Is too much Pixel Density Bad??

gdanmitchell · Jan 25, 2008

E.J. said:
For the umpteenth time DIFFRACTION!!!!!!!!
Here's a turorial:
http://www.cambridgeincolour.com/tutorials/diffraction-photography.htm

A follow up. May I suggest that you carefully read the information at the link you provided? In particular, take a careful look at the section titled "NOTES ON REAL-WORLD USE IN PHOTOGRAPHY."

Dan

--
---
G Dan Mitchell
SF Bay Area
http://www.gdanmitchell.com/

Bjoern Gustafsson · Jan 25, 2008

Hi!

Look at the thread "nikon 14-24 on Canon EOS with adapter (see 16-9)" I started in canon forum 1ds3/5d if you are interested in further reading:

The bottomline (maybe somewhat exaggerated): Even with the best lenses (Nikon 14-24 i the wide angle area) 1ds:3 become diffraction limited for all kinds of pictures demanding great DOF
(A typical lexample of course landscaping).

So you are wasting your money bying this camera because you cant use the resolution except for "shalllow" pictures like studiowork and portraits and so on.
That is not what Canon has said in their advertising

The diffractionproblem is now hidden behind the bad quality of canon´s wide angle lenses.

MY solution: Stick to my 5d and buy the converter mentioned in the thread

dnjake · Jan 25, 2008

Tiny digicam sensors do pretty well on resolution even with very high pixel densities. Lood at Phil's g9 test. Its 12 megpixel sensor does not get quite the resolution of the 5D. But its not bad. Clearly if you could manage a lens with the quality of the g9's across a full frame sensor with the g9's pixel density, you would get much better resolution than any current DSLR.
--
David Jacobson
http://www.pbase.com/dnjake

Oscar Latorre · Jan 25, 2008

Not possible,

the AA filter is needed in digital processing. Main reason is that it filters extra noise (high frequency) meanwhle preserving the image details (low and high frequencies of the image). Thus we eliminate the hight frequency of the noise (which occupies all frequencies).

That is in an ideal world. As ideal AA filters do not exist, then some details, the sharp borders of an image (medium-high frequencies) are affected. The image because softer than without AA filter, but otherwise it would be much noisier.

Regards.

Oscar Latorre · Jan 25, 2008

Sorry, a small mistake:

Image -> Low and Medium Frequencies
Noise -> All Frequencies

With AA filter we eliminate high frequencies of the noise, only low and medium frequencies of the image and noise are not filtered.

Regards.
--

Para hacer buenas fotos sólo son necesarias dos cosas: un CAMARÓN y TALENTO. Y para que no haya ninguna duda:

CAMARÓN: aparato enorme que saca fotos.
TALENTO: no tan rápido.

jonrobertp · Jan 25, 2008

most of their files are in the 50's. Interpolation used much of the time. Large size files images are definitely in demand. Alamy has 11 million images for sale.
--
Have a great and wonderful Day !! Cheers !!

See China's Best (new !! ) at http://www.jonrp.smugmug.com

fred vachss · Jan 25, 2008

In this discussion it's worth remembering that image detail and DOF are related through:

DOF

This quantifies what others have been saying about the usability (or lack thereof) of high resolution sensors for landscape work where high DOF is required.

That said, this limit can be beaten with multiple image DOF stacking (i.e. bracket focus in multiple frames at wide apertures to limit diffraction then combine in PP). Play this game and we won't hit diffraction limits until sensor resolutions of well in excess of 100 MP - though realistically the space bandwidth product of most 35mm format lenses is less than 50 MP.

Malcolm L · Jan 25, 2008

This is a combination of 6 images taken with the 5D and the 100mm macro all at f8 to limit diffraction.

Malcolm

James A. · Jan 25, 2008

Nice Shot! How long did the whole process take you?
--
My Gallery: http://www.pbase.com/jeimuzu

Malcolm L · Jan 25, 2008

I've done several dozen similar shots and it gets easier. This one was one of my first attempts and I probably spent 3-4 hours getting it right. I used Helicon Focus which is a good program, but you can also get CombineZM for free and it works nearly as well.
http://hadleyweb.pwp.blueyonder.co.uk/CZM/News.htm

The big problem is that the image size changes as you move the focal point. This results in some ghosting but the built-in editing tools serve to reduce this. I always end up doing a lot of manual editing in CS3, but the end result is always worth it to me. You should see this printed large. Everyone comments on it.
Malcolm

Nathan Yan · Jan 25, 2008

E.J. said:
Nathan Yan said:

Noise aside, there are no disadvantages to more MP. However, at a
certain point you would reach diminishing returns because current
lenses may not be able to resolve that much detail. It will never
make things worse, but it won't help much in the detail department
after a certain point.

Click to expand...

Yes there are, diffraction being the primary problem.

No....

How do more megapixels make diffraction worse ? You can always resize the 100MP image to a 25MP image, and the "diffraction" is the same - it's a property of the lens.

Bjoern Gustafsson · Jan 25, 2008

Yes the picture was wonderful. I knew stacking existed but not that it could produce such results.

What about some intelligent macroprogram which could handle stacking, HDR and mergering pictures for panorama at he same time. The demands for static pictures should be pretty high-anyway do you thing it will ever be created or is it just a fantasy

Regards,
Bjoern G

E.J. · Jan 25, 2008

Each pixel is its own lens and has diffraction properties associated with it.

You may want to read this for a better understanding:
http://www.cambridgeincolour.com/tutorials/diffraction-photography.htm

--
EJP

E.J. · Jan 25, 2008

But the point is still that as you increase the pixel count in a fixed area, the diffraction limit for aperture gets worse. The assertion in this thread is that you can just keep adding pixels forever and the only negative is noise and file size is just false. Certainly the extra sampling will reduce that effect but there is a limit, not limitless like so many here are claiming.

--
EJP

bobn2 · Jan 25, 2008

--
Bob

fred vachss · Jan 25, 2008

E.J. said:
Each pixel is its own lens and has diffraction properties associated
with it.

You may want to read this for a better understanding:
http://www.cambridgeincolour.com/tutorials/diffraction-photography.htm

--
EJP

Umm... Are you talking about microlenses here or some other lens of which I'm unaware. Microlenses have field of view issues, but I don't think their diffraction limited resolution plays any significant role in the resolution of the overall imaging system.

BTW. I thought I kinda understood diffraction (having been in the field for the last 20 years), but I'm always prepared to learn something new.

E.J. · Jan 25, 2008

In the linked article in the Visual example section, pick a spot size, say the f/11 one and then scroll through the different sized pixels by holding your mouse cursor over each one - I think it makes for a good illustration of how the pixel size is important in the debate.
--
EJP

fred vachss · Jan 25, 2008

E.J. said:
In the linked article in the Visual example section, pick a spot
size, say the f/11 one and then scroll through the different sized
pixels by holding your mouse cursor over each one - I think it makes
for a good illustration of how the pixel size is important in the
debate.
--
EJP

Oversample a blur spot and the result isn't any blurrier. I certainly agree that you will eventually reach a point of diminishing returns, but oversampling won't make the resolution worse.

Just to be clear: Say your diffraction limited spot is 10 um in diameter. Imaging this onto 10 um pixels or four times as many 5 um pixels will give you roughly the same sized spot, but the 5 um pixel result won't be any larger.

fred vachss · Jan 25, 2008

E.J. said:
But the point is still that as you increase the pixel count in a
fixed area, the diffraction limit for aperture gets worse. The
assertion in this thread is that you can just keep adding pixels
forever and the only negative is noise and file size is just false.
Certainly the extra sampling will reduce that effect but there is a
limit, not limitless like so many here are claiming.

--
EJP

I really don't understand this point. Please give me a bit more detail about the mechanism of this deterioration.

E.J. · Jan 25, 2008

Yes I understand that. In the calculator for diffraction limits at the bottom of that link, enter 21MP, 35mm, and f/8 - the calculator comes back with a YES for diffraction limited. Now just change the calculator to 16MP and it now says no. Certainly whether or not the difference of an identically sized print from the two is noticeable or not is unlikely.

But try this, take a 1DsII and shoot a controlled scene at f/8, f/11, f/16, f/22 and then compare the RAW file sizes. Canon compresses (losslessly) its RAW files so the file size should be smaller if you are starting to lose detail. On a 1Ds2 you will just start to see a very slight file size reduction at f/16 and more at f/22. On the 1Ds3 you actually just start to see the file size reduction at f/11 and by f/22 the file size reduction is more than 10%. The only way this can be (assuming all other things equal) is that you are starting to lose resolution due to diffraction. The fact that this starts to occur at a wider aperture on the 21MP camera over the 16MP camera is telling. At least to me it is

We have actually done this test in an extremely controlled situation and it is repeatable.
--
EJP

Is too much Pixel Density Bad??

Veteran Member

Well-known member

Senior Member

Forum Enthusiast

Forum Enthusiast

Forum Pro

Senior Member

Senior Member

Senior Member

Senior Member

Senior Member

Well-known member

Leading Member

Leading Member

Forum Pro

Senior Member

Leading Member

Senior Member

Senior Member

Leading Member

Keyboard shortcuts