Have sensors really passed the resolution of lenses? NO!

I keep reading that statement, that they have.
Does this mean that putting a good lens on a 24 megapixel camera (not
to mention a 36mp) rather than an 16 megapixel camera will yield no
extra resolution? Has this been anyone's experience, even with some
marginal lenses, like Sony's 16mm pancake? I'd urge people to
actually do some tests rather than accept it at face value.
Reason being, there are tiny sensored cameras out there that will
resolve a decent amount of detail with their lenses and therefore,
scaling up those sensors to the size of larger sensors would disprove
the idea that sensors have now out-resolved lenses.
This could be proven quite handily by mounting a DSLR lens on a
diminute Pentax Q with an adapter, or even using Nikon's V1 as both
cameras, where their sensors scaled-up to say APS size for example,
would exceed the pixel count of any (including the D800) DSLR, and it
would be shown the lenses are capable of delivering even more detail.

 

The answer is, as you well, know "it depends". With some DSLR lenses, 10
MP may well be enough to capture all they can offer when wide-open. When
viewing a "normal size" print at "normal" distance, or looking at an image
on a computer screen or CRT display. not pixel peeping, could you even
tell the difference between a 16 MP and 24 MP print? I doubt it.

To get the full resolution out of a 36 MP sensor will likely require the
best of lenses, and some of the best of photographic techniques (such as
tripod, best focusing, reducing mirror-slap etc. etc.)

 

Consider that many consider the resolution of Kodachrome 35mm film to be
around 100-250 mega pixels.

 

consider that many people are stupid. there's absolutely *nothing* that
can support that nonsense.

 

According to the modulation-transfer diagram for Kodachrome 25 a good
estimate for an equivalent digital image is about 20 megapixels.

http://www.kodak.com/global/en/professional/support/techPubs/e55/e55.pdf

(70 cycles/mm * 2 pixels/cycle * 35mm) * (70 * 2 * 25)

 

This is as deeply and well thought out as any post from you I have
read. Of course, I only read your posts when I am in the mood, so you
may have done better and I wouldn't know.

It's an interesting question. It seems from your heading that your
opinion is no while David Taylor's more qualitied opinion in his post
is that it all depends, which is intuitively more appealling.
Postings to newsgroups understandably limit how full an explanation
can be, which is a good reason to link to outside resources. The
conclusion reached by Luminous Landscape, using their greater
resources and probably coming from a more knowledgeable perspective
than most of us have, came to the conclusion the same conclusion as
David Taylor and they definitely aren't alone.

Quoting Luminous Landscape's conclusion in full:

'So, do sensors outresolve lenses? It depends on the lens you use, the
properties of the light, the aperture and the format. Small format
sensors may have surpassed the limit, this is, in most cases they are
lens-limited in terms of resolution. It is easier to correct
aberrations for a smaller light circle though, so you can approach
diffraction-limited resolutions for lower f-numbers. The signal-to-
noise ratio, however, imposes an inflexible limit to the effective
resolution of the whole system, mostly due to photon shot noise.

Sensors for larger formats are approaching the diffraction limit of
real lenses, and it is more difficult to get high levels of aberration
suppression for them. The point is that you cannot fully exploit the
resolution potential of high-resolution sensors with regular mass-
produced lenses, particularly for larger formats.

You cannot compare the limits of two different photographic systems
looking at a print because the variables that determine the subjective
perception come into play. Different systems can provide comparable
results on paper under certain conditions (the circle of confusion
reasoning explains how that is possible), but the limit of a system
must be evaluated considering the pixel as the minimum circle of
confusion.'

For those who want a deeper understanding of the question, or want to
see if the conclusion is sound, the link is:

http://luminous-landscape.com/tutorials/resolution.shtml

Your approach to 'proving' that sensors haven't outstripped lens
resolution seems a weak and way to much bother. In deference to
people who know more than I do, which seems reasonable, I have to go
with the less catagorical answer than either 'Yes' or 'No' and say
'Maybe', in spite of how commonly you hear 'Sensors have outstipped
lens resolution.'

Vance

 

Well, that stuff has been around for a while, but in truth, they are
referring mostly to scans of the film and the relative size of the files
that result from those scans. It's not resolution, it's the recording of
the grain structure of the film, which is why a course grain film will
often require even more memory than higher resolution, fine-grained film.
Apart from specialty films meant for ultra-high (by 35mm film standards)
resolution max-out at about 8 megapixels, compared to equivalent digital
images.

 

You want more than 2 pixels/cycle[1]. You want to seriously
oversample[1]. You want 0% response, not 10%[2].

(130 cycles/mm * 3 pixels/cycle * 3 (oversample))^2 * 24 * 36
=> >1 GPix. See?

-Wolfgang

[1] http://clarkvision.com/articles/sampling1/index.html
[2] http://clarkvision.com/articles/scandetail/index.html#modulation

 

Yes, and you can extend the MTF to the whole system, including any
filtering on the sensor (e.g. anti-alias filtering), the effects on any
processing, the MTF of the screen or printer, and the MTF of the eye, to
predict the image on the retina.

However, it can be convenient to take some defined point of the MTF (e.g.
30%), or the integrated area under the MTF curve, and a single
"resolution" equivalent measure, but as this excludes the shape of the MTF
curve is doesn't tell the whole story.

David

 

If you take 30% as the limit, the lens resolution will be an extremely
low one. MTF charts are typically made up to a max of 40 lp/mm and
already at 40 lp/mm the MTF is very, very low.
On the other hand with 4 micrometer pixels we are at 125 lp/mm, and you
can imagine what kind of MTF you get at 125 lp/mm - probably just a few
percent.

 

?????

I really don't understand what you mean.

 

Don't worry, Alfred. Browne doesn't understand either.

 

What you said:
QUOTE
Lenses do not have a resolution. There is something called
modulation transfer function (MTF) which tells you what
the response is at a certain spatial frequency.
/QUOTE

However, as you go up in spatial frequency, the contrast ratio falls
off. When you can no longer usefully distinguish between two close
lines you're at the limit for that lens. MTF is the means to
establishing the lens resolution.

Ref: http://en.wikipedia.org/wiki/Modulation_transfer_function

Q The optical transfer function (OTF) of an imaging system (camera,
video system, microscope etc.) is the true measure of resolution
(image sharpness) that the system is capable of.
/Q

And: http://en.wikipedia.org/wiki/Optical_resolution#Lens_resolution

 

Now it gets more clear, but the way you initially put it "higher spatial
frequency" contrast _is_ lens resolution" means nothing. You were
meaning something else. By the way, first time I hear of this OTF.

 

What set me off was when you said "lenses do not have a resolution".
(They do). And then brought up MTF (which is how you define/find the
resolution)

It's not useful to photography - as we do it anyway.

MTF is a component of OTF (See the equation). We're concerned with
magnitude (causes exposure), not phase - MTF "discards" the phase info
from OTF.

If a phase sensor could be done - and very small, I guess resolutions
could be magnitudes higher and no lens at all would be needed.

 

On 25/02/2012 19:55, Alan Browne wrote:
(...)

If you record phase as well as modulation in essence you'd have digital
holography, no?

 

You certainly have the information to focus at any distance and resolve
for any desired DOF. Not sure about holography.

The article in wikipedia is a good start, and it it:


[1] A hologram represents a recording of information regarding the light
that came from the original scene as scattered in a range of directions
rather than from only one direction, as in a photograph. This allows the
scene to be viewed from a range of different angles, as if it were still
present.

[3] A holographic recording requires a second light beam (the reference
beam) to be directed onto the recording medium.

and much more, of course.

 

Absolute accuracy is neither the goal nor is it possible.
This is a useful guideline.

If you want to indulge yourself in pointless pedantry then go ahead.

 

Plenoptic camera maybe?

Yes, but that's for analog holograms, with a physical recording medium
such as photopolymers or photorefractives.

If you look at <http://en.wikipedia.org/wiki/Digital_holography> you'll
find "The phase-shifting digital holography process entails capturing
multiple interferograms that each indicate the optical phase
relationships between light returned from all sampled points on the
illuminated surface and a controlled reference beam of light that is
collinear to the object beam"

Not that I truly understand any of it... ;-)

 

? (Okay, I'll go look it up...)

Sure, like the Lytro. Though the way I read the para on Wikip, it's not
as described in our flight of fancy above.

Me neither - but I believe it entails multiple POV's, not a single POV.