hubsand wrote:
If I still had my Contax N 17-35mm, I'd follow DXO's example and create a set of sharpening actions that target the mid-frame resolution dip. In the 17-21mm range, the results would be outstanding.
Please give an idea as to how to identify 'Area B' on the images to apply sharpening to this specific area only. I am very interested in these PS actions, since I own converted CZ N 17-35. PM me if this is off topic. Thanks. Pavel
belsha wrote:
Now one could indeed critize the rating system. In the 17-35 Contax vs Nikon test, it is true that the better shot at one focal range/aperture is always rewarded with a 1 point advantage, regardless if it's superiority is enormous (for exemple Contax Nikon 17mm wide open edge performance vs Contax ), quite substantial (generally speaking Contax mid frame (region "B", I think) vs Nikon) or barely visible or totally invisible to me (for exemple 35mm f5.6 center performance, but there are many more). At times, the ratings seem totally counterintuitive to me: at 24mm f5.6, the Canon L clearly looks A LOT sharper to me than the Contax, but gets a 0 vs 1 for Contax......Show more →
Thanks for the proof-read! The rating given at 24mm / f5.6 is a typo, and has now been corrected in line with your observation which seems spot on to me. The danger with awarding more than one point for each subcategory is that it all becomes a bit subjective: does this performance deserve 3 or 4 out of 5? It's much less controversial and open to general agreement to say that lens A is better than lens B in this particular shot, though it doesn't allow for any fine grading.
I'm still quite interested in the idea of a Parker-style system: in which a lens has the opportunity to earn 100 points in various weighted categories. That's even more subjective, however, and conflicts with a basic premise of the site: avoiding reduction to a single number, when the truth is rendered in shades of grey. As you say, Mike Hatam's Olympus v Zeiss 18mm is a classic case: I hope it's put an end to 'my Zuiko is better than your Zeiss' type debates . . . that would be better at what, exactly?
PLOG1964 wrote:
Please give an idea as to how to identify 'Area B' on the images to apply sharpening to this specific area only. I am very interested in these PS actions, since I own converted CZ N 17-35. PM me if this is off topic. Thanks. Pavel
If N 17-35 users would find it useful, I could write a set of actions that would automate the sharpening routine and give some extra attention to that annulus horribilis. If someone PayPal'd me a Fred Miranda-style $10 contribution, I'd sort it.
However, in honesty, I'd rather just tell you how to do it . . . just make a three point graduation in the mixer, apply it from the dead centre of the frame to the far corner in Quick Mask mode. That will create a fuzzy torus in the frame centre. You'll need to experiment a bit to match it to the lower resolution part of the field. Exit Quick Mask and apply a USM of something like 150/0.5/0.3, followed by another wide radius USM if you like it that way. The contrast could also be perked a little here with a curves adjustment. Then invert the selection and USM 100/0.4/0.3 or some such, plus a softer wide radius USM. Sharpening at f2.8 would require a slightly different action to target the corners.
> ...The danger with awarding more than one point for each subcategory is that it all becomes a bit subjective: does this performance deserve 3 or 4 out of 5?
By coincidence, the superiority of a mark out of 100 over a simple win/lose makes a good analogy to why a sensor's analogue (ok - 12 bit) gain gives a more accurate picture than the more volatile response of a film grain. A mark out of 100 doesn't have to be perfectly accurate for the running total to tell you a lot more about a lens' overall performance. If you managed to keep errors (noise) in the 10-20% range then I think this method would provide a much more informative scoring system. You could even make a rating calculator so that the site visitor could personalise the outcome by skewing it towards the factors that matter more to her. She could click buttons saying 'corner resolution: not important', 'distortion: very important', 'contrast: crucial' et cetera. I could build it for you in javascript if you like.
tom in mpls wrote:
Can you post some pics from the diStoRtaGoN?
Here's one. This lens scores higher for distortion than any other lens I have used. It's an incredible design. Two of the elements are made from jelly and they have to be kept moist so there is a small spittoon on the top of the lens body. The manual says you can chew tobacco to get that really creative graduated brown sky effect popular with landscape pros. I tried but felt sick and had to lie down.
I'm glad you didn't succumb to the temptation to use the tobacco grad, brainiac: the shot would have looked so contrived. For those interested, here are Seuss' published distortion data for the lens: http://www.16-9.net/erase4.gif
It's a bit of barrel, bit of pincushion, bit of waveform, bit of moustache, hint of gullwing, touch of frost, soupçon of fubar.
MarkJones wrote:
I wonder have you had chance to confirm your results with the CN17-35 with other samples of the same lens?
In using the CN17-35 I have noticed that it can produce nice 3d like images. Does the Nikon 17-35 do that?
The N 17-35 is very effective in this regard: it has typical Zeiss plasticity. The Nikon isn't too shabby, either: the better Nikon lenses have a signature that is closer to Zeiss than Leica, but often without such attractive bokeh.
