d3x diffraction limited at f8?!?

rayman 2 said:

Iliah Borg said:

D3X allows to register the diffraction at the aperture values where
lower resolution cameras fail to register it.

Click to expand...

True but thats only right for diffraction limited lenses.
Otherwise you register things like aberation more then diffraction
see this wiki...
http://en.wikipedia.org/wiki/Diffraction_limit

Click to expand...

--

Diffraction limited means the lens is accurate to 1/4 wavelength of yellow-green light, some mfgs will cheat and do it in red light which means the tolerance is slacker because the wavelengths are longer. Most camera lenses are no where near diffraction-limited wide open, they must be stopped down, then the lens "becomes" diffraction limited, i.e., the lens throughput is no longer effected much by the aberrations present in the wide open lens.

However, true lens resolution suffers because lens resolution is proportional to the diameter of the clear aperture. In other words, a 4" wide aperture can resolve 2x as well linearly as a 2" wide aperture. So, the lens produces a sharper image stopped down by controlling the aberrations, but at some point the loss of resolution from stopping the lens down becomes visible.

I don't get it. Diffraction limitations are tied to the medium recordng the image.

Bottom line.

Again, my D700 looks great at f/16. D300, not so much. Same lens.

I used my Olympus because the small sensor starts to see the effects of diffraction sooner than my D300. Top left to bottom right, f3.5, f5.6, f8, f11, f16 , f22

Although diffraction simply means the bending of light around a sharp edge (diaphragm blade edges). As the diaphragm gets smaller a greater percentage of the light entering through the aperture is refracted - causing blurring. However, with digital sensors (as well as with film) the physics of light also enter the picture - as well as the concept of airy discs and circle of confusion (and the chosen parameters). As pixels get smaller, it becomes more difficult for these pixels to contain this information (airy discs). Now diffraction mixes with airy discs and pixel size. That's what I understand diffraction limiting to be. Pixel density IS part of the equation.

Scouting around I found this site. It explains it much better than I can - and includes many illustrations and a calculator. http://www.cambridgeincolour.com/tutorials/diffraction-photography.htm
--
Steve Bingham
http://www.dustylens.com
http://www.ghost-town-photography.com

Thank you for that link. I have read it many times myself - mostly researching why Ansel Adams liked to shoot at f/64 and why my beloved (and hardly used now) D70 didn't like f/64.

Thom Hogan said:

Not so fast. I've got huge landscapes with much better resolution
than the D3x is capable of that were shot with a 6mp camera. This is
my point: people are too much looking for the "all-in-one" solution.
Some will be better served with stitching a D3 at f/11 than trying to
work around the diffraction on the D3x.

Click to expand...

Stitching has its own problems, which can often be worse. Images from a stitcher such as PTGui or AutoPano Pro won't quite look the same as shooting the same FOV with a single shot, at least not at wider FOV's. You're also likely to hit DOF problems sooner due to the longer focal lengths, although this can be fixed by changing the focus point between shots if you're careful and the scene lends itself to that approach (no trees sticking up in the foreground, etc).

I use stitching when I can. I've found the PC lenses to be particularly useful for this since I can do a quick 3-shot sequence and then flat-stitch them in Photoshop to get about the same resolution as a D3x shot. But the margin for error is much slimmer when stitching, as you have to worry about wind, movement, changing light, etc. Sometimes there's just not time as things are happening too quickly. Plus, the workflow becomes a lot more involved, especially if you're also bracketing for HDR/tufuse/etc. So yeah, having a D3x to reduce the need for stitching would be really nice, just not at $8K.

--
Jeff Kohn
Houston, TX
http://www.pbase.com/jkohn
http://jeffk-photo.typepad.com

Rigth and I could wear camos stay flat on the ground for two, three days and shot wildlife with my 100mm, Nikon and others have really to stop ripping us with all those 400, 500 and 600mm lenses,

My dear Thom I (we) have understood your opinion on D3X price but:

In UK , D3x is about GBP 4570 ($6500) + 17%vat = GBP 5350 = Eu 5500 : I paid EU 4800 for D3 on Dec 2007 , NOT so big a difference : early adopter always been ripped and always will have to pay more for the last toy .

The only FAIR comparison (full pro cam : best body , best seal, best AF , fast and long mileage mechanic) is with Canon 1DS M3 , the other cameras fall short one way or another.

This new camera is NOT a "Jack of all trades" but this was declared from the start, you are the universally aknowledged Nikon gear teacher, we expect from YOU to teach US when to use it and how to get the best results. (Nelson said, before each battle , "England expects every man will do his duty" ;-) ) .

My very best regards

Livio

A major note worthy factor is Adams shoots with a large format, north of 4x5 at many times. Lenses made for 8x10 cameras would be wide open at f8 or smaller. f64 is not that far fetched.

To the OP: Defraction also relate to the lens design a lot more than the sensor's. A 50 1.4 and a 500 f8 would not have the same signs of defraction as the lens stops down.

MyOatmealIsLumpy said:

Thank you for that link. I have read it many times myself - mostly
researching why Ansel Adams liked to shoot at f/64 and why my beloved
(and hardly used now) D70 didn't like f/64.

Click to expand...

--

I've cranked the D3x through my Barnack application and f/11 seems like an reasonable upper limit.

Photosite frequency contrast (at 84 lp/mm) drops to 50% at f/9. Taking the somewhat more pessimistic approach of looking at 2x2 Bayer blocks (RGGB quadruples) is thus 2 x f/9 = f/18.

So, if 50% contrast is an acceptable threshold for getting proper "credit" for pixels at given pixel pitch, then the diffraction limitation would lie in the range f/9 - f/18, depending on the colour of the subject and the quality of the demosaicking algorithm employed. I would go for f/11 as a practical limit.

Notice that this doesn't take the AA filter response into account. Typically, this reduces contrast quite a lot when moving towards the Nyquist limit, just like any remaining abberations would do in a real-life situation.

However, f/11 remains a practical hard limit if per-photosite detail is of interest. Which it should be often, why else bother with shelling out $8K for a 24 MP monster like the D3x.

On the other hand going above f/11 also have its merits, e.g. for DoF requirements (equal bluring over a range will be perceived as "sharp" since sharpness perceptually is relative), of for time blurring reasons (think misty waterfalls, and ponder over why the G10 with f/8 as max includes a 2-stop ND-filter).

The graph below shows the photosite frequency as a grey vertical line and the Bayer frequency as a green vertical line. The red line is the contrast at f/11 (see the slider 2nd from the top):



Notice that 50% contrast is approximately between the two vertical lines.

Another more extreme approach is to see when the D3x sensor will have absolutely no detail at photosite level. That is the MTF 0% figure below:



For most humans however, contrast below 9% cannot be perceived. Thus, at f/18 no visible photosite contrast can be obtained, even with the imaginary construct of a perfect lens.

If you like, you can play around with the figures yourself for the D3X and a ton of other cameras:
http://stegmann.dk/mikkel/barnack/

Best regards,
Mikkel

Thom one thing that I am intrigued about, is it that for instance does the D3X need a better lens to exploit its potential than say the D3, or whatever, That is, it is "performance limited" by optical quality or would it actually look worse than with a lesser camera given indifferent lens quality, which as what seems to be implied with some writings on these issues. If the answer is a page of maths with circle of confusion calculations, feel free to ignore the question. feel free to ignore the question anyhow, i appreciate you have other things to do!

Atto

Thom Hogan said:

of it against each other in order to achieve the results we seek. The
D3x is a more stellar camera within a more narrow range than most
previous Nikon bodies. And it's not very forgiving of mistakes,
whether than be in handling or in sticking a poor lens on it. But
then, some people learn from making mistakes ; ).

--
Thom Hogan
author, Complete Guides to Nikon bodies (19 and counting)
http://www.bythom.com

Click to expand...

dchphoto said:

Why reduce the number of stitches? If a scene you want to cover can
be done in five d700 stiches at any given focal length, then why not
do five d3x stitches?

Click to expand...

The question here is "how many pixels do you really need?" Some people seem to think that bigger is better, period, and yes, they'll do the same number of stitches with a D3x. But as you add megapixels, you start really taxing your computer and software. I've already got D70 stitches that'll bring my computer to a standstill for a day. Throw the same number of D3x images into a pano stitch and you might not hear from me for a week or two ; )

--
Thom Hogan
author, Complete Guides to Nikon bodies (19 and counting)
http://www.bythom.com

Thom Hogan said:

dchphoto said:

Why reduce the number of stitches? If a scene you want to cover can
be done in five d700 stiches at any given focal length, then why not
do five d3x stitches?

Click to expand...

The question here is "how many pixels do you really need?" Some
people seem to think that bigger is better, period, and yes, they'll
do the same number of stitches with a D3x. But as you add megapixels,
you start really taxing your computer and software. I've already got
D70 stitches that'll bring my computer to a standstill for a day.
Throw the same number of D3x images into a pano stitch and you might
not hear from me for a week or two ; )

Click to expand...

Well, quite obviously you need a bigger computer, or a processor farm...need to get your priorities right

--
Bob

Livio Spallone said:

Rigth and I could wear camos stay flat on the ground for two, three
days and shot wildlife with my 100mm, Nikon and others have really to
stop ripping us with all those 400, 500 and 600mm lenses,

Click to expand...

Here in the US, you'd be arrested for approaching many animals that closely ; )

Lisperit said:

My dear Thom I (we) have understood your opinion on D3X price but:

Click to expand...

I suspect you haven't understood my opinion on the D3x price. Most haven't. I haven't objected to the price. I've objected to releasing, marketing, and supporting the camera as Nikon did at the price they did. And the reason I objected is exactly the reason why I've got hundreds of emails in my InBox that say "is Nikon nuts?" I'm sure Nikon's got them, too ; ). If Nikon wanted to price the D3x at US$7999 here in the US, they needed to set user expectations correctly and they needed to seriously improve their customer handling at that level. They did neither.

Lisperit said:

The only FAIR comparison (full pro cam : best body , best seal, best
AF , fast and long mileage mechanic) is with Canon 1DS M3 , the other
cameras fall short one way or another.

Click to expand...

The D3x is a better camera than the 1DsIII by a small but clear margin. But the 1DsIII is also a US$1500 cheaper camera here in the US. And that "best body, best seal, best AF, and other stuff you cite is exactly the same in a body that sells for US$4100 at the moment. Thus, there is a US$3600 cost associated with the change in image quality. That's a big premium (and not likely to hold), and Nikon did LESS in promoting and marketing and supporting this new camera than the one without that premium.

Lisperit said:

we expect from YOU to teach US when to use it and how to get
the best results.

Click to expand...

Working on it. First pass: stay at or below ISO 1600 and f/8 and use only a handful of lenses that the "right stuff." Don't yet know what the best exposure and post processing regime is. (And before someone replies "but I can shoot at ISO 6400 handheld at f/11 and get good results, please re-read the quoted bit: "best results.")

--
Thom Hogan
author, Complete Guides to Nikon bodies (19 and counting)
http://www.bythom.com

bobn2 said:

Well, quite obviously you need a bigger computer, or a processor
farm...need to get your priorities right

Click to expand...

I could easily put together the processor farm, as it's easy enough to rent one these days. The problem is the software, not the hardware. Nothing's changed in my 30+ years in high tech: it always devolves to software.

--
Thom Hogan
author, Complete Guides to Nikon bodies (19 and counting)
http://www.bythom.com

Thom Hogan said:

use only a handful of lenses

Click to expand...

Most of those, as far as I can see, are diffraction-limited at f/5.6 and wider

--
http://www.libraw.org/

Merging multiple images with different focal points - software to do so and PS technique - http://www.photoacute.com/press/pressrelease20Jan2008.html
http://www.photomigrations.com/articles/0603200.htm
Will

William Carson said:

Merging multiple images with different focal points - software to do
so and PS technique -
http://www.photoacute.com/press/pressrelease20Jan2008.html

Click to expand...

Do they have full-size examples from a high resolution camera, together with the source files?

--
http://www.libraw.org/

Thom Hogan said:

bobn2 said:

Well, quite obviously you need a bigger computer, or a processor
farm...need to get your priorities right

Click to expand...

I could easily put together the processor farm, as it's easy enough
to rent one these days. The problem is the software, not the
hardware. Nothing's changed in my 30+ years in high tech: it always
devolves to software.

Click to expand...

People always blame the software, you just need to add another couple of hundred processors to the farm!

--
Bob

First pass: stay at or below ISO 1600 and f/8 and use

Thom Hogan said:

only a handful of lenses that the "right stuff." Don't yet know what
the best exposure and post processing regime is. (And before someone
replies "but I can shoot at ISO 6400 handheld at f/11 and get good
results, please re-read the quoted bit: "best results.")

--
Thom Hogan
author, Complete Guides to Nikon bodies (19 and counting)
http://www.bythom.com

Click to expand...

I ask because a lot of my studio/location portrait/model/kids nursery shoots that I currently use a D2x for are shot at and beyond f/11 to maintain sufficient dof.

I've kept with the D2x thinking that one day something like the D3x (once the price becomes more realistic new/used) would end up being the replacement,

I commonly soften parts of the image afterwards but I want max quality for those areas that I demand be in focus....are we saying that the niche for the D3x is not going to accommodate such studio shots if you want - as you say - best results?

--
http://www.tigershootsback.co.uk

William Carson said:

Merging multiple images with different focal points - software to do
so and PS technique -

Click to expand...

I think you're missing the point. While software that does depth of field stitching is available, when we're talking about multiple stitched images (pano, HDR, DOF) you need software that can distribute the processes in parallel amongst a great number of processors. I don't know of any. Indeed, I'm reminded of where we were back in 1975 when I was using a CDC6600 running SPSS. The core processor had 16KB of memory. I had several megabytes of Arbitron and Neilsen data. Guess how long it took the world's fastest computer to run my statistical analyses?

--
Thom Hogan
author, Complete Guides to Nikon bodies (19 and counting)
http://www.bythom.com