Discussion in 'Nikon' started by evan_bedford|2, Nov 2, 2017.
I'm glad to see that the electronic focus indicator also works with the ancient Nikkor PC.
Finding, however, that the Sony A7s does much better (relative to the Df) with the PC Nikkor lens in the corners and edges of the shots. Wonder if it has anything to do with the fact that I have 2 adapters in between the lens and the A7s body. Therefore the A7s sensor uses the central part of the PC lens. Would the central part of the lens have less distortion and more sharpness than the outer part of the lens?
Why 2? What are they?
I don't know what the two adapters are and hence can't figure out the "therefore". But those adapters likely only move the lens to the correct flange-to-sensor distance on the A7S body and you should have the same FOV on both the Df and the A7S. The lower resolution A7S sensor may not show issues as clearly as the slightly higher resolution Df one.
What's not so good about the corners and edges on the Df?
As Dieter explained, for any given amount of shift, the Df and A7s should be using an identical area of the image circle. - Unless one of those adapters allows tilt as well.
However, I'd be surprised if 2 stacked adapters didn't cause some vignetting with the lens shifted near to its maximum extent.
Hmm, you're right. Now that I check, both cameras show precisely the same field of view. (I use a Nikon to M-mount, and an M-mount to Sony adapter on the Sony). And now that I check more closely, the corners and edges seem no different than sharpness in the center. However, I still can't seem to nail focus on the Df -- even at f11. I've tried both the live view with magnification, and the little dot between the two triangles, but something seems crappy on the Df output. I used the focus peaking combined to digital zooming to nail focus on the Sony. Both cameras are on tripod with timed shutter release. Attached photo shows the Df above and the A7s below. The A7s image is slightly smaller, since it has fewer pixels.
Did you focus stopped down or wide open on the Df?
Thanks ilkka. That may be the key. I was probably focusing stopped down before. Just did it again, wide open, and the focus seems to be better. The A7s output sometimes still looks a bit sharper, but it's hard to compare 12 mp to 16 mp output, and I may just be splitting hairs unnecessarily.
Were the above RAW images similarly processed? If not, the difference in sharpness could easily be put down to the degree of in-camera sharpening applied to JPEG output.
Actually, I'm pretty sure they were both jpegs to begin with. But I'm getting better with Df focusing now. Image below shows the Df above and the A7s below. Both shown at 200%. Both from the outer fringe of the scene.
Glad it's getting better, Evan. The Sony output has some ringing around sharp edges, which says to me that JPEG sharpening settings are in play (along with the colour shift). I hope you're now happy with your set-up! I imagine newer PC Nikkor lenses are designed to allow for the sensor stack thickness, which may be different from the Sony (so you may see a benefit to Nikon with a newer lens) - but somehow I doubt this was a design goal for this particular lens!
"I imagine newer PC Nikkor lenses are designed to allow for the sensor stack thickness...."
- Nah! That would require a sensor in the lens to communicate the amount and direction of shift. Nikon ain't that smart.
The 24mm PC-E Nikkor still shows horrible fringing when shifted, same as the old 28mm f/3.5 PC lens. Although some bodies partially correct it as if its lateral CA. On uncorrected RAW files it's blatantly obvious.
If it is really the ”same” the only way you could know that is if you have made a side by side comparison. Could you share it with us? Thanks. There are not so many tests of PC lenses published.
When the 24 PC came out we did some testing with also the 28 PC and 35 PC (unfortunately I don’t have the images easily accessible as that was many years ago and I didn’t feel important to keep them) and and at least at that time the 24 PC was felt subjectively the best quality. The 24 produced images with richer colors and better sharpness than the 35 which looks comparatively dull. The 24 PC unshifted also was the sharpest 24 we had access to at the time (including 24-70G, 24 AF-D and 25 ZF). The 35 and 24 PC both seemed to exhibit fewer issues when shifted than the 28. I guess with CA the question is if it is such that software corrects it well or not.
I still have the 35 so it should be possible to do a side by side with the 24, but I never owned the 28.
I don't doubt that the 24mm PC-E is optically superior to the 'old' (but not long out of production) 28mm f/3.5 PC when unshifted. That was never what I meant or claimed.
What I'm saying is that there's still uni-directional fringing visible on images from the 24mm when shift is applied. And that because there's no way the camera can 'know' the direction or degree of shift, then there's also no quick software fix available.
This 'shift fringing' is entirely different from normal CA and Loca. As Andrew implies, it probably has more to do with the angle of rays passing through the sensor's IR filter, and the angle at which the microlenses receive light. Although the 35mm f/2.8 PC-Nikkor doesn't show it as far as I can see.
It would be interesting to compare the degree of fringing between sensors that incorporate an AA filter, and those that don't.
Ilkka, here's a sample I posted ages ago showing the shift-fringing issue with the 28mm PC-Nikkor.
Camera was the D700, which AFAIK doesn't do any inbuilt CA correction.
Left - unshifted, right - full shift.
Edit: I don't have the 24mm PC-E. I was put off buying it by this very fringing issue. I felt that for its cost it should be fully usable with no fringes visible.
For what it's worth, I only meant allowing for the optical thickness of the sensor stack as part of the optical design of the lens (see Roger Cicala's article) - obviously something that's different for film. Although Nikon don't seem to have stayed particularly consistent with their sensor stack size. I've no idea what kind of adjustments could be made, especially for tilt/shift, it was just a thought.
Software is pretty good at autocorrecting these days; shift you can obviously fix by treating the image as a crop out of a larger image circle. Tilt is harder.
"shift you can obviously fix by treating the image as a crop out of a larger image circle."
- I'm not so sure about that Andrew.
'Shift fringing' is not like lateral CA. It doesn't radiate from some central point, but smears high-contrast edges perpendicular to the direction of shift.
I haven't characterised it in any detail, except to say it seems focus independent and variable along the shift axis, but consistent across the frame normal to the shift axis.
Incidentally, I don't remember seeing it when I played with a Samyang 24mm T/S, but I had limited hands-on time with that lens under less than ideal testing conditions.
Hmm. The tilt-shifts I've used have shifted the entire optical group (there's no lens on the mount side of the shift); I don't know whether that's true for all designs on 35mm. Unless the mount is causing some vignetting, the effect of shifting should be the same as that of using a wider lens, and cropping (although the rear nodal point of the lens must come into play). It's possible that Nikon offset their sensor microlenses to compensate for the exit aperture location (I know Leica have, historically, but then they have some lenses with rear nodal points very close to the sensor), but for issues specific to the lens, I'd expect it to be possible to create a larger image in an editor of which the capture is a subset, then correct lateral optical aberrations relative to the lens centre. Or at least, I'd expect it to be a lot better than trying to correct relative to an uncentred lens.
I'll try to do some experiments with my tilt-shifts some time soon.
A little quick and dirty testing with a D7100 and the old 35/2.8 PC and the much more modern 85/2.8 D T/S micro suggest that the 35 shows noticeable color fringing when shifted vertically to the max on DX, and also a fair amount horizontally, though maybe a little less. It does not show up everywhere, but it's noticeable if you pixel peep at something like a newspaper. Shifting a bit less radically improves it a bit, but it's still there. The 85 is much better behaved, and in fact seems to show less color fringing shifted than it does straight on. I did both these at F5.6, indoors with flash, at about five feet, just shooting newspaper taped to the wall.
I've never found the color shift on the 35 particularly bothersome in actual use, using it mostly straight and shifting sideways for panoramas and shadow control. I rarely crank it all the way up or down.
I'm pretty sure the fringing is caused by an interaction between the sensor and shifted lens. It's definitely not CA that's native to the lens, because there's no sign of it in the corners/edges of an unshifted image.
As I said, the shift-fringing doesn't appear to radiate from a centre, at least not one anywhere within the image circle.
I don’t normally use full shift on a PC wide angle because the effect would be too extreme to my taste. I discussed this with an architectural photographer who looked at my images and advised to avoid full shift to avoid an effect of the building falling over. I realize a full shift would produce artifacts especially if done along the long axis of the frame such as in your example. I think it is good to stay within +-7 or 8mm. There should be fewer problems when shifting in the short dimension of the frame. I shoot these lenses at f/8 to f/11 normally. On the 45 and 85 I sometimes use wide open as well, but on the 24mm stopping down is mandatory if shifting a lot, to maximize the image field.
Once weather is favourable I will look at the 35 and 24 and see what kind of a CA effect there is at shifts and apertures I would use. I will try to get a hold of the 19mm also, a bit later. It is supposed to alleviate the field curvature effect of the 24.
Separate names with a comma.