of money by buying a super zoom digicam...

The reason I posted the reference is to show others that you've already seen all the problems with your position, but chose (and still choosing) to ignore them.

Chato said:
The science of optics is the science of optics.

So, how does your "science of optics" explain the resolution figures that Phill is getting for the small-sensor digicams?

nickleback · Apr 11, 2007

Chato said:
How do I check the EXIF data? I downloaded both images and can't
find it in Photoshop...

I had no problem finding the EXIF data.

From the Pany image:

Camera Make = Panasonic
Camera Model = DMC-FZ50
Exposure Time (1 / Shutter Speed) = 10/2000 second = 0.00500 second
Lens F-Number/F-Stop = 45/10 = F4.50
Exposure Program = aperture priority (3)
ISO Speed Ratings = 100
Exposure Bias (EV) = 66/100 = 0.66
Max Aperture Value (APEX) = 30/10 = 3.00

Max Aperture = F2.83

Focal Length = 227/10 mm = 22.70 mm
Image Width = 3648
Image Height = 2736
Digital Zoom Ratio = 0/10 = 0.00
Focal Length in 35mm Film = 107

...etc. I cut out a bunch of stuff as it made the post too long. As you can see, it was shot at 1/200, f/4.5, ISO 100, 22.7mm (107mm equivalent).

The Canon 400D shot was shot at 1/60, f/9, ISO 100, 50mm (80mm equivalent)

Chato said:
However it appears that the settings used certainly varied between
the two tests. I doubt if "all things were equal."

Pretty darn close.

Chato said:
Here is a side by side, interpolated up to twice normal, but no
other processing

You've just proved that the numbers given in the review are correct. You conveniently picked the horizontal resolution, which the 400D is a better at, and left off the vertical resolution, where they are tied.

Chato said:
But I can do a simple test FOR you. I have a Nikon D1x, 5 megs, and
a whole batch of primes and zooms. And I have a Canon 540 P&S,
6Megs and 35 to 140 "equivalent" lens.

So tell me what paramters you wish, and I will present you with the
results...

The D1x is a strange camera. It has rectangular pixels. The D1x will smoke the A540 for horizontal resolution (since the D1x is effectively a 12mp camera, horizontally), and the A540 will smoke the D1x for vertical resolution (since the D1x is effectively a 3mp camera, vertically). Go ahead and take some shots if you like, but here's what you'll probably get:

http://www.dpreview.com/reviews/nikond1x/page25.asp
http://www.dpreview.com/reviews/canonsd700is/page3.asp

D1x: horizontal 1600, vertical 1150
SD700: horizontal 1400, vertical 1500

(yes, I know that's the SD700is review, but there's not A540 review on this site and the SD700is has the same sensor as the A540)

--
Seen in a fortune cookie:
Fear is the darkroom where negatives are developed

TrapperJohn · Apr 11, 2007

'crop factor' only applies with legacy film lenses, or in the case of Canon, 24x36 image circle lenses used on an APS sensor. If you're one of those who were led into dslr's based on the legacy film lens pitch, I could see why you'd be disappointed. Not only do you lose resolving power, but you have edge problems as well, especially with wide angles.

Not all lenses fall into that category. Nikon DX, Canon EF-S, and any 4/3 lens does not impose a 'crop factor'. They produce an image circle tailored to the sensor size.

Most P&S cameras do the same, though their IQ issues are related to price point, not a specific flaw in lens design.

nickleback · Apr 11, 2007

Chato said:
nickleback said:

Click on the image and check the EXIF data, it is about 20mm IIRC.
The Canon was shot with a 50mm lens. The distance is about the
same. Yet, somehow, both resolve the same detail. According to
you, this is not possible. Perhaps you should rethink your
theories.

Click to expand...

However it appears that the settings used certainly varied between
the two tests. I doubt if "all things were equal."

One more little thing...

Don't put words into my mouth. I never said "all things were equal". Look at the words I actually wrote.

--
Seen in a fortune cookie:
Fear is the darkroom where negatives are developed

zygh · Apr 11, 2007

No myth, just poor understanding of the principle by some people, that's all.

--
http://www.flickr.com/photos/zygh/
http://zygh.deviantart.com

Chato · Apr 11, 2007

nickleback said:
Chato said:

nickleback said:

Click on the image and check the EXIF data, it is about 20mm IIRC.
The Canon was shot with a 50mm lens. The distance is about the
same. Yet, somehow, both resolve the same detail. According to
you, this is not possible. Perhaps you should rethink your
theories.

Click to expand...

However it appears that the settings used certainly varied between
the two tests. I doubt if "all things were equal."

Click to expand...

One more little thing...

Don't put words into my mouth. I never said "all things were
equal". Look at the words I actually wrote.

I'm quoting myself - Moreover the only cherry picking I did was to show the horizontal lines. I "choose" this for the following reason:

In the example I posted the horizontal lines are almost EXACTLY twice as large in the Panasonic image as in the Canon image. So what the actual parameters used in this test are, I have no idea.

Second, I never said the EXIF data wasn't there, I asked you how to get it. I downloaded both images, and PS 7 didn't show any data, and neither did CS (Didn't try CS2).

So I assume that I'm doing something wrong. It never occured to me that you are being dishonest. So I ASKED you how to secure that data.

As for the test I suggest, I asked YOU to set up the parameters, your answer SEEMS to suggest that there are none that I could use.

I have a D1x and MP wise it more or less matches the Canon that I have. I also have a D2x that I could use, but that of course is a 12 MP camera.

So, could you suggest such a test with the equipment I DO have?

I have a Nikon prime at 50mm, a neat Tamron 28-105, a Sigma 35, a Sigma 180 and others.

Dave

nickleback said:
--
Seen in a fortune cookie:
Fear is the darkroom where negatives are developed

Chato · Apr 11, 2007

Petteri Sulonen said:
I really don't see what it is about this that (a) you find so hard
to understand and (b) bothers you so much that you keep harping on
it.

It's very, very simple: if the lens you're using resolves more
lp/mm than your sensor, for purposes of resolution, the physical
size of the lens and sensor do not matter.

In other words, a 45 mm lens on a small sensor resolves exactly as
much as a 450 mm lens on a sensor ten times larger with the same
pixel count, if the 45 mm lens resolves ten times more lp/mm. This
is usually the case with small-sensor cameras and their lenses: the
lenses project very small image circles but with very high
resolutions.

You only hit a limit when diffraction comes into play, which it
doesn't with current digital cameras at their open apertures.

So, Chato my friend, that 45 mm on the little sensor is the
equivalent of the 450 mm lens on the big sensor, for purposes of
resolving power. No marketing hype about it.

What's more, bringing "crop factors" into it is just plain silly.
There's nothing magical about the 24 x 36 mm frame size that makes
telephoto lenses for it "real" telephotos.

It should be easy for you to quote from the scientific literature to back up your statement. Aside from the numerous photography sources that I've used, I used a PHD Paper devoted specifically to refuting your statements.

If my point is so obviously wrong, you better write to the university this author is a Fellow at, and get him removed from his position before he spreads more false information. The PDF below makes for more interesting viewing then the HTML file link I posted before....

https://archive.ugent.be/retrieve/3673/Verhoeven+2007+-+Did+the+Digital+ (R)Evolution+Change+the+Concept+of+Focal+Length.pdf

Dave

Petteri Sulonen said:
Petteri
--
http://www.prime-junta.net/
http://www.flickr.com/photos/primejunta/

Petteri Sulonen · Apr 11, 2007

...the problem is that you don't understand what you're quoting.

Consider this; perhaps it'll help you sort out the confusion in your head about the topic.

Take a 450 mm lens.

Shrink it by a factor of 10.

What do you get?

You get the same lens, only in miniature: it produces exactly the same image and resolves exactly the same amount of detail (in lp/height) -- only ten times smaller. Specifically, you get the same field of view, but an image circle that's 1/10 the size, and ten times sharper as measured in lp/mm.

In other words, the equivalent -- only smaller.

This is why small-sensor digicams resolve just about the same amount of detail as dSLR's with comparable pixel counts.

Again, at some point diffraction will throw a monkey wrench into the works, since although everything else scales, light itself doesn't. This is why there's no point stopping down a digicam lens past f/5.6.

Petteri
--
http://www.prime-junta.net/
http://www.flickr.com/photos/primejunta/

Chato · Apr 11, 2007

avidday said:
avidday said:

Just because you have x number of MP’s on the sensor, that has a ten
times cropping factor only means you have the field of view of a telephoto
and NOT the resolving power of a telephoto.

Click to expand...

[snip]

Unfortunately that ignores the very definition of resolving power.
Lord Rayliegh (1906 Nobel prize winner for physics) observed in
1896 that the spatial resolving power of a lens was directy
proportional to the aperture ratio and the wavelength of the light
being focussed. His relationship

theoretical maximum spatial resolution = 1.22 * light wavelength *
focal length / diameter

is one of the cornerstones of modern optical physics. A short focal
length lens has the same theoretical resolving power as a longer
focal length lens if they share the numerical aperture ratio.

The telephoto lenses that we are discussing are designed for gathering and concentrating light. This information is then focused by the lens onto the sensor.

We are now (not theoretically - but NOW) comparing the lens of a digicam with an "equivalent" FL of 450mm with a real 450mm telephoto. What does the sensor have to do with this? Nothing.

What can be resolved with this digicam lens? About the same amount (or less) as a decent prime of the ACTUAL mm. Now if you hung a 50mm Prime that gathers as much light as a 450mm telephoto AND had a sensor capable of "seeing" all this light - You would be correct.

But such equipment doesn't exist - neither the lens nor the sensor. Not only that, but NO ONE would even dream of creating such a rediculous lens.

If you monitor the threads that appear on this board, you will see that quite a few people honest to God believe that their cameras easily match real telephoto.

In other words, in the Real World, you are just repeating numbers that simply have no meaning.

Dave

nickleback · Apr 11, 2007

Chato said:
In the example I posted the horizontal lines are almost EXACTLY
twice as large in the Panasonic image as in the Canon image.

No they are not. Measure across the horizontal lines at 20. I get about 46 pixels on each image. The reason the lines look thinner on the Canon shot is that you are looking at the range 1800 - 2600, whereas on the Pany image you are looking at 1200 - 2200.

Chato said:
So what the actual parameters used in this test are, I have no idea.

They are well defined, same parameters as use in all resolution tests on this site. Shoot at the sharpest aperture and fill the frame vertically with the chart.

Chato said:
Second, I never said the EXIF data wasn't there, I asked you how to
get it. I downloaded both images, and PS 7 didn't show any data,
and neither did CS (Didn't try CS2).

It's there, all the applications you mentioned are perfectly capable of extracting the EXIF. Beats me why they didn't for you. FWIW I just tried it on CS, shows up just fine. Click on File-> File Info.

Chato said:
As for the test I suggest, I asked YOU to set up the parameters,
your answer SEEMS to suggest that there are none that I could use.

Shoot something with detail, both vertical and horizontal. A res chart has these elements, and you don't even need to bother shooting because they are here on this site. But go ahead, setup a shoot. It doesn't matter what I say, the cameras will perform as they will perform. Pick focal lengths that give you equivalent FOV and fire away.

Chato said:
I have a D1x and MP wise it more or less matches the Canon that I
have.

It does and it doesn't. As I already mentioned, horizontally the D1x is like a 12mp camera, vertically it is like a 3mp camera. I've already pointed out the res charts shot on it and on a Canon camera similar (same sensor, different lens) as the Canon you own.

Chato said:
I have a Nikon prime at 50mm, a neat Tamron 28-105, a Sigma 35, a
Sigma 180 and others.

Wow, I'm impressed. It really doesn't matter. You can pick the finest Zeiss macro, stick it on the D1x, and the A540 will outresolve it vertically, and the D1x will outresolve the A540 horizontally. Spend all that time and all you'll have done is proved the charts here are fair.

--
Seen in a fortune cookie:
Fear is the darkroom where negatives are developed

Popping the Myth of the Focal Length Multiplier I

Senior Member

Forum Pro

Forum Pro

Forum Pro

Forum Pro

Leading Member

Forum Pro

Forum Pro

Senior Member

Leading Member

Veteran Member

Forum Pro

Forum Pro

Forum Pro

Senior Member

Forum Pro

Forum Pro

Forum Pro

Forum Pro

Forum Pro

Keyboard shortcuts