NikonGear'23
Gear Talk => Lens Talk => Topic started by: Airy on April 20, 2017, 20:11:21
-
... very much at daytime, for now. I got mixed feelings : on a high MP camera, it does not look good, compared with e.g. the 50/1.2 or the Voigt 58/1.4. On a low MP camera (Df), I won't say it shines, but the results are more attractive. Contrast wide open is relatively high, and it does not flare easily.
The first two at f/1.2, last one at f/2.8 ; #2 is slightly cropped.
-
A dreamlens, Airy, congratulations ! But it seems it is best for poor light conditions, where it really shines !
What about comparison with your 1,4/58 Voigt and your Nikkor 1,2/50 ?
Amitiés, Didier
-
The first two shots are amazingly good considering they are at f/1.2.
-
Honestly, direct comparison (using D800, LV with Zacuto) was a disaster.
1) brickwall shooting, 1.5m distance : Noct displays significant field curvature and focus shift. Worse, focussing (via LV) close to the distant edge will reveal that it never gets sharp there. By comparison, the Voigtländer has no significant focus shift (the focus point remains in the sharpness zone), has a flat field, and does no mush near the far edges. I never bother about the corners, by the way, where it would be vain to expect decent sharpness wide open ; these are no OTUSes.
2) close focus shooting (about 0.5m) : focus shift gets even worse (to the point that the focus point will be outside, and in front, of the sharp area). All lenses display lots of LoCA, the Voigt being most "baroque" (LoCA seems to be from Rubens' palette, but the pics have better contrast). But I do not really care, because I see no relevant "use case" for a Noct at such a distance, opposite to 1.5m (portraits).
On the other hand, doing real shots using the Df, the Noct does not disappoint. More examples to follow. I have the impression it is a good portrait lens, if you overcome the focus shift, which is possible by combining training (focus in front of the desired sharpness plane) and "statistical shooting".
So far, I have not made up my mind, whether I should keep it or not. So I keep shooting.
-
The first two shots are amazingly good considering they are at f/1.2.
I confirm, f/1.2
But since the pics are not really tests, but honest attempts to get decent images, I pushed the sharpening and clarity a little bit. These won't turn mush into crispness. So when you see sharpness, it means the lens is sharp indeed. Downsampling helps too (pixel size = 1600, long edge)
The figures (in LR6) : the bar & wine store, sharpness at 35. The two fatties : sharpness 45, clarity +7.
By the way, it took me 7 attempts to get the one with the fatties sharp (the scene was not strongly lit, while the ambience was sunny). DOF is ridiculously low, and LoCA will immediately deteriorate the blacks.
-
This one at f/1.4. For some reason, the door, which I thought I had brought into focus, does not look very sharp (setting = +30 in LR6 ; shadows raised) ; focus shift is minimal between f/1.2 and f/1.4, so blame my poor eyesight. I was more interested in getting a smooth light background.
Bokeh is so-so, because of strong outlining of shiny points ; the outline is green due to LoCA behind the focus plane.
-
The Noct on D800 is indeed a disaster in terms of getting the proper focus. Same story on an A7 Sony, by the way, so we cannot just put the blame on the OVF it seems. LiveView on D800 is also a hit-and-miss affair.
On my Df, the Noct is easy to focus and I have no problems getting the focus where it should be.
I did careful comparison with Noct and the 'Neo-Noct' on a D800, by selecting a test subject that was pretty flat and inclined. As the 'Neo' had AF, it could be focused more precisely in this case, but apart from that, the old lens did well against its modern counterpart and in fact was sharper at f/1.2 than the other at f/1.4 for the on-axis part of the subject. Stronger field curvature than the 'Neo' was obvious though. I ended up keeping my Nocts and declined the offer of the AFS 58/1.4 at a good price.
-
With the Zacuto, a tripod and lots of patience, focussing on the D800 remains possible - the other problem being the short throw of my AIS version... But that does not explain it all (I mean, the poor results). Shooting at an angle did not provide evidence that my Noct was better than the other two. Once again, in the field and with the Df, it is another story. I have yet to understand why.
-
Attempted subject isolation at f/1.2. Framing is not quite OK : the "V" of the two guys in front is part of the intention ; I should have paid more attention to the arcades in the background and to the column that interferes with the player's profile. But I could not stay much longer. Also, the sharpness of the player could be better, but you got the idea.
-
Another one at f/1.2. Note the slightly "swirly" bokeh, caused by the catseye effect, and enhanced by the decent vignetting: we have no "clipped corners" effect, unlike 50/1.2. Sharpness +45, but contrast and clarity settings left at their standard level.
This kind of pics makes me feel special about the Noct, even at daytime.
-
Still f/1.2. Absolutely no tricks - standard contrast and sharpening, no clarity push. No vignetting change (there is no profile for that lens in LR, by the way)
-
I did careful comparison with Noct and the 'Neo-Noct' on a D800, by selecting a test subject that was pretty flat and inclined. As the 'Neo' had AF, it could be focused more precisely in this case, but apart from that, the old lens did well against its modern counterpart and in fact was sharper at f/1.2 than the other at f/1.4 for the on-axis part of the subject. Stronger field curvature than the 'Neo' was obvious though. I ended up keeping my Nocts and declined the offer of the AFS 58/1.4 at a good price.
I tested the "Neo" and found the tested copy rather impressive ; impression shared by an employee (and Noct owner) at my favourite shop in Brussels, who also made a comparative report for a Belgian photo magazine. Whatever the difference may be, my impression is that for night shots, lenses such as the Tamron 45/1.8 made the (Neo)noct look older, if not obsolete. With good coma correction, good coatings, and stabilization, such kind of lenses pretty much fill the bill. This is also why I keep investigating the daytime behaviour of the Noct: the Tamron is very neutral (too neutral maybe) and opens at f/1.8, whereas the Noct has an interesting personality.
-
The DoF at 50mm f/1.2 is so shallow I found I can't stand free and expect to hold the focus. Slight body movement easily moves the focus point of the lens before or after the chosen focus spot on the subject. At the least one must sit down or brace against a solid object to keep the camera position from moving during and after focus. The DoF of a 50mm (51.6mm) lens at f/1.2 and a focus distance of 1.5 meters is 0.06 meters, 60mm or 2.36 inches. A 58mm at f/1.2 drops to 0.05 meters, 50mm or a tiny bit less that 2 inches. I cannot stand steady enough to hold the focus.
I found on my D300s I could focus better than I could stand still. Trying to hand hold a 50/1.2 at f/1.2 and 1.5 meters is photographer error for me. Others may have less problems standing perfectly still. I'm guess I'm just wobbly. If I lean on a counter or sit in a chair the number of in focus frames markedly improves. The first time I tried my 50/1.2 AIS on a friend's D300 I missed the focus four times. Then I leaned on a kitchen counter and nailed the focus. I wish that meant I could nail the focus 100% of the time by just bracing myself.
I don't find the focus screen on the D800 or D300s to be nearly as good as the one in my D2H. Also with the D2H I could use the DK-17m (1.2x) eyepiece. I own a 1.2x eye piece for the D300s and the DK-17m fits the D800 but I can't see the data in the finder with these.
Dave Hartman
I used DOF Master and a circle of confusion of 0.03mm. I believe the calculator is for an 8x10 inch print. 0.03mm is probably out of date so generous. Maybe I should have used 0.025mm? The DoF figures are certainly not for 100% viewing of a D800 file.
-
David, you are absolutely right, and this is why I keep shooting without any tripod - but taking enough shots so I can select a sharp one.
Against camera shake, I typically shoot twice (at 1/FL speed) or three times (at 1/(2*FL)) just to be on the safe side.
Against focus errors when shooting those things wide open, I repeat until the image I get and zoom on the rear screen is sharp enough. So yes, 7 shots happen to be necessary at times. But it is still less cumbersome a procedure than lugging a tripod.
With models, one has to select patient ones. For family shots, better grab the 50/1.8 G anyway.
-
Against camera shake, I typically shoot twice (at 1/FL speed) or three times (at 1/(2*FL)) just to be on the safe side.
Funny how skyrocketing ISOs have been balanced by skyrocketing resolution demanding higher shutter speeds. ISO 12.8k is nice, but gone are my days of shooting a 50mm at 1/50!
Lens looks to perform quite nicely.
-
What is the size of a single pixel for a D800 censor? After the Bayer filter does its thing what is the "size" of a single pixel in the image file? Is it the size of four pixels, a red, two green and a blue?
What circle of confusion should be used for 100% pixel peeping? :D
El Diablo David
-
Vast subject, worth another thread. In the present case, I only can say that the Noct dislikes high MP sensors. My subjective impressions (when it is properly focussed) is that Noct IQ wide open ranges from "remarkable" to "junk". I still do not understand how the 50/1.2 plays it even with the Noct when the D800 is behind, whereas on Df, I would immediately know my preference.
... to be taken with a pinch of salt, since I have only a few hundred shots done and none at night. I hope to be able to propose "comparative" shots, which is tricky because I'd have to change my vantage point (if some nifty fifty were used as a yardstick), to keep the framing identical (an thereby altering the perspective and possibly the lighting).
As for handling, the Noct is a pig. Focus shift, shallow DOF and short throw = no fun. Fortunately, weight and form factor are ideal on Df. If you want to stay comfortable, go for a Voigt 58/1.4, my sweetest lens so far.
-
Although I can understand the appeal of statistical shooting, I would like to humbly relate that I've been following an approach that is quite different.
I don't know whether it will work for everyone, but I believe that manual focus can be trained similar to throwing darts or basketballs, or playing an instrument.
You have to force yourself to get it right on the first try.
I did and still do the following practice method.
Pick a very fast shutter speed to make it just about the focus (like 1/1000s).
I stand with my camera in front of my belly, arms relaxed and my left hand on the focus ring. I put the lens on infinity between shots to have a consistent starting point.
I pick a target (for starting out, targets on an inclined plane like a bookshelf at an angle or an isolated target like a sharp pencil in a mug are best), I look at it and try to concentrate on its appearance. I then raise my camera to the eye while inhaling, and while exhaling I turn the focus ring in one continuous motion (no overshoot because that is too difficult for the brain, at least mine), stop and press the shutter. Then I lower my camera again, inhaling and exhaling, before doing the next turn. The whole cycle lasts 2-3 seconds only. I do five of the same target, then calmly evaluate the results. Afterwards, I go to the next target, picking a nice variety of focus distances between MFD and a few meters (for close to infinity focus I do rely on the green dot, otherwise I try to ignore it and concentrate on the visual appearance of the target). I will do no more than 5-10 targets a day, and then do something easier for the rest of the day (like shooting at smaller apertures and work on composition and exposure).
The method has the advantage of not straining the eyes so much because they can relax between the shots, while standing there turning the focus ring back and forth around is very tiresome for the eyes, and small differences are harder to perceive than a continuous focus action. You benefit from observing the target with your naked eye and having a short-term memory of it when you see it again through the viewfinder. Your brain learns how much to turn the ring based on the distance (if you prefer, you can also start from MFD all the time). You maintain a healthy breathing cycle and a relaxed stance, both of which make the whole thing more enjoyable, while also giving you a stable platform for slower shutter speeds. You also train yourself to be quick, which is handy for street and portrait shooting. Because you are not overshooting while turning the ring, you will learn to have a sense of the degree to which you missed focus even without looking at the image (like when throwing a ball, you usually know that it will miss as soon it leaves your hand, so over time you will be confident that it hits once it leaves your hand).
It's not that trying to nail focus by going back and forth is bad in itself or cheating, but I find that I'm not able to train myself to get better that way (essentially, it's an open control loop and you never have to make a commitment). I will reserve this for the really hard targets, but while practising I try to make one continuous turn and then stop.
If you try this, I would be interested to hear about your experiences.
-
As for handling, the Noct is a pig. Focus shift, shallow DOF and short throw = no fun.
Lenses should be fun on all camera's, not?
-
Lens handling fun obviously varies. What matters most is the result,, but.
-
Yes I understand. Thanks for your detailed experiences with the lens. That's why I asked. ;) I agree that results matter. It all depends also on your personal goals and shooting style.
-
Funny how skyrocketing ISOs have been balanced by skyrocketing resolution demanding higher shutter speeds. ISO 12.8k is nice, but gone are my days of shooting a 50mm at 1/50.
Down sample and you're covered. :)
Last night I was trying to find out the DoF for a 58mm lens at f1.2 with a subject distance of 1.5 meters for a D800 when view at 100% pixels. If my logic was correct (which is not assured) the DoF would be a touch less than 8mm total (about 4mm fore and aft).
Dave Hartman
-
Down sample and you're covered. :)
Unfortunately, not quite. Camera shake will result in a loss of contrast. Downsampling will make the blur less apparent for sure, but the contrast will not be restored. This is what I observed with the D800 on my first day of usage - blacks would appear "less black".
-
Wide open ALL my lenses have less contrast. Especially pronounced on f/1.2 and f/1.4
-
Simone, thanks for reminding me to practice quick focusing. Many of my lenses have too-short and non-matching focus throws, so I tend to forget.
Practice and familiarization with the lenses also helps with predicting and controlling the actual, deliverable depth of field much more that charts and tables.
I often, but not always, practice what Airy has coined as 'statistical shooting'. Not only against my camera shake, but against subject movement. Combined with instant focus check via playback, it is one of the great economies of digital.
With a non-moving photographer and subject, the imprecision of the modern focusing screens seem to demand repeated racking back and forth before settling on the placement of the plane of focus.
-
My 55 f1.2 is not a noct but it sure is complicated to focus on the D800
I can only focus with liveview when done with a tripod, and it looks better if the liveview is done in video mode, than switch to photo before taking the shot
I second the comment about the patient model and a photographer retaking the use of a MF lens
I'll practice Simone method, about the speed: the weight of the 55 seems to help a lot , and I get less motion blur than with my other lenses
-
Today's finding, using test charts : indeed, on Df (16 MP), the Noct is better than the Voigt 58/1.4. On D800 (36 MP), the reverse was true.
In other words, the Noct is the better lens on relatively low res sensors, not on high res.
I still struggle to understand why this is the case, but it corroborates by practical shooting experience: on Df, the Noct does not disappoint. It is, so to say, a high contrast, low res lens, but the latter weakness is simply not apparent when the pixel count is 16 million (on FX).
The test chart shooting confirmed the better contrast, similar (apparent) resolution, lower coma of course, and lower (less "technicolor") LoCA. The focus shift is there and, if uncorrected, the focus area chosen at f/1.2 - f/1.4 will drift backwards, to the extent that the chosen focus point will slightly be out of focus, with slight purple fringing revealing the back-focussing. But the phenomenon, while annoying, does less damage that on D800.
I think I'll keep this lens - I really like the way ot draws at f/1.2-f/2, better that the 50/1.2 I think (and wide open, there is much less overall haze, increasing the subject isolation). But I'll never mount it on the D800, and can only hope that Df successors won't pack many more pixels.
Best use case: portraits, definitely, where the extra 8mm of focal length are quite welcome. As said before, for night shots, the Tamron 45/1.8 makes more sense. Now I am going to search for the patient model.
-
I've heard somewhere that Noct was designed to reduce the coma while somewhat sacrificing the resolution. The resolving power of Ai(s) 50/1.2 is higher.
So, the resolution of the 16MP sensor may not high enough to reveal the weak point of Noct, as opposed to the 36MP one.
-
I've heard somewhere that Noct was designed to reduce the coma while somewhat sacrificing the resolution. The resolving power of Ai(s) 50/1.2 is higher.
Coma near the edges of the frame is a fault of spherical 50-55mm f/1.2 and f/1.4 lenses wide open and generally cured by stopping down. The aspheric front element of the Noct is there to control the coma.
Various correction of aberrations can confict so too much correction of coma in a spherical design will increase another aberration to an unexceptionable degree. The Noct was designed for shooting at night at large apertures, f/1.2 to maybe f/2.0 where a spherical design displays significant coma. If the Noct isn't as sharp as the 50/1.2 AI/AIS at f/5.6 it doesn't matter. Its primary design consideration isn't for shooting at f/5.6.
As a non-owner with no training in optical design I'm probably typing trash but that's my humble understanding.
Dave Hartman
-
I've heard somewhere that Noct was designed to reduce the coma while somewhat sacrificing the resolution. The resolving power of Ai(s) 50/1.2 is higher.
So, the resolution of the 16MP sensor may not high enough to reveal the weak point of Noct, as opposed to the 36MP one.
The trade-off "resolution vs. coma reduction" is obvious, away from the center. Far sides never get sharp when the Noct is wide open, regardless of focus compensation.This becomes apparent when shooting a +/- flat subject at an angle, here for instance:
-
The Noct never was conceived as a "flat-field" design. Despite its current popularity as a lens for portrait and dreamlike close-ups, it shouldn't be compared to lenses designed for those purposes.
-
... however, what I was referring to is *center sharpness*. I thought naively that lens ranking would remain the same, no matter the sensor resolution; on low resolution sensors, the gap might decrease, and on very low resolution sensors, the performance might get identical (when sensor resolution gets lower than both lenses resolving power).
In reality, I observed a reversal, the best becoming the worst, and this is surprizing: Voigt best and Noct worst on D800 by a wide margin, the contrary on Df by a narrower margin. I'll redo the test this afternoon and share the results. No direct comparison with the 50/1.2 because of FL difference, requiring even more precautions. Common wisdom is, the 50/1.2 overtakes all other Nikkors (50-58) at f/2.0.
-
The Noct never was conceived as a "flat-field" design. Despite its current popularity as a lens for portrait and dreamlike close-ups, it shouldn't be compared to lenses designed for those purposes.
Of course not, and that's not an issue I'd want to raise any further. See crops (center and right side) of previous picture; on the right, the focus zone is slightly moved forward, and good sharpness, while still there on the left part of the right side crop gets completely lost on the right side (the outermost, and darkest, by the way) :
For flat field, go for Voigtländer 58/1.4 for instance, or most 50-55 macro lenses.
-
Common wisdom is, the 50/1.2 overtakes all other Nikkors (50-58) at f/2.0.
I believe that originates with Ken Rockwell. I don't believe it includes 58mm lenses, only 50mm lenses, and I don't believe this includes recent G-type 50mm Nikkors.
Dave Hartman
-
It does not originate from Mr. Rockwell, but from another tester who compared several fast fifties including a Noctilux. The MTF charts dealt only with the center.
You are right in saying that this older test, which I published somewhere on Nikongear, did not include the "Gs"; nor did it include the Summilux-R, Summicron-R, Zeiss, Voigtländer and others.
-
A 100% crop from a pic (lower), only the center part being relevant here. Camera was the Df. Left, Voigtländer 58/1.4; right, Noct Nikkor. Aperture is f/1.4 in both cases.
The Noct resolution is a bit lower, but contrast is higher. That sums it up.
At f/2.0 (third picture, this one using the D800), the situation remains the same.
-
Just made a quick & dirty comparison of center sharpness and contrast, on D800 and at about 1.4m distance: Noct, 50/1.2 AI, 50/1.8G, Zeiss Milvus 50/2, and Summicron-R (Leitax mount)
The Noct is distinctly more contrasty than all others, except Zeiss 50/2 that is both sharper and more contrasty at f/2. The 50/1.8G is on a par with the Noct.
At f/1.2 and f/1.4, the Noct is less hazy than the 50/1.2. At f/2.0, where the 50/1.2 is supposed to overtake the competition, The Noct is still more contrasty, and a wee bit sharper too. The difference is such small that it could be due to focussing precision. By the way, the 50/1.2, being an AI lens with longer focus throw, is generally easier to focus. I often read that, but direct comparison makes it obvious.
At f/2.0, the Noct and the 50/1.8G are roughly equivalent. At the same aperture, the Summicron is slightly sharper, but displays some haze and blue-ish fringes.
-
Most of these 'normal' lenses are pretty much similar in performance once you get to f/5.6 or perhaps a bit earlier. However, shooting night scenes makes the difference immediately obvious. The 50/1.2 really has bad coma, other f/1.4 Nikkor normals less so, and the Noct shines above them all.
Apart from the focusing throw, I see little difference between my Nocts (currently, one AI, one AIS). My sample of the AI is a tad sharper in the centre than the AIS, but has more field curvature. For night shots they are pretty much the same in their behaviour.
I have a feeling the Nocts do somewhat less well on digital than on film, particularly off axis. However, unfortunately at present I cannot examine that surmise further. Perhaps other Noct users can pitch in here?
-
I confess I never had a film Nikon. My brother did, though. I started with Canon (AE1 borrowed from my father, then T70, T90) before going digital.
-
It does not originate from Mr. Rockwell, but from another tester who compared several fast fifties including a Noctilux. The MTF charts dealt only with the center.
You are right in saying that this older test, which I published somewhere on Nikongear, did not include the "Gs"; nor did it include the Summilux-R, Summicron-R, Zeiss, Voigtländer and others.
NO! Ken Rockwell didn't credit his source?
-
I did not say that. He may have come to similar conclusions on his own, but he generally does not care about measurements.
-
By the way, did somebody keep a copy of the "tales of the 1001 nights" page about the Noct ?
-
The internet never forgets:
https://web-beta.archive.org/web/20160315172834/http://www.nikkor.com:80/story/0016/ (https://web-beta.archive.org/web/20160315172834/http://www.nikkor.com:80/story/0016/)
EDIT: Modified link to better version with samples
-
Gracias !
-
Airy
I've never done direct comparison with D800E or Df and when I shot on film I never got critical focus with it so EVERYTHING was out of focus.
I'm not unhappy with it on the D800E - still brings out that 3D roundness (have written about it before) that the other 50s (apart from Neo-noct) can't do well even on subjects 50m away. Getting correct focus is really hard and doubly so on the D800E which is very discerning. But that character or 'drawing' or whatever one wants to call it, is unique.
Mine does lose contrast wide open which is much more noticeable on D800E vs D700 or Df
JJ
These D800E and 100% crop
-
Mine also has lower contrast at f/1.2-f/1.4 compared with itself at smaller apertures (big jump at f/2). Wide open however, it still beats the other fast lenses I have around.
More to come but, for the time being, days are full and nights too short.
-
Daytime, nevertheless f/1.2
-
All relatively close focussed and wide open (but the second one, f/2.8). Firenze, yesterday.
-
Another one wide open. Focus slightly missed (wrong eye), but it was actually much darker than it seems here.
-
Nice work Airy
Can I ask how you tested the focus shifting? Mine doesn't seem to shift but I need to replicate your setup to make sure we are comparing the same thing
JJ
-
I've used the Spyder Lenscal target (with the oblique ruler) at about 1.5m, Noct mounted on D800, LV focussing using the Zacuto eyepiece.
-
According to my experience, focusing a Noct with any accuracy using this setup is next to impossible. Try the Df or similar to learn if you can improved and more consistent results.
-
...I forgot to mention : tripod...
-
According to my experience, focusing a Noct with any accuracy using this setup is next to impossible. Try the Df or similar to learn if you can improved and more consistent results.
Or perhaps an A7RII :)
-
By all means - if one has better luck than I did with those models ...