6
\$\begingroup\$

First off, I'm deliberately referencing this question and have read both the original question and every answer, and I just wanna make sure to note that I know how to focus manually and am in fact moving the focus ring.

I recently bought not one but two old Nikon 50mm f1.8 Series E lenses. One of them is the newer variant, one is the older one (see also this page for visuals). I first purchased the newer model, but then realised that the aperture blades are oily, and because I found another lens very cheap, I also bought the older model and that one does not suffer from oil on the blades.

I own a Nikon D3300 which does not use a split-image prism and is generally not recommended for manual focusing. Nikon DSLR's -- even those cheaps ones -- do have a focus dot in the viewfinder which can be used even with old manual focus lenses. Having said this, here's my problem:

Using the 'newer' Series E on my Nikon D3300, I get sharp images when the focus dot indicates the image is in focus (The dot is visible and does not flicker). See this image:

Sharp image using new lens

However, on the older Series E, I get completely blurry images when the focus dot indicates the image should be in focus. (See the following crop, taken at F1.8, 1/2500th, ISO 200)

Blurry image despite focus dot indicating sharp image

I've double and triple checked that I indeed set the focus pixes to the middle element and have tried to account for every kind of user error I could think of. Lastly, I have used the Live-View to focus manually and the images that I get from that are certainly sharper (see following photo) than using the focus pixel. They are not as sharp as possible, just serve to illustrate the issue. Additionally, when I use Live-View to focus, switch back to the main viewfinder, the focus dot is not visible, indicating the camera does not think the image is in focus.

Better image using Live View

How can this be? There are no electronic contacts between camera and lens, so the focus dot should work 'optically' and should not care what lens is in front of it. I've tested this on another Nikon DSLR and the same result happened there, too.

\$\endgroup\$
7
  • \$\begingroup\$ It's not a shutter speed that is allowing vibration to blur the image, is it? \$\endgroup\$
    – Rory Alsop
    Commented Apr 20, 2017 at 16:39
  • \$\begingroup\$ Looks like vibration blur to me... \$\endgroup\$
    – BobT
    Commented Apr 20, 2017 at 17:24
  • 1
    \$\begingroup\$ @BobT at 1/2500 with a 50mm? You'd have to be turning cartwheels, not getting "vibration". \$\endgroup\$
    – hobbs
    Commented Apr 20, 2017 at 20:55
  • \$\begingroup\$ Oops... Didn't see that data. \$\endgroup\$
    – BobT
    Commented Apr 20, 2017 at 21:34
  • 1
    \$\begingroup\$ @benrudgers, it seems you have completely misunderstood the OP's question. \$\endgroup\$
    – lightproof
    Commented Apr 21, 2017 at 8:39

3 Answers 3

5
\$\begingroup\$

I think everyone is struggling with the example itself, but let me suggest an answer to the underlying question: How can the lens be badly out of focus when the green dot indicates it is in focus while, for the same lens using live view as a focus technique, it is approximately in focus.

There are two possibilities that come to mind. One is that the lens is badly flawed in some fashion that is throwing off the Phase Detection AF system (causing it to mis-read the focus state), but still allowing it to form a reasonably acceptable image when actually in focus. That seems unlikely but I guess possible. The optics involved in PD focus do not rely so much on apparent focus as angle of incidence of an object's light on different sensors. At least conceivably some form of distortion might mislead PD sensors. Have not seen it, but can imagine it is possible. Maybe.

The other is that the lens/body combination has an extreme back or front focusing issue. Focus, on most DSLR's in optical view finder mode (not live view) depend on a very precise match between the optical path for the AF system and the sensor. Fractions of a millimeter can through this off badly, causing the AF system to think it is in perfect focus when the subsequent image is not. Many bodies permit this to be fine tuned, the D3300 does not so far as I know.

Focus with live view uses the main imaging sensor so there are no path length issues and no fine tune is ever needed.

To confirm that is the issue, you need a more continuous set of subject matter in the frame, so you can see not just that it is out of focus, but where it did focus. A simple way to try this photograph a pole or similar vertical item (even a person's legs) in tall grass, where you can see where on the grass before/behind it actually focuses. The trouble with your shot is NOTHING is in focus, apparently, so it is hard to tell if it missed focus before/behind, or was just blurred for some other reason.

If it is actually focusing in front/behind consistently, you are probably out of luck with that lens (lacking AF Fine Tune, the alternative is sending both together to Nikon, crossing your fingers they do not mess up the body for other lenses fixing it for that one -- but likely they would just say "not supported"). Even if it had fine tune, your example makes it appear too extreme to fix with it (emphasis on "fine" tune).

Or... you may find something wrong in your experimentation and examples. Such things are always best done multiple times, on a tripod, locked down, and repeated from scratch checking all the settings each time. It's VERY easy to let experimental error creep in (I know from making many mistakes testing lenses and focus and fine tune).

Update: As discussed in the comments below, even if your camera had AF Fine Tune (and it does not), it likely would be unable to apply that to a manual focus lens.

\$\endgroup\$
11
  • \$\begingroup\$ Excellent root-cause analysis. \$\endgroup\$
    – scottbb
    Commented Apr 21, 2017 at 21:03
  • \$\begingroup\$ Thanks for the long and detailed answer! Here's another sample shot I just did. Please excuse the abhorrent light situation, this was shot at ISO 400, f1.8 at 1/60th. Link to new sample shot In this sample, the digit 7 of the calculator was supposed to be in focus (as the focus dot indicated.) So you are correct, the issue is backfocusing. However, what is the actual problem? Without having developed the film from my Nikon FE, I can say that whenever I use the exact same focus I got on the FE on my Nikon, the images are sharp. \$\endgroup\$
    – 0xdead10cc
    Commented Apr 21, 2017 at 21:48
  • \$\begingroup\$ I don't get what exactly the 'match of the AF-system' is. Nikon produced so many lenses for their cameras, all with different focal lengths and different physical dimensions, and there is something that is shared by all of them that in this lens is plainly wrong? How come I can focus accurately (at least so far I'm convinced) using the split-image prism? What is the actual issue? Is the lens broken? Is the issue the interaction between lens and camera and the lens itself is fine and this is just one example of old and new tech not getting along? \$\endgroup\$
    – 0xdead10cc
    Commented Apr 21, 2017 at 21:54
  • \$\begingroup\$ It does appear to be back focusing, and in that case does not look too far back for a body with AF Fine Tune to adjust, but not the one you are using. As to the comment above, while there are lots of difference in lenses over the decades, the F mount has been fairly standard in terms of the basics. From the new example, it just looks like it back focuses, which is not that unusual even for brand new lenses. You just don't have a camera that will let you adjust. \$\endgroup\$
    – Linwood
    Commented Apr 22, 2017 at 1:54
  • \$\begingroup\$ @Linwood AF fine tune does not work with manual focus lenses. They don't have a CPU and the camera has no way to identify them. \$\endgroup\$
    – lightproof
    Commented Apr 23, 2017 at 9:51
1
\$\begingroup\$

The issue with your images on the older lens don't have anything to do with focus accuracy.

I am positive that if you inspect the lens, you will find that it has extreme internal hazing and likely a lot of dust in the lens. Examine the lens with the aperture wide open (f/1.8) using a bright light source, and inspect the transmission of light from both ends of the lens. You should see that at least one of the internal lens elements is hazed, if not all. Despite noting that there is no oil on the aperture blades, if you inspect with a bright light, I believe you may see that they are all actually a little bit shiny, in line with the aperture blades being completely covered in oil, instead of shiny edges observed when the oil is just starting to work its way into the blades. Note that some dust is normal even in new lenses, but dust and matter towards the rear of the lens can have a greater effect on the image than dust closer to the front element.

The blurry image is actually in focus. If you look closely at the streetlamp head in the center of the crop, the close side (bottom right edge) has a slight amount of green chromatic aberration and the far side (top left edge) has a slight amount of magenta/purple chromatic aberration. This is expected behavior from a lens shot wide open at f/1.8, with a subject directly in focus, and a steep change in image contrast around the subject edges due to lighting. Also, the fence and the bricks behind it are still in focus (relative to the streetlamp). Any additional focus inaccuracy here would completely blur away the wires in the fence and the gaps between the bricks, but you can still make them out if you examine the image. The effect of haze and dust in your lens is to reduce contrast, and you can see that the features are there but much lightened relative to the other image.

Your camera is still able to determine focus as the haze in the lens is not enough to prevent phase AF from working on your D3300, provided you are using the focus dot indicator from the viewfinder and not live view. Live view cannot be trusted without visual aids and a closer magnification level, as the back of your D3300 has ~921k dots of resolution, but the sensor has 24.2M effective dots. You're only seeing about a 1/4 size view (downsampled) of the entire scene, and more of the image will appear to be in focus in live view than actually is, especially at f/1.8.

Last, how you focus with the lens will make a difference too, especially when working with very fast apertures or extremely close magnification (macro work). If you travel past the point of focus to a spot behind the streetlight lamp, then turn the focus ring back until you get dot confirmation, the closest surfaces of the streetlight lamp may not be in focus. Similarly, if you start far in front and close distance with focus until the dot lights up, the furthest visible areas on the streetlamp head may not be in focus. The bigger the subject, the more pronounced this effect will be. To make up for this, you should focus through and back, and then adjust focus to be somewhere just inside of the close side, using the travel amount from close/far side focus to give you an idea of how much working distance you have. It's harder to make this kind of adjustment in live view, as a lot of the scene will appear to be in focus when focusing close-to-far; and as you have noticed, your internal confirmation dot will almost always disagree with your visual confirmation this way.

\$\endgroup\$
4
  • \$\begingroup\$ Bold, assertive claim, but you make a compelling case. +1 \$\endgroup\$
    – scottbb
    Commented May 18, 2017 at 2:44
  • \$\begingroup\$ Thanks for your input! As interesting as your ideas sound, I don't think you are correct in this case: Firstly, both lenses look about the same to me when inspecting them with a bright light. Secondly, I'm positive the aperture blades are not covered completely in oil, as I can see specks of oil on the aperture. Third, I shot the scene above from quite far away, focused to Infinity, so that even if there was a back- or front focus, pretty much everything after the streetlamp should be in focus regardless of whether the actual subject was in focus. Cont... \$\endgroup\$
    – 0xdead10cc
    Commented May 22, 2017 at 21:46
  • \$\begingroup\$ Forth, when using LiveView I'm zooming in and I'm focusing while zoomed in, and as such I am in fact using a "closer magnification level". And finally, I have tested this lens on my Nikon FE and there is a discrepancy between when the split-image prism tells me the image is in focus and when the focus dot on my D3300 tells me it is, suggesting the lens itself might be fine, but just does not play nice with automatic focus confirmation. \$\endgroup\$
    – 0xdead10cc
    Commented May 22, 2017 at 21:47
  • \$\begingroup\$ To add to this: Using the Nikon FE I focused to Infinity, marked the exact position on the focus ring (curiously this is not the Infinity mark) and used this setting on my D3300. The image comes out sharp. When I use the internal focus dot, the camera tells me to focus past Infinity. \$\endgroup\$
    – 0xdead10cc
    Commented May 22, 2017 at 21:57
0
\$\begingroup\$

Not exactly the answer, but I had a similar situation with Helios-44 58mm f/2 lens (Soviet copy of Zeiss Biotar). My D700 was showing the focus confirmation dot when in reality there was a very pronounced back focus. At first I accounted that to M42-Nikon adapter which I used, as it has a glass element in it (which was actually not the case), but I soon found mentions of this problem elsewhere on the internet. Unfortunately, there were no explanations of the cause, only acknowledgment of the fact that it happens with some lenses.

I don't know how exactly Nikon's focus confirmation works, but I believe the reason is that some lenses render the image in a certain way optically that tricks the system to believe it's in focus. It could be either excessive spherical aberration (you focus wide open, so there could be lots of it depending on the lens), or any other - sorry, I'm not an optics engineer.

Eventually, I sold the lens as I couldn't use it. But there is a possible way to deal with such issue: there are aftermarket AF confirm chips that can be installed on manual lenses with Nikon mount, some of them are programmable and even have "AF adjustment". It's not really AF adjustment, as you install the chip on manual lens, but it adjusts the focus point perceived by the camera's electronics. I bought on such chip for it's EXIF capabilities (you can program lens focal length and F-number), but I found it has some limitations, so the camera's "non-CPU lens data" still works better. I never used the AF point tuning though.

Judging by your photos, it looks like your older version is overall noticeably less sharp than the newer one. If this is in fact the case, then there is definitely something wrong with your lens. Both old and new 50mm Series E lenses are identical in terms of optics and should give you almost identical results. They could differ, of course, as some lenses are a hair sharper than others, but not by that much. You should check for signs of your lens being dropped in the past, as well as the rear screws for the screwdriver marks to see if anyone have disassembled it. Any of this could lead to possible misalignment of the lens elements. Of course you could just have a very bad example.

\$\endgroup\$
2
  • \$\begingroup\$ Yes, the example is the issue. While it is hard to focus properly, I have certainly taken sharp images with this lens. One workaround that sort of works is to focus so that the focus dot appears and then to move the focus ring such that you're moving the focus in front of the object you're taken a photo of, just enough to make the focus dot blink. In my case, this results in decently sharp images. Obviously it's not as quick, but then again, using manual focus lenses on newer bodies is never quick. \$\endgroup\$
    – 0xdead10cc
    Commented Apr 23, 2017 at 11:25
  • \$\begingroup\$ using manual focus lenses on newer bodies is never quick - that depends on your skill mostly, but on top bodies you often can swap focusing screen for the one that has split image and/or microprism aid. I have the same screen as my FE in D700 and it works great. It's pricey though. Concerning confirmation dot - yes, your technique could work if you know the exact state image is in focus. In my case it differed depending on the subject distance. Another thing is to always focus moving focusing ring in one direction, because other way around is usually less precise. \$\endgroup\$
    – lightproof
    Commented Apr 23, 2017 at 11:41

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.