Super Wide Angle Lens Question

[johnmsanderson]

There's a few images I've taken with the 58mm XL where it appears the foreground is crisp sharp whereas detail farther away looks rather soft in comparison. I've done work at F/22 and F/32 where this is the case. Is this simply a trade off of such a wide lens -- coverage vs sharpness?



Sample crops from this image, scanned on Epson V750 (i know, I know...)

Was around 8 seconds at F/32 with center filter

My focus was pretty far out, so infinity was well within DOF.


http://john-sanderson.com/files/58xl/Iona_58xl.jpg

http://john-sanderson.com/files/58xl/tree%20crop%20background.jpg

http://john-sanderson.com/files/58xl/tree%20crop%20foreground.jpg

thanks for any feedback...

johnny

[Ole Tjugen]

Classic case of hyperfocal use, I'd say.

Point is that with the hyperfocal method, foreground and background will be "acceptably sharp". For a given definition of "acceptably sharp". But remember that anything right at the limit, as when using the "true" hyperfocal technique, will be "almost unacceptably unsharp"!

Looks like you need to decide what level of unsharp you find acceptable. Either that, or read Harold Merklinger's "The INS and OUTS of focus", available as a free download somewhere here on the internet. Personally I prefer that method in 99% of the pictures where I have a choise.

[lenser]

Johnny,

Have you checked to be sure of absolute alignment between front and rear standards and their relationship to the subject?

If you are accidentally out of parallel, as in not purposely using controlled Scheimflug adjustments and therefore too far out of intended focus plane adjustments, you could easily exceed the focus limits of both the lens and the subject planes.

I've done this when going from one image where I intentionally was using the Scheimflug principle to accommodate focus planes on the subject, and then go to another scene where I forgot to restore the default adjustments on the standards before making the shot and had my focus messed up all to hell.

By the way, I also own a 58mm XL and find it quite sharp from front to rear as long as I don't mess up as described.

[ic-racer]

If you want the distant trees in focus you have to focus on them, they are obviously not in focus. "well within DOF" was proven wrong by your post. If you are goning to closely examine your images (as you should !) you need a DOF table based on smaller circles of confusion.

In my own work I want more resolution on distant objects and closer objects can have less resolution. This requires focus at infinity. Using a 'hyperfocal' distance will get you just what you have and I agree with you that it is not acceptable.

[Bob Salomon]

What did you focus on? You should be focused 1/3rd into the dof zone you want. And at f32 you are into diffraction.

[johnmsanderson]

Ole and ic -- The Schneider DOF tables claim that if one focuses at more than 15 feet at f/32 then infinity is within DOF. I focused on the base of the tree, figuring everything would come into focus as I stopped down. The lens is very dark on the GG especially at that time of day so I was limited in what I could check was in focus once stopped down.

lenser -- I'm pretty sure everything was vertically parallel, I did not want to use movements because of the tree. I remember checking that the tree was in focus top to bottom. Even if I had a slight amount of front tilt applied it should have thrown the top of the tree out of focus, I'm thinking.

Bob -- I focused on the base of the tree.

Thank you for the replies, any more advice would be appreciated.

I'm looking at the INS and OUTS of Focussing...

"If we want anything at infinity to be critically sharp,
focus at infinity."

Aha!

[Nathan Potter]

John, if I scale the tree branches from your expanded image section, a decently resolved one seems to be roughly 40 um wide (40 l/mm, assuming 4X5 original frame). Let's call the resolution nominally then about 20 lp/mm. The lens should do at least twice that good or better at critical focus so you are probably COC limited at that near infinity distance as others have suggested despite Schneiders exhortations. This unless you are experiencing some kind of distant wind phenomena.

A Schiempflug fix by using the distant trees and some near point only puts critical focus at those two points and all points connecting between in the plane which runs thru those two points. All points outside the plane (well, above and below) will exhibit continuously increasing COC as you move away from the Schiempflug plane. The rate of increase of COC is a function of the NA (Numerical Aperture) of the lens, the f/no. (f/no. = 1/2NA). Some optical expert here might venture to guess what that increasing function is mathematically, but right now my mind has been wined out from a stressful day.

Nate Potter, Austin TX.

[rdenney]

...This unless you are experiencing some kind of distant wind phenomena....

This isn't impossible, though with such a short lens it's hard to know how far away those trees really are. I've had photos ruined because of turbulent air, with shutter speed a lot shorter than 8 seconds.

To Johnny:

It's not diffraction. Diffraction reduces sharpness over the entire image area. The sharpest part of the image is the only place diffraction could be limiting sharpness.

Let's do some arithmetic based on the size of the enlargement. It appears as though you scanned at about 2000, which would make the enlarged sections correspond to about a 48x60" print. That is probably a bigger print than was used as the standard to create those tables.

DOFMaster assumes a circle of confusion of 0.1mm for 4x5. The circle of confusion is the size of a spot created by a point in the subject, where the fuzzy spot is small enough in the print to look like a point and not a fuzzy spot. At a 2x enlargement, 0.1mm becomes 0.2mm, or five lines/mm, which is about the limit of unaided human vision. So, DOFMaster's formulas are calibrated for 8x10 prints. If we select a circle of confusion standard for a 4x5-foot print, we'd need one for a 12x enlargement. That's about 0.015mm. If I use that standard within DOFMaster, the hyperfocal distance for a 58mm lens at f/32 is 23.2 feet, and the near limit is 11.6 feet.

That should have been plenty for this scene. But let's say you didn't focus on the hyperfocal distance of 23.2 feet, and instead focused at 20 feet. Then, the far distance of the depth of field range would be 144 feet--short of those distant trees. With a lens this short, it doesn't take much of a miss in focus to really affect the far end of the depth of field.

If that rock in the lower left is in focus (and we know for sure that it's closer to being in focus than the trees), then your focus point was very likely closer than 23 feet. Lenses this short always make objects look more distant, and I bet that rock wasn't nearly as far as 20 feet away. If it's 10 feet away and at the focus plane, then the far limit of focus might be only 17 feet away.

If you use the hyperfocal distance, you have to choose the correct circle of confusion standard, evaluate the result according to that standard, and then focus at the hyperfocal distance quite accurately.

A few degrees of tilt might have done quite a lot to improve the result, and it could have been done with backward tilt of the rear standard without much effect on perspective projection and without exceeding the lens's marginal 4x5 coverage.

Rick "conservative with depth of field with short lenses and preferring a loupe to a table of numbers" Denney

[johnmsanderson]

Thanks, Rick, Nate. This is starting to make more sense to me. The CoC threw me off. At least the image looks acceptably sharp printed at smaller sizes </assuage>.

The foreground objects were definitely closer than 11.6 ft, maybe 4 ft. It looks farther away in the image.

Lesson learned:

1) Adjust CoC on DOF tables to account for print size, and if this is not practical place focus closer to or at infinity and use a small LED in foreground to check focus on while stopping down.

2) Measure distances from camera position. Optical/laser rangefinder? Or perhaps my guesstimates will be more successful as I account for a smaller CoC.

Perhaps there is a way to measure distance with my android phone. Anyone done this?

3) While being cautious about vertical objects, apply a small amount of tilt.

These wide angles are something to master.