D2X and DOF calculators

[diglloyd]

[ QUOTE ]
Lloyd,

I looked at your layered DOF comparison on DIgital Outback. Indeed the difference is amazing. But I am wondering what the situation would be with the 1Ds II if one used Canon's TSE lenses and applied tilt. This would seem to increase the Canon's DOF well beyound what you have shown and possibly exceed the D2X.

[/ QUOTE ]

Yes, of course tilt changes the game, if the subject is of the right "shape" for tilt to work its magic of shifting the plane of sharp focus to coincide with the subject/sensor.

I have personally used the Canon 24 TS, and have been very unimpressed. Tilt or not, it's not an impressive lens, looking pretty nasty when fully shifted. That leaves the Olympus 24mm/f3.5 shift, an excellent lens (but not tilt), and the 45TS and the 90TS, which I haven't used.

If would be nice if both Nikon and Canon were to refresh their lens lines with models of 24/28/50/80/120mm shift/tilt lenses using Apochromatic designs and aspheric elements...but they probably consider that a niche market.

The Nikon 85/f2.8 PC-Micro-Nikkor, with its 12.4mm of shift is the sole example of a tilt-shift lens that I would consider outstanding (eg high resolving power, no color fringing). The Olympus 24mm/f3.5 comes darn close, but I would not put it into that elite group, based on shooting it on the D2X and observing significant color fringing. But it's the best there is at 24mm today, which means Canon's offering is pathetic, considering that Olympus designed their lens in the early 1980s.

[diglloyd]

[ QUOTE ]

I did recently use f/Calc to verify numerically Lloyd's stipulation that one see's about 1 1/3rd stop more DoF with the D2X vs. the 1DsMkII. (The result did not surprise me at all, but a friend insisted that this would only hold if one was using equal CoC's for the cameras. So I ran the numbers. It comes out dead on if one uses the respective pixel pitches of the cameras as the CoC in each case. Of course any camera will continue to see an improvement even as the CoC drops below the size of the pixel pitch.)


[/ QUOTE ]

Theory is great (and accurate in an absolute sense). However, when actually shooting on the D2X, there is more to COC than just the resolving power. The composition of the "spot" that is imaged changes as the lens is stopped down, with an increasing "halo" effect. I don't have the mathematical grounding (or inclination) to explain this technically, but it is accurate to say that on the Nikon D2X, anything past f8 inevitably degrades image contrast. f11 resolves just as much, but usually doesn't look as sharp, due to lower contrast. At f16, contrast worsens, and at f22 the game is over.

The only pseudo-exception to this rule is the 28mm/f2.8 PC Super Angulon, which performs better at f11 than at f8 in terms of resolution, but the contrast is still lower. On the whole f11 is better, but the rule still holds that micro-contrast is better at f8 than f11 on the D2X, even as theory would say that the resolution is the same (which it is). MTF charts do show the contrast loss; field results simply confirm it.

[diglloyd]

Correction to the previous post:

The only pseudo-exception to this rule that I've seen so far, is the 28mm/f2.8 PC Super Angulon...

[DougDolde]

To me the aspect ratio is a paramount factor. I am coming from 4x5 and really like that format.

On a Canon full frame sensor the 45mm TSE should be wide enough for stitching because you don't have the 1.5x factor that you do with Nikon. Stitching widens the effective angle of view as well. Plus you get tilt. So I would not even consider the 24mm TSE, but only the 45mm and 90mm TSE lenses.

If Canon comes out with a 12.8mp full frame camera at under $4K I think I would go that way. Of course the 1Ds II would still be an option although a pricey one.