more pixels are better!

Started Dec 14, 2008 | Discussions
John Sheehy Forum Pro • Posts: 27,400
Re: Are those figures for

Graystar wrote:

John Sheehy wrote:

You said something, or I wouldn't have replied. You made a complete,
independent statement:

You're attacking an answer without considering the question that the
answer is answering. That seems to be something you like to do.

I play be the rules of reason. I am quite cognizant of what depends on the link, and what does not. You made a complete, independent statement which stands alone, without context. You said the sensor breaks up the analog focal plane into "pieces". Some potential connotations of such a statement are completely wrong; it is a very poor description of what a sensor does, regardless of anything you may be replying to or linking to. It can launch someone into a completely incorrect idea about what a sensor is doing, and that is why I had to reply.

Myths, bad models and paradigms, etc, need to be nipped in the bud. They are extremely difficult to reverse once people have adopted them.

The
question asked was of compact sensors and their theoretical limits.
The answer to the question asked is that when a given sensor size and
lens combination is operating at their diffraction limit, increasing
the megapixel count will not increase the real detail of the capture.

That is not really true, either.

There is no brickwall limit on lens resolution. Resolution fades in contrast.

John Sheehy wrote:

That statement is relying on language to alter reality. Digital
cameras do not break up an analog images into pieces; they BIN
areas of analog focal plane into POINTS of information. The real
resolution of a camera is a combination of the lens, and the sensor
(including AA filter and microlenses).

That has to be the most absurd twisting of words to say the same
thing that I've ever read from you (although I should note that I
don't read everything you write.)

Really? Does dividing a pizza into 8 slices turn each slice into a number of identical pizza atoms? The latter is all we get with digital cameras.

-- hide signature --

John

Graystar Veteran Member • Posts: 8,373
Re: Are those figures for

John Sheehy wrote:

Does dividing a pizza into 8 slices turn each slice into a
number of identical pizza atoms?

There's no such thing as a pizza atom.

Les Olson Senior Member • Posts: 2,081
I ...

am just curious as to what happens when there is no space left for the title of the next reply.
--
2 November 1975.

'... Ma come io possiedo la storia,
essa mi possiede; ne sono illuminato:
ma a che serve la luce?'

bobn2
bobn2 Forum Pro • Posts: 72,009
Re: Are those figures for

Graystar wrote:

John Sheehy wrote:

Does dividing a pizza into 8 slices turn each slice into a
number of identical pizza atoms?

There's no such thing as a pizza atom.

Cheese strings? (Emil knows about strings, he might be able to answer)

-- hide signature --

Bob

Ralf Ronander
Ralf Ronander Contributing Member • Posts: 797
I´m not saying you are wrong...

...but if we have no practical way of experiencing the benefits of the our monster pixel cameras (very few have the means or the equipment the make very large prints on a regular basis), then what´s the point

As a pixel man myself I know there are several good reasons nonetheless, but to many this could be a valid objection.

bobn2
bobn2 Forum Pro • Posts: 72,009
Re: I´m not saying you are wrong...

Ralf Ronander wrote:

...but if we have no practical way of experiencing the benefits of
the our monster pixel cameras (very few have the means or the
equipment the make very large prints on a regular basis), then what´s
the point

As a pixel man myself I know there are several good reasons
nonetheless, but to many this could be a valid objection.

The point is that your original image contains more information which you can use in a number of possible ways to improve the quality of your output. As suggested already, high MP captures will tend to give more microcontrast, because they use a more benign part of the OLPF MTF, and this will be true at any scale because below Nyquist the OLPF function is falling and monotonic. The additional information could be used to allow sophisticated PP to make better decisions to separate image from artifacts, or it could simply be of use in PP cropping. Even if you don't need it now, having all that information in you files is an insurance in the future for unforseen uses of the image.
--
Bob

DaSonyGuy Forum Pro • Posts: 12,386
More pixels are not necessarily better...

Iforgetwhat8was4 wrote:

or are they?

Could someone try to give a clear but comprehensive explanation of
why more Mp doesn't improve IQ?

Given a fixed sensor size the camera with higher "resolution" will produce a physically larger image than one with lower "resolution" but one that is noisier and has reduced image quality, because smaller photosites equal more noise, fact.

I understand that more pixels means more noisy pixels, and i've seen
that in my own cameras.

Well then why do need to ask?

But should twice as many pixels give twice the dynamic range, simply
because you could sample two pixels for a value with twice the range
of a single pixel?

Completely the opposite. A single Bayer pixel is actually the combined output of at least four adjacent photosites, often more...Make the photosites smaller by increasing the "resolution" and it just increases noise whilst allowing slightly more detail to be captured (and I do mean slightly more) its normally at the expense of pixel level image quality....There is no such thing as a free lunch.

Isn't it a question of whether NR is losing detail that wouldn't be
there in the first place with less pixels?

Noise reduction does indeed reduce resolution but you have to trade that off with being able to capture usable shots...A clean low res shot can be more usable than a noisy high res shot...Everything is a compromise.

-- hide signature --
 DaSonyGuy's gear list:DaSonyGuy's gear list
Sony a7C Samyang AF 35mm F2.8 FE Samyang AF 18mm F2.8 FE Sony FE 28-60mm F4-5.6 Samyang AF 24mm F1.8 FE +10 more
bobn2
bobn2 Forum Pro • Posts: 72,009
Re: More pixels are not necessarily better...

DaSigmaGuy wrote:

Iforgetwhat8was4 wrote:

or are they?

Could someone try to give a clear but comprehensive explanation of
why more Mp doesn't improve IQ?

Given a fixed sensor size the camera with higher "resolution" will
produce a physically larger image than one with lower "resolution"
but one that is noisier and has reduced image quality, because
smaller photosites equal more noise, fact.

Brave man, wading into the discussion without bothering to read it, and offering a contentious (and wrong) view. OK then....

Statement 1, wrong. The file produced by a higher "resolution" camera is not physically larger. If by higher resolution you mean higher pixel count (they are not the same thing, since a camera could have higher resolution but a smaller sensor and thus have fewer pixels) then the physical size of the image is dictated only by the output medium. The file produced by the camera has no physical size, it is a collection of bits, which are immaterial.

Statement 2, wrong also - both because small pixels do not produce an image that is noisier and also because higher pixel count increases image quality, all else being equal.

I understand that more pixels means more noisy pixels, and i've seen
that in my own cameras.

Well then why do need to ask?

Fishing to catch chumps who'll make ill informed posts?

But should twice as many pixels give twice the dynamic range, simply
because you could sample two pixels for a value with twice the range
of a single pixel?

Completely the opposite. A single Bayer pixel is actually the
combined output of at least four adjacent photosites, often
more...Make the photosites smaller by increasing the "resolution" and
it just increases noise whilst allowing slightly more detail to be
captured (and I do mean slightly more) its normally at the expense of
pixel level image quality....There is no such thing as a free lunch.

Got that back to front. A single Bayer pixel is a single photosite, receptive to one of three light wavebands. There is no such thing as 'pixel level image quality' - a pixel cannot record an image, it can only record a single value (unless it's a Foveon pixel) to measure the light incident on its own area.

-- hide signature --

Bob

DaSonyGuy Forum Pro • Posts: 12,386
Re: More pixels are not necessarily better...

bobn2 wrote:

DaSigmaGuy wrote:

Iforgetwhat8was4 wrote:

or are they?

Could someone try to give a clear but comprehensive explanation of
why more Mp doesn't improve IQ?

Given a fixed sensor size the camera with higher "resolution" will
produce a physically larger image than one with lower "resolution"
but one that is noisier and has reduced image quality, because
smaller photosites equal more noise, fact.

Brave man, wading into the discussion without bothering to read it,
and offering a contentious (and wrong) view.

Yes, you are.

Statement 1, wrong. The file produced by a higher "resolution" camera
is not physically larger. If by higher resolution you mean higher
pixel count (they are not the same thing, since a camera could have
higher resolution but a smaller sensor and thus have fewer pixels)
then the physical size of the image is dictated only by the output
medium.

Oh dear...I guess I'll have to put it a bit more simply for you...

"Given a fixed size sensor" means the sensor is the same size in both the high resolution and the lower resolution camera.

The output image size is therefore determined purely by how many pixels a sensor has, meaning for all practical purposes pixel count, resolution and image size are one and the same thing.

As an example, lets assume you have a FF sized sensor (36mmx24mm) in both cameras being tested but one has a resolution of 12mp and the other a resolution of 20mp.

The image size produced by the 12mp FF sensor is something like 4000x3000 pixels (Perhaps not exactly in that ratio though because FF might not be exactly 4:3 format(?), but good enough to use as an example.) The image size produced by the 20mp FF sensor on the other hand is something like 5000x4000 pixels..Nearly 17% larger from the same sized sensor.

Which will produce the better QUALITY images...The one with the larger photosites, ie: the 12mp camera.

Which can capture more detail...Obviously its the 20mp camera but because it has smaller photosites its pixels are a lot noisier.

Well then why do need to ask?

Fishing to catch chumps who'll make ill informed posts?

Well he sure caught you, did'nt he.

But should twice as many pixels give twice the dynamic range, simply
because you could sample two pixels for a value with twice the range
of a single pixel?

Completely the opposite. A single Bayer pixel is actually the
combined output of at least four adjacent photosites, often
more...Make the photosites smaller by increasing the "resolution" and
it just increases noise whilst allowing slightly more detail to be
captured (and I do mean slightly more) its normally at the expense of
pixel level image quality....There is no such thing as a free lunch.

Got that back to front. A single Bayer pixel is a single photosite,
receptive to one of three light wavebands.

Your totally wrong...If I were you I would learn how Bayer sensors work before reaching for your keyboard and making embaressing statements.

There is no such thing as
'pixel level image quality'

Wrong again.

  • a pixel cannot record an image, it can

only record a single value (unless it's a Foveon pixel) to measure
the light incident on its own area.

Foveon sensors have much better pixel level quality than any Bayer sensor yet made, irrespective of sensor size or pixel count.
--
DSG
--

-- hide signature --
 DaSonyGuy's gear list:DaSonyGuy's gear list
Sony a7C Samyang AF 35mm F2.8 FE Samyang AF 18mm F2.8 FE Sony FE 28-60mm F4-5.6 Samyang AF 24mm F1.8 FE +10 more
bobn2
bobn2 Forum Pro • Posts: 72,009
Re: More pixels are not necessarily better...

DaSigmaGuy wrote:

bobn2 wrote:

DaSigmaGuy wrote:

Iforgetwhat8was4 wrote:

or are they?

Could someone try to give a clear but comprehensive explanation of
why more Mp doesn't improve IQ?

Given a fixed sensor size the camera with higher "resolution" will
produce a physically larger image than one with lower "resolution"
but one that is noisier and has reduced image quality, because
smaller photosites equal more noise, fact.

Brave man, wading into the discussion without bothering to read it,
and offering a contentious (and wrong) view.

Yes, you are.

Not me, I'm offering a contentious but right view.

Statement 1, wrong. The file produced by a higher "resolution" camera
is not physically larger. If by higher resolution you mean higher
pixel count (they are not the same thing, since a camera could have
higher resolution but a smaller sensor and thus have fewer pixels)
then the physical size of the image is dictated only by the output
medium.

Oh dear...I guess I'll have to put it a bit more simply for you...
"Given a fixed size sensor" means the sensor is the same size in both
the high resolution and the lower resolution camera.
The output image size is therefore determined purely by how many
pixels a sensor has,

Output image size has nothing to do with how many pixels a sensor has. You don't seem to have grasped this.

meaning for all practical purposes pixel count,
resolution and image size are one and the same thing.

And therefore you've got this wrong. Very wrong.

As an example, lets assume you have a FF sized sensor (36mmx24mm) in
both cameras being tested but one has a resolution of 12mp and the
other a resolution of 20mp.
The image size produced by the 12mp FF sensor is something like
4000x3000 pixels (Perhaps not exactly in that ratio though because FF
might not be exactly 4:3 format(?), but good enough to use as an
example.) The image size produced by the 20mp FF sensor on the other
hand is something like 5000x4000 pixels..Nearly 17% larger from the
same sized sensor.

Those are not image sizes, they are pixel counts written a different way. Size has to do with spatial dimensions, the things you measure in metres, feet or inches, not counts.

Which will produce the better QUALITY images...The one with the
larger photosites, ie: the 12mp camera.
Which can capture more detail...Obviously its the 20mp camera but
because it has smaller photosites its pixels are a lot noisier.

It's pixels are, but its images aren't. Now there's a paradox for you. You could try to work it out.

Well then why do need to ask?

Fishing to catch chumps who'll make ill informed posts?

Well he sure caught you, did'nt he.

I think you'll find, if you think about this properly, that it's not me that's been caught.

But should twice as many pixels give twice the dynamic range, simply
because you could sample two pixels for a value with twice the range
of a single pixel?

Completely the opposite. A single Bayer pixel is actually the
combined output of at least four adjacent photosites, often
more...Make the photosites smaller by increasing the "resolution" and
it just increases noise whilst allowing slightly more detail to be
captured (and I do mean slightly more) its normally at the expense of
pixel level image quality....There is no such thing as a free lunch.

Got that back to front. A single Bayer pixel is a single photosite,
receptive to one of three light wavebands.

Your totally wrong...If I were you I would learn how Bayer sensors
work before reaching for your keyboard and making embaressing
statements.

If you're totally wrong, everyone who's right looks wrong. My statement was correct in fact. If you go on disputing it, you could make a big fool of yourself.

There is no such thing as
'pixel level image quality'

Wrong again.

  • a pixel cannot record an image, it can

only record a single value (unless it's a Foveon pixel) to measure
the light incident on its own area.

Foveon sensors have much better pixel level quality than any Bayer
sensor yet made, irrespective of sensor size or pixel count.

Again, there is no such thing as pixel level image quality, because a single pixel cannot produce an image, the image can only be produced by an aggregation of pixels. You could stretch a point and talk about 'per-pixel image quality', which has a reasonably sensible interpretation (image quality, however you would measure it, divided by the number of pixels), but pixel level image quality is an absurdity. People who know more about sensors and image production than me (such as Joe Wisniewski or Iliah Borg) seem to not share your view about the qualities of the Foveon sensor. I suspect if it were so good, it would have been a commercial success. I see one of its major customers has gone bust. Your statement (and your tag) reveals that you are looking at this problem through Sigma coloured glasses ( with unfortunate metamerisms) rather than rationally.

-- hide signature --

Bob

ejmartin Veteran Member • Posts: 6,274
Perhaps you can explain...

... the following then:

http://forums.dpreview.com/forums/read.asp?forum=1018&message=30190836
http://forums.dpreview.com/forums/read.asp?forum=1018&message=30211624

http://luminous-landscape.com/forum/index.php?showtopic=29801&st=20&p=241562&#entry241562

Please use quantitative analysis rather than the erroneous internet mythology you have been spouting.

As an example, lets assume you have a FF sized sensor (36mmx24mm) in
both cameras being tested but one has a resolution of 12mp and the
other a resolution of 20mp.
The image size produced by the 12mp FF sensor is something like
4000x3000 pixels (Perhaps not exactly in that ratio though because FF
might not be exactly 4:3 format(?), but good enough to use as an
example.) The image size produced by the 20mp FF sensor on the other
hand is something like 5000x4000 pixels..Nearly 17% larger from the
same sized sensor.
Which will produce the better QUALITY images...The one with the
larger photosites, ie: the 12mp camera.
Which can capture more detail...Obviously its the 20mp camera but
because it has smaller photosites its pixels are a lot noisier.

-- hide signature --
DaSonyGuy Forum Pro • Posts: 12,386
Re: Perhaps you can explain...

ejmartin wrote:

... the following then:

http://forums.dpreview.com/forums/read.asp?forum=1018&message=30190836
http://forums.dpreview.com/forums/read.asp?forum=1018&message=30211624

Explain?...OK...Never heard of in-camera noise reduction?

Here are some relevent quotes from Phils 50D review to save me the trouble of repeating myself:

"The 50D produces comparatively clean output that demonstrates a well balanced mixture of chroma and luminance noise reduction. However at ISO 1600 and above it can't quite keep up with the D300 in terms of detail retention. The Canon's more heavy-handed luminance noise reduction inevitably leads to some blurring of fine detail. Output up to ISO 3200 is usable, ISO 6400 should be reserved for emergencies and the inclusion of the ISO 12800 setting was quite frankly pointless."

"Canon first introduced 'High ISO Noise Reduction' on the EOS 40D. While the first generation of this feature only offered one level of noise reduction, its latest incarnation gives you three (Low, Standard, Strong) which can be controlled via C.Fn II-2. Of course there is also an 'Off' setting. As you can see from the graph below the 'on' options appear to apply a fairly even mix of chroma and luminance noise reduction at all settings. When noise reduction is set at 'Low' its effect is pretty marginal, the 'Strong' setting leads to a visible loss of detail from fairly low ISOs. The 'Standard' setting seems to be, depending on your preferences, a pretty good compromise to work with. Please note that the 'Strong' setting reduces buffer space in continuous drive and that even at the 'Off' setting the JPG engine still applies some noise reduction to the image. If you want to get completely rid of NR shooting RAW is your only option."

"Switching to our benchmark RAW converter, Adobe Camera RAW equalizes image processing between the two cameras and allows us to get a much better idea of the level of detail actually captured. As you can see both cameras images look crisper and exhibit better detail but the 40D stills beats the newer model in terms of per pixel detail."

Bob does'nt think there is such a thing as "per pixel detail"...Are you still prepared to go along with him now?

The quote continues:

"Despite of a 22% increase in vertical and horizontal resolution the extra detail captured by the 50D is marginal."

Here I disagree...To my eyes the 40D is actually capturing MORE detail, at least going by Phils samples but again, this is probably due to much better pixel level quality, which in turn is due to the 40D's larger photosites.

Please use quantitative analysis rather than the erroneous internet
mythology you have been spouting.

I dont need to because Phil has already done it, right here on dpreview.
--
DSG
--

-- hide signature --
 DaSonyGuy's gear list:DaSonyGuy's gear list
Sony a7C Samyang AF 35mm F2.8 FE Samyang AF 18mm F2.8 FE Sony FE 28-60mm F4-5.6 Samyang AF 24mm F1.8 FE +10 more
ejmartin Veteran Member • Posts: 6,274
Re: Perhaps you can explain...

DaSigmaGuy wrote:

ejmartin wrote:

... the following then:

http://forums.dpreview.com/forums/read.asp?forum=1018&message=30190836
http://forums.dpreview.com/forums/read.asp?forum=1018&message=30211624

Explain?...OK...Never heard of in-camera noise reduction?

Not a property of Canon RAW data, which is what the linked posts are about. You might (re)read them. The RAW data, run through a reasonably good quality RAW converter (DPP was used for the analysis), has detail out to very close to Nyquist.

Here are some relevent quotes from Phils 50D review to save me the
trouble of repeating myself:

[snipped irrelevant diatribe about soft camera jpegs]

To my eyes the 40D is actually capturing MORE
detail, at least going by Phils samples but again, this is probably
due to much better pixel level quality, which in turn is due to the
40D's larger photosites.

Again irrelevant, when the question is what are the capabilities of the sensor rather than the mangling of the sensor data done by the camera jpeg engine. The RAW data tells a different story:

http://forums.dpreview.com/forums/read.asp?forum=1019&message=30001276

Please use quantitative analysis rather than the erroneous internet
mythology you have been spouting.

I dont need to because Phil has already done it, right here on dpreview.

ROFLMAO; Phil hasn't got a clue.

-- hide signature --
ejmartin Veteran Member • Posts: 6,274
To be more specific...

This 50D image, developed from RAW via DPP
http://theory.uchicago.edu/~ejm/pix/20d/posts/tests/Noise/E50DhSLI01600.jpg

when downsampled to the pixel dimensions of the 40D (using PS Bicubic), yields this image:

http://theory.uchicago.edu/~ejm/pix/20d/posts/tests/Noise/E50DhSLI01600-downsamp.jpg

The downsampled image has a noise spectrum much the same as the corresponding 40D image (from RAW via DPP):
http://theory.uchicago.edu/~ejm/pix/20d/posts/tests/Noise/E40DhSLI1600.jpg

The Imaging-Resource multitarget images, developed from RAW via DPP, and run through Imatest, show resolution out to the respective Nyquist freqeuencies; and the 50D, downsampled to the pixel dimensions of the 40D, outresolves the latter by a rather slight amount.

So, contrary to your claims,

1. Careful treatment of 50D RAW images shows resolution exceeding the 40D by the ratio of pixel pitches.
2. Noise of the 50D is the same as that of the 40D at comparable image scales.

3. Resampling the 50D image to the 40D dimensions yields slightly more resolution, with the same amount of noise, and mitigates interpolation artifacts.

And now you've read it on the internet, so it must be true ;-p

-- hide signature --
DaSonyGuy Forum Pro • Posts: 12,386
Re: To be more specific...

ejmartin wrote:

This 50D image, developed from RAW via DPP
http://theory.uchicago.edu/~ejm/pix/20d/posts/tests/Noise/E50DhSLI01600.jpg

when downsampled to the pixel dimensions of the 40D (using PS
Bicubic), yields this image:

http://theory.uchicago.edu/~ejm/pix/20d/posts/tests/Noise/E50DhSLI01600-downsamp.jpg

The downsampled image has a noise spectrum much the same as the
corresponding 40D image (from RAW via DPP):
http://theory.uchicago.edu/~ejm/pix/20d/posts/tests/Noise/E40DhSLI1600.jpg

The Imaging-Resource multitarget images, developed from RAW via DPP,
and run through Imatest, show resolution out to the respective
Nyquist freqeuencies; and the 50D, downsampled to the pixel
dimensions of the 40D, outresolves the latter by a rather slight
amount.

So, contrary to your claims,

1. Careful treatment of 50D RAW images shows resolution exceeding
the 40D by the ratio of pixel pitches.
2. Noise of the 50D is the same as that of the 40D at comparable
image scales.
3. Resampling the 50D image to the 40D dimensions yields slightly
more resolution, with the same amount of noise, and mitigates
interpolation artifacts.

And now you've read it on the internet, so it must be true ;-p

Everyone knows that downsizing an image hides image noise so its hardly supprising your technique of downsizing the larger 50D image to the size of the 40D image makes it look less noisy!...Your technique is therefore completely flawed and the results therefore irrelevant.
--
DSG
--

-- hide signature --
 DaSonyGuy's gear list:DaSonyGuy's gear list
Sony a7C Samyang AF 35mm F2.8 FE Samyang AF 18mm F2.8 FE Sony FE 28-60mm F4-5.6 Samyang AF 24mm F1.8 FE +10 more
Ralf Ronander
Ralf Ronander Contributing Member • Posts: 797
Re: To be more specific...

Not to argue against your conclusions,
but the shots are too diffeent for a meaningful comparison.

Different object sizes (50D downsampled image vs 40D), different lenses and different exposure. Deeper shadows in the 50D image. Different object placement (in the right hand part of the image).

Possibly the latter could be in fovour of the 50D (the scale better placed in the focal plane) but the other factors are mostly in favour of 40D.

Ralf Ronander
Ralf Ronander Contributing Member • Posts: 797
Re: To be more specific...

You are totally wrong.

You really should do a lot of studying before making claims of what "everyone" knows. Especially since you don´t know yourself.

DaSonyGuy Forum Pro • Posts: 12,386
Re: To be more specific...

Ralf Ronander wrote:

You are totally wrong.
You really should do a lot of studying before making claims of what
"everyone" knows.

In thinking everyone was intelligent enough to know it, yes, but you have proved that is'nt the case.
--
DSG
--

-- hide signature --
 DaSonyGuy's gear list:DaSonyGuy's gear list
Sony a7C Samyang AF 35mm F2.8 FE Samyang AF 18mm F2.8 FE Sony FE 28-60mm F4-5.6 Samyang AF 24mm F1.8 FE +10 more
ejmartin Veteran Member • Posts: 6,274
Re: To be more specific...

Ralf Ronander wrote:

Not to argue against your conclusions,
but the shots are too different for a meaningful comparison.
Different object sizes (50D downsampled image vs 40D), different
lenses and different exposure. Deeper shadows in the 50D image.
Different object placement (in the right hand part of the image).
Possibly the latter could be in fovour of the 50D (the scale better
placed in the focal plane) but the other factors are mostly in favour
of 40D.

The use I was making of the two images was mainly for a noise comparison; for that purpose they are eminently suited. If I had the two cameras in hand I surely would have made the exposures, lenses etc equal. But that is irrelevant for noise, so long as one chooses two patches of the image of equal tonality in the two images. I'm not sure why aesthetic factors such as object placement, tone curve etc should matter as long as one is comparing say a white patch of tonal value 200 in one image to a white patch of tonality 200 in the other for the purpose of determining the noise spectrum of each.

A different set of images (the IR multitarget shots) were used for a resolution comparison using Imatest, and showed that both cameras have a similar ratio of MTF50 to Nyquist. For this test it is important that the camera be properly focussed on the test chart; it seemed to be.

Once one has determined that the noise spectrum is identical at comparable scales and the resolution is close to the ratio of pixel pitches, I think it reasonable to conclude that

1. Both sensors have the same quantum efficiency, and therefore the higher resolution sensor is not noisier than the lower resolution sensor, since they are gathering the same number of photons overall in each part of the image.

2. The higher resolution sensor has the potential to deliver higher image resolution in proportion to the linear pixel count, ie the camera does not limit increase of image detail (though other factors such as lens diffraction, camera shake, poor light, etc might).

-- hide signature --
ejmartin Veteran Member • Posts: 6,274
Re: To be more specific...

DaSimpleGuy wrote:

Everyone knows that downsizing an image hides image noise so it's
hardly surprising your technique of downsizing the larger 50D image
to the size of the 40D image makes it look less noisy!...Your
technique is therefore completely flawed and the results therefore
irrelevant.
--

Um, no. And if you want to see that, print all three at a common output size.

The point of the downsampling exercise was to counter the argument that the 50D is a noisier camera than the 40D. Since one can generate an image with identical noise and resolution as the 40D (if that is what one wants) from the image data of the 50D, it cannot be noisier, or have less resolution. The 40D image data is a (proper) subset of the 50D image data.

It is true that, not downsampled, the 50D has high frequency noise that the 40D doesn't. That is because the 40D doesn't have high frequency anything, including image detail, beyond its Nyquist frequency.

-- hide signature --
Keyboard shortcuts:
FForum MMy threads