I know your not a "full test" kinda site...or kinda guy, but one of the things that really burned me on my first outing when I purchased my D2X is soft pictures even using proper techniques. Went on an outing to Yosemite and came back with better than half of my pictures unusable due to softness (using proper technique....MLU, Tripod, cable release...etc) Very little printable....and most reduced to bearly acceptable web postings due to the hiding effect of downsizing.

It was my uncomfortable introduction to the effects of pixel density and how it affects diffraction.

Currently I dont shoot stopped down any more than f/11, and I am guessing that with FF....at 21MP....your getting pretty close to the same pixel desnity of a 10MP camera with a DX sensor which would limit you to close to the same.

Since this is a real world limit to quality output, have you either tested or just plain noticed where diffraction starts to be a detriment with the new Canon your playing with?

Currently I am a Nikon guy...but I am sure I could expect similar numbers from any Nikon offerings that might have close to the same pixel count....and that would weigh heavily in a decision to purchase a D3 or a D3X (or whatever they might call their competition for the 1Ds MKIII when it comes out)

So if you ended up noticing any limitations in that arena...any comments you might have would be greatly appreciated.

Roman

I think that Michael will answer to you better than I, by whom he has already had the 1DSIII in the hands.

But I believe that canon is not going to present a product, that in quality of image, is bad to a 1DS2 or one 5D, and probably approaches much the result of a P30,

Canon in the technical report spoke of an increase in the capture surface of pels, with respect to the 1DS2, evidently, as Michael says, will have to use the best lenses, so that any defect will be intensified by the hi-res.

Of all way, like all the others, we are to the expectation of which they give the new model to us, to make a personal test, of the yield each one is able to obtain, I believe that is only vay to obtain the peace of mind.

F 11 is likely the diffraction limit with the IIIs, as it is with the IIs (and with the P45+ back on MF).

I strongly recommend you go

http://www.cambridgeincolour.com/tutorials...photography.htm

While it does not deal with the IIIs, it is the best explanation I've seen on diffraction limits. Further, I have tested extensively with a variety of high end cameras and lenses and find the conclusions in practice are precisely the same as the theory!

I'm often surprised how many people routinely stop down a 16 megapixel camera to f32 without realizing they are then reducing the image resolution to that of a camera with half the number of pixels - or less. OK if you only want 8 by 10s, but it's death to 16 by 20s.

Bill

Thanks everyone.....

I am pretty up to date on what diffraction is......and even have that site bookmarked. I had to learn after my Yosemite fiasco.

JUst been bounding aroud the idea that while I want lots of megapixels...right now my D2X is limiting due to hits lack of usable DOF when I put long lenses on...limiting my creativity.

The more I think about it...the more I might actually just get a D3 and hot rod it by getting the AA filter removed. Maybe getting some Zeiss glass and maximizing my resolution that way and still have my more creative DOF.

I can use the saved money to do the mods and put into the glass.

Roman

If you need DOF and feel stopping down degrades resolution too much, then regardless of camera brand, the only single-capture solution is to use lenses with tilt (and that won't help in all situations). Otherwise, make composites like the photographer of the Cambridge in Color site.

Well...that would be a great option if there were tons of tilt lenses for 35MM units. And yes I do understand that Horseman makes a set of tilt bellows for 35MM based systems. But then photography is getting too complicated and I would find myself fighting the process and losing the moment....this would also negitively impact portability.

Thanks for the suggestion....but not a good fit.

Still interested in the answer my origonal question....it would go a long way between choosing an option that works.

Roman

Roman,

If you read Sean McHugh's (Cambridge in Color) essay carefully, you will we that diffraction is a property of light and not the sensor. The resolution of the image projected on the plane of the sensor is the same for a D3, D2x, or D70. The following quote explains the issue well.

"Are smaller pixels somehow worse? Not necessarily. Just because the diffraction limit has been reached with large pixels does not mean the final photo will be any worse than if there were instead smaller pixels and the limit was surpassed; both scenarios still have the same total resolution (although one will produce a larger file)."

His diffraction calculator assumes the system is diffraction limited when the diffraction spot (Airy disc) diameter equals the circle of confusion as shown below. The pixel size does not enter into the equation.



The full resolution of the 1DsMIII will be obtained when the diffraction spot is no larger then the pixel size, which is 7.2 microns. At apertures smaller (larger f/number) than f/5.6 the resolution resolved by the camera will fall, but will not be worse than what would be obtained by the 1DMIII. By way of comparison, the pixel size of the D200 is 6.1 microns and the D2x is 5.5 microns. However, the cropped sensors have more depth of field. With my D200, optimum aperture for a diffraction limited lens is also f/5.6.

Bill

Roman,

There is no solution to your problem, as I wrote to you months ago on DPreview already.

It is impossible to get at the same time perfect sharpness and infinite DoF with any of the current high end digital solutions, regardless of the size of the sensor. The closest thing is the D2X that you have.

The only solutions around this are tilt lenses of DoF stacking, both do indeed take time.

The good news is that:

1. Infinite DoF is not always a key requiement for succesful landscape photography,
2. Taking more time is often a good way to take better images.

Cheers,
Bernard

Bjanes.....So...you’re saying photo site size has nothing to do with diffraction?

When light (in whatever conditioned form) meets with the sensor and is collected....is where the rubber meets the road.

If all other things are equal....and you put a sensor into the same body and lens combination that has tighter pixels than another...it will experience diffraction earlier than the other sensor in the flow of f/stops. (starting at wide open...heading toward closed down) I understand that what kind of diffraction hits the sensor plane is created by the lens.....distance from the sensor plane...etc all come together. But we are dealing with pro bodies in 35MM systems that have fixed variables in that arena....and the only variable is the sensor placed in the system.

When that becomes the ONLY variable......then sensor (pixel size) size has everything to do with it.

Then we stop dealing with what happens to the light and change the variable of what the light is being broadcast upon.

It will be why Nikon will use the pro body the current D3 is using and I can GUARANEE you that when they put the sensor of 20 something MP in it ....I can guarantee you that it will be more diffraction limited than the exact same body with the 12MP sensor in it.

And while all the math is nice to understand WHY it is happening....I am more interested in results than math.

Simple problem.....

I don’t like being limited to such a small aperture....as anytime I put a long lens on my D2X to compress a scene....I cant get to an aperture that allows that to happen.

My D2H with 4MP...didn’t have that problem.....the only difference was the sensor.....same lenses...same focal plane....same everything.

I just want to find the best balance......no hocus pocus.....just what works.

I will be renting them both (the D3 and its big brother when it comes out) and I can place good money that it will follow suit with every other camera that has come down the pike in a fixed set of variable. The D3 will allow a tighter aperture than the big brother camera will. If you would like to place money on it ....I will be happy to take a bet to the opposite if you like.

So ...this leads me back to my original question.....not about the theory...but about practical application and where it lies with the current 20+ MP offering from Canon.....as I will guarantee you that the 20+MP offering from Nikon will have similar performance (might not be exact...but similar)....and it really is that simple after all the smoke mirrors and math have left the scene.

Roman

Not looking for a solution....just the best compromise. My goals are realistic. I am guessing that the extra elbow room of the pixels in the D3 at the same amount...will be a solid gain. And I am also seeing it might be a better option than more megapixels that reduce themselves to 12MP of resolution when I need to stop down....might as well spend less money and just have the 12MP option as my balance.


I dont mind time AFTER the shot....I am only talking about reaction time to light out in the field.


Never said it was.....but there is NOTHING wrong with trying to find the best balance you can in existing hardware....now is there? :~)


And a good way to miss the light when it comes and goes. Would rather have a simpler system and be ready....than have a cumbersom one and be fumbling with a settings while the light dissapears.


Roman

There really is no difference than with film - except that maybe (and only maybe) film has lower resolution than current top end digital sensors (meaning perhaps you didn't always see it - grain can improve the sense of acutance) and bearing in mind that crop sensors need more magnification for a given print size so worsening the impact of diffraction effects. If you stop down beyond the 'diffraction limit' for a given sensor you get the same result that you would have got with film - if it could record the information. Ultimiately it's a trade off between depth of field and overall sharpness.

By way of interest I have a copy of Galen Rowell's Mountain Light that I was completely enamoured with when I bought it some years ago. Now looking at it I still love a lot of the pictures, but on inspection most are not all that sharp:) This may be due to the print process, but equally Galen often used small apertures to get depth of field and so it's possible that diffraction effects reduced the absolute resolution and sharpnes of some of his work.


Mike

And you do bring a solid piece of the puzzle. With Galen…he DID stop down….he was also notorious for using “cheaper” lenses (lighter) for his “chasing the light” runs which also adds to the equation.

The other point you inadvertently bring up is…while quality of the actual capture is a solid goal. The quality of light…the emotional power of the capture ( a quality composition with solid light and a compelling subject) will often overweigh ultimate sharpness. These are all part of my assessment of where I want my equipment to go.

If you can’t tell yet….I also have a copy of Mountain Light…and have been enamored with Galen as well.

I agree totally with your points….and it is why I am laboring over where the best balance lies with the wonderfully advanced equipment we now have before us.

Initially (even as recently as a few weeks ago) my focus was on more MP. We have technologically gotten to a point were we need to find the balances more than just go for the obvious marketing….and why I might actually be considering the D3 over any bigger brother.

I have a hunch as well that I might also be dealing with a much better capture of light at each pixel at the 12MP FF option…..maybe helping in the micro contrast quality of light in my captures. I am thinking that diffraction dims that down a bit as well when you start getting to the stronger end of the problem. Kinda like 14bit vs. 8 bit…the finer the graduations…the better the end result…especially in larger prints.

Now that last part…was just a hunch..no math to prove it…just my basic vision of what is going on.

Roman

No, the effect of pixel pitch is exactly the OPPOSITE of what you are thinking it is. Given the same f-stop and analog diffraction, the bigger pixels obscure the details of the subject more. The diffraction limit is matched as a ratio quicker with the smaller pixels, but that is totally irrelevant, as big pixel size is a bigger obstacle to resolution. This is also true of things like motion blur. People wrongly assume that because any given analog blur covers more pixels, that you need faster shutter speeds to get the same quality of stopped action, but again, the bigger pixel only spreads more confusion over the image, and actually forces the blur to be wider. Motion blur is less blurry with smaller pixels, AOTBE, and can be deconvolved with greater resulting resolution.

Basically yes. Go back and re-read the quote from Sean McHugh several times. If you want to get full resolution of the 1DsMIII you will have to use an aperture of f/5.6 or so, but you can stop down further and still get the same resolution as with the original 1Ds.

The smaller pixel size of the new camera just gives you more options. You can have the higher resolution if you need it and use proper technique. The same considerations apply to film, but since the film can't resolve at the level of the 1DsMIII, the differences are less apparent.

Bill

Ok...I am going to put this to the test.....I need some information.

Currently I have a D70 (6MP DX sensor) and a D2X.

Not a biggy on math...so if somone could help me understand what I should stop my D2X down to to put out the same quality of my D70 at say F/8.

I would love to shot both side by side and see if the diffrence has any more of an effect than just a decrease of resolution. (see if there is more to this degredation than the math shows)

Roman

The pixel spacing of the D70 is 7.8 microns and that of the D2x is 5.5 microns. The D200 is 6.1 microns.

Diffraction spot sizes for green light (530 nm) are 2.6, 3.6, 5.2, 7.2, 10.3, 14.2, 20.7 and 28.6 microns for f/2, 2.8, 4, 5.6, 8, 11, 16, and 22 respectively. When the diffraction spot is larger than the pixel size, MTF falls

For resolution figures with the D200 and the Zeiss Makro-Planar f/2, look at this link:

http://www.photozone.de/8Reviews/lenses/ze..._50_2/index.htm

Best resolution is at f/4 where the diffraction spot is 5.2 microns. That would be near optimum for the D2x as well. With the D70, you could use f/5.6 or slightly smaller. For your tests I would use f/4 through f/16. The differences can be rather subtle and you will have to use meticulous technique and a very good lens to see differences shown on the Photozone site.

As a general rule, I think that Thom Hogan's recommendations work fine. I think you are worrying about differences that will not show up in practical field photography.

Bill

System resolution is essentially a product of the sensor resolution and the lens resolution. There are no sudden cut-offs. If you keep increasing the resolution of one, say the sensor, whilst keeping the resolution of the other, say the lens, unchanged, then the increases in system resolution will get smaller and smaller for each additional percentage increase in pixel count.

The 1Ds3 has around the same pixel density as the 20D. I can confirm that a 20D at f22 produces a more detailed and sharper image than the 5D, using the same lens from the same position, but cropping the 5D image to the same FoV as the 20D image and then interpolating the 5D file size to that of the 20D.

If I'd been using a 1Ds3 instead of a 20D, there would have been no need for cropping.

You can see the results in the following thread. My post is the last.

http://luminous-landscape.com/forum/index....anon+100-400+is

This is one of those things where I believe, perhaps naively or conveniently, there's no clear answer. For me the question is too subjective: what makes the nicer picture, diffraction-limited DOF or DOF-challenged sharpness?

I think each one of us will have a particular leaning, and may also depend on the specific shot. For me I feel sharpness is important, but not the only and often not the most important factor. I feel we have become too obsessed with sharpness, perhaps because it's easier to measure, and tend to play down or forget other factors which I believe often have more impact at the emotional level that (to my again naive or convenient perspective) is the main thrust of my photography.

With my 1Ds II and Zeiss / Leica lenses, I usually shoot at f16 / f19 because the DOF is often what I prefer and am willing to forgo the sharpness for the DOF.

But that's for me, and as always YMWV ...

I am afraid you have the wrong system :-), but as you probably know, Canon has 24, 45, 90, and since you mentioned him, Galen Rowell stated somewhere that 90% of his best images could have been made with 24mm or 85mm.

Roman,

I only know some bits of your work, but your most successful images among those I have seen were all shot in a misty atmosphere where light was probably not changing at all.

Cheers,
Bernard

It is very important - as the Cambridge site acknowledges - just what size prints you're going to make.

I tested my Mamiya/P45+ combo using the standard 80 mm lens at F8, 11, 16 and 22. At 1:1 on my computer screen, the shots at f8 are the sharpest, with f11 being very, very close. f16 is somewhat degraded and f22 is simply not sharp.

BUT...I then printed 8 by 10s and you really cannot see the difference in ANY of the shots.

THEN...I did 16 by 20s - and what I had seen on my computer screen showed through again. However, the f16 shot looked quite acceptable - until you put the f11 print beside it!

The moral of all this is simply that if you're shooting typical "book size" prints, as opposed to exhibition prints, you are not going to see the difference.

My recommendation is to test, using your most used lens and subject matter and print. Then act accordingly.

The morality is also, if you're shooting typical 'book size' prints, you don't need a 1Ds3. The Olympus E-3 should be more than adequate.

There's a fallacy which started some years ago on a Canon EOS forum with a thread titled, "The D60's sensor is too small to stop down below f11". It was supported by some impressive mathematical formulae from the OP relating to DoF and resolution and the thread became the longest thread ever, on this particular forum. It was a fine example of the ancient Greek story about a group of philosophers arguing throughout the night about the number of teeth a horse has, until some bright spark in the early hours of the morning suggested the group take a brief amble across the road to a field where a horse was peacefully grazing, and actually count the teeth. The matter was resolved.

But we don't want that to happen, do we!

With my D2X I consistently get a good print up to 30"x45"....with the process I have gone through great lengths to perfect.

I don't want to print any bigger....just wondering how much more detail I can get for that largest size I offer to my customers.

Roman