Help judging performance of my FE55z, FE35f2.8z horizon/infinity

Started Jul 5, 2016 | Questions thread
JimKasson
JimKasson Forum Pro • Posts: 26,613
Re: Full sized images
1

l_d_allan wrote:

JimKasson wrote:

l_d_allan wrote:
Since you don't care whether the sky blows out or not, seems like you could use a longer exposure ... the EXIF shows EC of -1.3 EV. Do recall if you used "A" mode with EC, or "M" exposure with EC ignored?

A mode with EC set to -1.3. Noise is not a problem in this scene, and I didn't want to blow the highlights on the leaves, which don't show up in blinkies or the histogram.

Good point.

  • I don' t have that lens, so have no basis of comparison on what the center, near edges at the APS-C off-axis, and extreme corners should look like.
  • I also don't enough experience with TNIH to comment on perceived IQ.
  • We apparently differ on whether validly done TNIH captures tends to be better, worse, or about the same as a validly done brick wall IQ test for GCVBCT.

Ideally, we could test with a very large brick wall 500m away.:-)

Roger C at LR has wistfully written about renting a billboard in order to get far enough away ... and not just for super-telephotos as I recall.

  • If so, have you checked out how far trees at the "left horizon" are vs the "middle horizon" vs "right horizon"? My experience is there is a definite "learning curve" to estimating distances, but with practice you can get pretty good at it ... such as golfers and hunters.

Everything is well over the critical 200m.

Is that from your "Object DOF" studies? With the earlier link regarding the Otus 55mm at various apertures focused at 10,000 meters (10 km).
http://www.dpreview.com/forums/post/58031944

Yes.

If so, the "200" seems associated with the MTF-50 label with units of cy/ph.

You're thinking about the chart wrong. To get to 200, take the f.2,8 (yellow( curve, and track it back to where it starts bending down at 500m. Then convert 500m for a 55mm lens to 200m for a 35mm lens by multiplying by the square of the focal length ratio.

That puts the relevant MTF50 over 1600 cy/pn.

Does that indicate relatively poor (if not grossly unacceptable) 200 cy/ph IQ? Wrong link? Low or no comprehension on my part?

See above.

My experiences seem different, but the FL and aperture are very different:
With a fast zoom at 4x longer 200mm but two stops slower f/2.8, focus still mattered more than I expected. My rusty math is that for DOF purposes, I can't use the the chart for the Otus 55mm f/1.4 and the 200mm f/2.8.

with a 200mm lens, you'd multiply the 500m by (200/55)^2 = 3.6 and get almost 2 km. That would be incredibly conservative for a zoom, but maybe not for the Nikon or Canon 200/2 lenses.

Even ignoring that the Otus is sharper by at least 100 cy/ph at f/2.8 that the zoom, aperture is non-linear square, but FL may be linear?

As you have seen, hyperfocal distance varies as the square of the focal length.

Or is it more complicated than that, and I need to carefully read 100's of pages of TLW blog articles ... after retaking some math classes and taking one or several courses in Optics?

Sounds fun to me...

Back to my fast zoom at 200mm f/2.8:
There were neighborhoods perhaps several miles away. Focus was still critical. With magnified LiveView, focus on the roof tiles of the Red house was obviously invalid for the Blue house just to the front of it or rear of it.

I've notices with my Nikon 500/4 that I can focus on something half a mile away and have hills 5 miles away be obviously OOF.

Focus at near infinity to Cheyenne Mountain signal towers at perhaps 20 miles away was obviously unacceptable for the Red house maybe 2 miles away. Also unacceptable for signal towers maybe 5 miles away?

I've never tried it, but my speculation is that at dusk and if I focused on a partial moon, that might not to appropriate for the Cheyenne Mountain towers.

A doubt that; see the above 200mm lens calculation.

Going back to the 200m issue with your Otus 55mm ... I will reluctantly concede that if I made a large'ish print with the focus distance recommended by the Object DOF approach, that DOF / CoC might be adequate ... if I used the full frame.

To me, that ignores the factor of cropping, which I don't think can and should be ignored.

Cropping linearly affects cy/ph.

If the lens + sensor is critically focused and is capable of A3+ 13x19" prints from 25% of the full frame, that's how I want to focus. And that includes for GCVBCT.

I think the argument that "as long as you can make large prints at 300 dpi, it doesn't matter" is maybe obsolete with Super-Exmor's with 42 mpx that encourage and allow significant cropping. I'll admit I need to think that through more carefully.

I'll look for the .DNG's. If necessary, I'll add to my "to do list" to re-take those images on another super clear day, and be that much more careful.

Is that consistent to the "extinction graphs" for a 200mm lens at f/2.8? It might be just an average 'L' prime, but it's a zoom with a reputation for stellar IQ, world class build quality / control.

I will attempt to read and comprehend what I think of as your Object DOF related extinction graphs. Maybe a light will go on?

Is there a graph on your web-site that would have 'extinction lines" that reflect a high quality 200mm lens?

Take the 55mm graphs and multiply the distances by 3.6.

Would I be better off testing my admittedly less capable FE55, and attempting to extrapolate to your Otus 55?

Those lenses aren't very far apart at all  in the middle of the frame, at f/4 and narrower.

If I went back thru the previously ignored 20+ articles of the Object Mode + Protocol, would I find a "extinction graph" for the FE55?

No. I never modeled it. I haven't collected sufficient data.

I realize those graphs aren't answering the questions I'm asking. I'm being slow how they relate to GCVBCT, other than maybe how critical alignment is or isn't.

BTW, my impression is that you ended up your Object Model DOF in general or specific agreement with the paper that had the concluding summary like, "focus at infinity with the aperture being FL / 5". Or did you closely study that, more or less reject it,

Yes.

and came up with a different, but perhaps equally unconventional approach?

No.

I'll address my confusion on the issue of just how accurately it seems you'd have to have the horizon trees at the center, APS-C rule-of-thirds, and extreme corners in terms of distance in a separate post later.

And really, I want to be once again in error ... that TNIHT (tilted near-infinity horizon testing) is a relative "piece of cake" compared to brick-wall testing and still provides valid captures for whether to return a lens.

I learn less on the rare occasions when I'm correct than on the more typical posts when you point out my ignorance / misconceptions / carelessness.

I hope the above helps.

Jim

-- hide signature --
 JimKasson's gear list:JimKasson's gear list
Nikon D5 Sony a7 III Nikon Z7 Fujifilm GFX 100 Sony a7R IV +4 more
Post (hide subjects) Posted by
(unknown member)
(unknown member)
(unknown member)
(unknown member)
(unknown member)
osv
osv
osv
osv
osv
osv
osv
osv
osv
(unknown member)
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow