The actual focus distance is irrelevant when considering the original D800 "left focus" problem.
Whichver AF point is selected, the part of the subject covered by that sensor - assuming you don't re-compose - should be in focus. If your D800 had the problem, that wasn't the case when the leftmost AF points were used.
>So if I focus far left (76 on the scale), and 76 is the sharp >focus point, I should be good?
Basically, yes. The original D800 "left focus" problem was very obvious - in cameras which suffered from it. Quite a few people put a lot of time and effort into using things like Lensalign - and there's nothing wrong with that, but a simple test setup with a flat, high-contrast target parallel to the sensor and in good light should indicate whether your camera is OK or not.
<<>So if I focus far left (76 on the scale), and 76 is the sharp >focus point, I should be good?>>
As said "yes". When I bought my D800 my testing was very unscientific. I simply popped a 50mm f1.8 lens on, focused on the shops book shelf and moved the focus point (51 points) from far left, far right, top and bottom. Each was in sharp focus. Job done
Sun 13-Oct-13 03:37 PM | edited Sun 13-Oct-13 03:48 PM by gforeman
Here are two shots I just took. First was far left focus
point 76, 2nd was center focus point 64 on the horizontal
scale. These are full size out of camera:
I can't seem to figure out how to make these a LINK and not
embed them. They are huge. Just copy and paste.
Sun 13-Oct-13 05:41 PM | edited Sun 13-Oct-13 05:43 PM by gforeman
Sorry you will have to copy/plaste. Too large to embed. They
all seem to be focused on 64, even though the focus points
were different in Live View
I don't think you have a problem. I see little or no difference in sharpness between the three shots. As mentioned, if your camera was faulty, mis-focus with the left AF point would be a LOT more obvious.
How did you test the 24-70? Difficulty with zooms I find is that rarely and in my experience with lens align, both settings throughout the range were the same. I only ever regarded this testing as reliable at a single focus length, or perhaps use an average of both ends. I now only and rarely use it for primes.
>How did you test the 24-70? Difficulty with zooms I find is >that rarely and in my experience with lens align, both >settings throughout the range were the same. I only ever >regarded this testing as reliable at a single focus length, or >perhaps use an average of both ends. I now only and rarely use >it for primes. > >Richard.
I had the 24-70 set for 50mm.
I was just trying my Tamron 28-75 f/2.8 with the D800, and I cannot even get it to focus in Live View. I must have something wrong. I'll try it on my D700.
I had the Tamron 70-300mm VC until very recently and it worked faultlessly on both my D800 and D7100. I remember only using it to focus once using live view on my D800 with no issues. Can't speak about the 28-300mm, but would have expected a similar result on both my cameras.
If you can get it to focus ok via the viewfinder, and the resultant images are acceptable, then perhaps the camera setup in someway is to blame. I really have little experience of live view with those lenses, but extensive experience with my wide angle 16-35 and 24-70mm Nikon's I find that the live view works very well on both my cameras.
>I have two Tamron Lenses. 28-75 and 28-300 VC. > >In live view (on the D700 or D800) they will not get a good >focus. I can focus through the viewfinder and it's okay, but >never in Live View. > >This really bothers me. These are both only a few years old >(got them when I got my D700). The 28-33 has actually never >been used.
Normally this is just a third-party lens problem, they will not AF in LiveView (i.e. with Contrast Detect AF). I had to send my Sigma 50mm f/1.4 in for a firmware update and when it came back, it would then focus in LiveView with CDAF.
Somewhat of a common problem, and this may be what you are experiencing...
I did shoot them in live view with a cable release. Maybe I am doing it wrong...
I just removed it from the tripod, and doing real-world test shots and it looks good. I don't really shoot live view much. The fact the Tamron's do the same on both cameras means it is not probably a camera issue.
Mon 14-Oct-13 06:14 AM | edited Mon 14-Oct-13 06:15 AM by briantilley
>I did shoot them in live view with a cable release.
Ah - in that case I respectfully withdraw my previous conclusion...!
The D800 (like the other current Nikon DSLR's) has two completely separate AF systems - one used in Live View and the other in normal viewfinder-based shots. The "left focus" problem that you were testing for only ever happened with the latter system, so testing using Live View does nothing to show whether your camera has the problem or not.
My other comments, that whatever is under the chosen AF point should be sharp and that if the problem is present the effects are pretty obvious, still apply.
> >>I did shoot them in live view with a cable release. > >Ah - in that case I respectfully withdraw my previous >conclusion...! > >The D800 (like the other current Nikon DSLR's) has two >completely separate AF systems - one used in Live View and the >other in normal viewfinder-based shots. The "left >focus" problem that you were testing for only ever >happened with the latter system, so testing using Live View >does nothing to show whether your camera has the problem or >not. > >My other comments, that whatever is under the chosen AF point >should be sharp and that if the problem is present the effects >are pretty obvious, still apply.
I'll do another test today, using the viewfinder for focus only, but after the 100 or so test shots last night (handheld) and using the focus pot all the way to the left, I think I may be okay.
Hi Brian I am a new D800 owner purchased about a month ago. I don't think I have the left focus problem, but would like to ask your opinion.
When I asked my dealer about the issue he told me that yes, there was a left focus issue with early units at initial release, but that Nikon corrected the problem. Also to me that if I had the issue it would be very apparent in the viewfinder when trying to use left focus points.
>When I asked my dealer about the issue he told me that yes, >there was a left focus issue with early units at initial >release, but that Nikon corrected the problem. Also to me that >if I had the issue it would be very apparent in the viewfinder >when trying to use left focus points. > >Would you concur with his statements?
I'm not Brian and I'm climbing out far on a limb by disagreeing with him, but I think your dealer is full of it about the view finder. Short of using a magnifier (and maybe not then), I doubt you could see, much less would it be "very apparent" in the viewfinder. I think it would be possible to see it on the LCD if you knew what you were looking for, but I doubt, and certainly have never heard of anyone, seeing it in the viewfinder.
I think those who have followed the story from the beginning believe Nikon figured out the problem and made it better in later shipments. But I think we believe that mostly because it's the only sane thing they could have done, and the number of complaints did drop off. But Nikon never really admitted to a problem, and certainly never let it be known there's a specific range of serial numbers or anything so definitive (but think about it - if they did, everyone below that would want a fix or new camera because it was "defective").
All that said, I wouldn't let this stop me from buying one. Especially now. And I'm not really upset I had to have mine fixed -- they gave me no hassle and fixed it on the first go.
I would agree, Nikon is not going to admit anything, its that old cost effective thing, like how many owners are going to notice the focus problem. Those that do and send in their cameras, we'll fix it, just like they for the most part did with the D600 dust issue.
So as a new D800 owner in my case how does one know if I have a left focus problem or not? Do I try a test as has been mentioned here to put on my 50MM 1.8 lens set it to f/2.8 and run through the focus points left to right and top to bottom, then look at the images in LR to see if each one is in focus?
I am not familiar with a lens align unit or know anything about using one.
Sun 20-Oct-13 12:37 PM | edited Sun 20-Oct-13 01:02 PM by Gromit44
>So as a new D800 owner in my case how does one know if I have a left focus problem or not?
You can test for the left AF issue using a simple high contrast b&w target - e.g. a series of black vertical lines printed on a piece of white A4 paper or card.
a. Mount the target on a wall and illuminate it with a constant light source. b. Mount the camera on a tripod and fit the 50mm lens. c. Switch to manual exposure and open the lens to its widest aperture - take a test shot to ascertain correct manual exposure. d. Use a cable release or remote shutter release. e. Get the camera's sensor as parallel as possible to the wall. Don't move the camera from now on.
1. Make a small aiming cross near the centre of the target.
2. Slide the target along the wall so that the cross is under the camera's CENTRE AF point - defocus the lens, refocus using AF and take one picture. Switch to Live View, position the LV focus point over the cross and take another picture. Defocus lens and turn off Live View.
3. Repeat #2 for the FAR LEFT AF point.
4. Repeat #2 for the FAR RIGHT AF point.
You now have six shots in three pairs:
PAIR A: CENTRE AF and CENTRE LV PAIR B: FAR LEFT AF and FAR LEFT LV PAIR C: FAR RIGHT AF and FAR RIGHT LV
Live View produces sharper focussing than AF and we know that the problem only affects AF - so look at the difference in sharpness between the two shots in each pair.
If the camera has the left AF issue there will be a bigger difference in PAIR B shots than there is in PAIR A or PAIR C shots.
If the camera doesn't have the left AF issue, PAIR B shots will have a similar difference to PAIR A and PAIR C shots.
Bear in mind though - the centre of a lens is always sharper than the edges, so the centre shots will be sharper than the left or right shots with both LV and AF focussing. So a straight comparison between PAIR B and PAIR C should show up the problem if you have it.
I'm not sure I did this test correctly or not, but here are six images taken using my 50mm 1.8 on my D800. I used a tripod and cable release, taking center af, left af, right af, then lv center, lv left and lv right using af in lv as well as out of live view.
When looking at them in LR, I can't see much difference.
I'd appreciate any feedback. If I did the test incorrectly I can do it over again.
Wed 23-Oct-13 11:57 AM | edited Wed 23-Oct-13 02:57 PM by Gromit44
I've looked at these in Photoshop but can't tell much because:
a) Each shot has three targets instead of one and I don't know where the focus point was in each case. b) The jpegs have been resized to 1200x801 - too small to assess focus on any one target even when viewed at 100% or greater. c) They're underexposed and the white balance is off. d) The lighting appears to have changed between shots - e.g. left AF is brighter than left LV even though they're both f/1.8 @ 1/40s.
This is how I did it:
1) Instead of using three targets, use only one. 2) Distance from front of 50mm lens to target - 70cm approx. 3) Focus on the exact same spot on the target for each shot (see green arrow below). 4) The distance from camera to the focus spot on the target must not change at all between shots. 5) The camera sensor must remain perfectly parallel to the target. 6) The camera must not move at all between shots - slide the target across instead (see 'Positioning the target' below).
Order for taking the shots:
Select CENTRE AF point in viewfinder. Slide target so that focus spot is exactly under CENTRE AF point. Defocus lens by turning ring to closest focus. Focus using AF. Take CENTRE AF shot. Defocus lens by turning ring to closest focus. Switch to Live View and move LV focus box over focus spot on target. Focus in LV. Take CENTRE LV shot. Switch LV off.
Select LEFT AF point in viewfinder. Slide target so that focus spot is exactly under LEFT AF point. Defocus lens by turning ring to closest focus. Focus using AF. Take LEFT AF shot. Defocus lens by turning ring to closest focus. Switch to Live View and move LV focus box over focus spot on target. Focus in LV. Take LEFT LV shot. Switch LV off.
Select RIGHT AF point in viewfinder. Slide target so that focus spot is exactly under RIGHT AF point. Defocus lens by turning ring to closest focus. Focus using AF. Take RIGHT AF shot. Defocus lens by turning ring to closest focus. Switch to Live View and move LV focus box over focus spot on target. Focus in LV. Take RIGHT LV shot. Switch LV off.
Processing the files:
Upload 6 NEF files to computer - label them CENTRE AF, CENTRE LV and so on. Open the files in Photoshop, crop them so that only the square target box is left. Save as TIFFs using a name suffix (don't overwrite original NEFs).
Take CENTRE AF shot and add 100% width to the right of its canvas. Import CENTRE LV shot and position it on the blank area of the canvas. You now have CENTRE AF and CENTRE LV side by side in one Photoshop file. Save it as 'PAIR A - CENTRE AF & CENTRE LV'. Repeat the Photoshop process for the other two pairs.
Positioning the target:
I mounted the target to a piece of hardboard. Then I nailed a wooden batten to the wall. To position the focus spot exactly under the camera's Left, Centre and Right AF points, the board was slid horizontally along the batten - this is much easier with an assistant. Four cable clips were tacked to the wall to hold the top of the board flat against the wall (the board slides under the plastic arms http://images.beatsons.co.uk/images/products/zoom/1350489529-93962100.jpg).
The green arrow shows the focus spot on the target - the centre of the centre circle.
Grom, Thanks for the detailed explanation of how to take the photos. What I did was very similar to in that I set the camera up by first taking a test shot in manual mode.
Then I took the first shot labeled afcenter with the focus point on the center circle on the target, then I moved the focus point to the left and shot the center circle on the left target, then the same on the right target. The camera was stationary on a tripod during each shot.
Then I repeated the same steps using Live View.
The images are very small due to the limits of a maximum of 300K on the website.
Can I assume that in the final process you describe of cropping and adding to the canvas, that when I save the tiff for web in PS that I can get it small enough to again upload?
If so I can try it again today, moving the single target as you suggest.
If you're going to re-do the test, then one other step that I would recommend - as also mentioned by Simon above) is to de-focus the lens by setting it minimum (or maximum) focus distance before each test shot, to make sure that you start from the same degree of "unfocusedness" each time.
>The images are very small due to the limits of a maximum of 300K on the website. > >Can I assume that in the final process you describe of cropping and adding to the canvas, that when I save the tiff for web in PS that I can get it small enough to again upload?
To reduce file size for uploading to the website, start with each side-by-side file in Photoshop and save it as a JPEG (also save the .psd file just in case).
If the JPEGs are still too big, crop more tightly but don't cut off any of the target focus spot (the vital bit) - crop something like the one below.
The reason for using the side-by-side trick is to have both the AF & LV shots from one focus point viewable at the same time - that way we can gauge the difference in sharpness more easily. So, what we need is three of these - one for left, one for centre, one for right (include ident labels if possible):
Thu 24-Oct-13 08:53 AM | edited Thu 24-Oct-13 08:58 AM by Gromit44
That's the way to do it Jim - exactly right.
If we look at the Centre pair (top image above), AF and LV are very close in sharpness. LV may be a tiny, tiny bit sharper.
If we look at the Right pair (bottom image above), LV is sharper than AF but not by much - the difference is still tiny.
Now for the Left pair (middle image above), again LV is sharper than AF but only by a small amount - the difference is very close to that of the Right pair.
Everything here is just as we'd hoped to see.
Both Centre pair pics are sharper than the Left or Right pair pics because the lens is always going to be sharper in the centre. The Centre AF point is very, very good - almost as good as LV.
The Left pair and Right pair pics are softer than the Centre pair and LV is slightly better than AF in both cases (as we'd expect - see ***) - BUT - the crucial point is that there's no significant change in the AF to LV differences of these two pairs. Hence, I would say your AF is as good as it gets and your camera definitely does NOT have the left AF issue. I'd be very, very happy with the AF focussing on that body.
*** AF (Phase Detection) is faster than LV (Contrast Detection) - it uses a separate sensor for focussing but the light path is more complicated and it's usually less accurate than LV.
Gromit44 and others replying to this issue, I really appreciated this post and the detailed description of how to test for the left focus issue. I've read through a couple other tests and they are not quite as clear, but close. This is the best I've seen. I just bought (2) D800e's (wife+me) and was very concerned about the left focus issue. I did this test on both cameras and both are fine and focus correctly. I used a 50 1.4g lens to test and it worked well. For others who want to know my buying details on these bodies to help in their ordering choices... ordered from B+H in early Nov., 2013... about a week later... got serial numbers in the 3012xxx range. Both are working beautifully.
>I'm not Brian and I'm climbing out far on a limb by >disagreeing with him...
Don't worry, Linwood - we thrive on discussion and (polite) disagreement...!
I've never owned a D800 (yet...) so my views on this are all second-hand or worse, but much of the information I've read about the left-side focus issue indicated that, when a particular D800 suffered from the problem, it was bad enough that the lack of sharpness was very obvious. I thought some people reckoned they could even see the result in the viewfinder but maybe that's not the case.
It's true that Nikon never came out and said that "this problem is fixed from camera serial xxxxxxxx onwards", but if it were still happening we would certainly have heard much more "noise" about it. It's safe to assume that it has been rectified and is no longer something to worry about.
What I'm really saying is that it should not be necessary to undertake complex testing, or view images at large magnifications, just to establish whether a camera has the problem, and the chances are that a current or recent example will be fine. If that's what the dealer meant - "the problem is easy to spot, and newer D800's are probably OK" - then I think it's a fair statement
>All that said, I wouldn't let this stop me from buying one.
Sun 20-Oct-13 12:16 PM | edited Sun 20-Oct-13 12:21 PM by Ferguson
If helpful to see one data point, below are the images I used to show Nikon, and they agreed it needed repair.
The left image in each was live view manual focused, the right was AF. More than one was taken in each but these were representative.
Enlarged the lack of clear focus was obvious.
The third is the same scene from the center focus point. The right looked almost identical.
Very clear enlarged, not so much as a full frame. I'd say the top scene is maybe, perhaps about the quality of what you see in a viewfinder? Maybe not, depends on your eyes.
In fact if I look at the out of focus left frame at 4x6 I really don't notice at all, at 8x12 I can see it but you don't notice at first glance. It does show up easily at that size if it was something more natural, like the eyes of a portrait. Then your first look is "something is wrong".
I had convinced myself (without testing) I had no problem for quite some time before I started hitting combinations where it showed, and tried to look more closely.
The fourth is the same scene, with left focus (live left, AF right, same magnification as the top, same camera distance) with a 14-24/2.8 at 14mm. This also shows it fairly clearly, but I think the wide angle's deeper DOF and 2.8 make it a bit more subtle. It was hard to see on my 24-70/2.8 at all (I can't quite explain that since it should be more shallow DOF).
I'm sure different cameras had more or less problem of course.
<< If that's what the dealer meant - "the problem is easy to spot, and newer D800's are probably OK" - then I think it's a fair statement>>
I agree, the dealer said as much to me and was pleased to help me conduct my extremely unscientific tests in store. He also said that after checking a few earlier cameras that were returned, the problem was obvious when they did have the left focus problem.
I suppose that's the beauty of buying from an official Nikon high street retailer, as he told me that the shop had taken the problem on board and issued a replacement new camera at best, or loaned the same model until the camera was repaired. I believe that he's done the same for folks who found immediate D600 problems though I can't confirm that.
Mon 14-Oct-13 11:57 AM | edited Mon 14-Oct-13 02:21 PM by Gromit44
>If I center focus, then move the focus point far left and shoot again, isn't the distance to the far left further than the center? So wouldn't the focus be off because of distance? >
As I understand it the answer to this question is no because the camera doesn't use the main imaging sensor for autofocus. Instead it uses a dedicated AF module (in the D800 it's called the Multi-CAM 3500FX AF sensor module).
Light from the lens passes through the main mirror and onto a secondary mirror behind. The secondary mirror redirects it to the AF module. Inside the module light passes through a small 'separator lens' which produces two images and these land on a bunch of tiny sensors - one pair for each AF point. The distance between the two images on any given pair of sensors represents the amount of defocus for the corresponding AF point. This is the same principle as rangefinder focussing - perfect focus is only achieved when there is zero distance between the two images (i.e. they coincide exactly).
So if you're using the centre AF point, the camera compares the images from the 'centre pair' of sensors and sends a focussing instruction to the lens (let's call it instruction A). If however you're using the left AF point, it compares the images from the 'left pair' of sensors and sends a different focussing instruction to the lens (let's call it instruction B). The difference between instruction A and instruction B would cater for the difference between distance 2 and distance 1 in your diagram.
At least, that's my understanding of how the AF system works. There's some useful info on these pages.
I've tweaked the post above Richard - it's a reads a bit better now.
This is the light path in abbreviated form: Lens > main mirror > AF mirror > AF module (separator lens > AF sensor pairs).
I can't find a picture of the innards but there must be 51 AF sensor pairs in the D800's AF module (Multi-CAM 3500FX) because the camera has 51 AF points. The camera picks the pair that corresponds to the AF point you've chosen in the viewfinder and measures the distance between the two images from the separator lens. The distance is converted into a plus or minus reading which is sent to the lens. The AF motor inside the lens then turns one way if the reading is positive and the other way if it is negative.
This may be obvious but it comes up often in the whole left-focus discussion.
The classic focus and recompose is fundamentally flawed. It works with slower lenses with deep depth of field, but it doesn't work with wide aperture lenses, and those are often used on the left focus test.
Due to misunderstanding, people either try to focus and recompose to check left focus, or understanding their left focus is off they focus and recompose to avoid the left focus point. Both have issues.
To see the difference see the diagram (inspired by the diagram of the OP).
In a classic recompose situation, one focuses head on (as on the left), then rotates the camera without changing focus. The expectation is that the subject remains at the same distance (true) so will remain in focus (false).
The reality is that the in-focus area is a plane, not a portion of a sphere. So when you rotate, the in-focus area near the subject actually moves behind the subject.
If you are using (say) an 4 lens, this might be fine, switch to an F1.4 lens and the subject now is awfully blurry.
Testing for the D800 never involves recomposition like this. Working around the left focus issue with recomposition sets you up for disappointment if your lens is fast.
The "fast lens" issue is mostly an issue for those whose focus, so to speak, is on the best possible bokeh. Just because a lens is fast, e.g., f/1.4, doesn't mran it always has tonbe used wide open. Rather, all lens can be stopped down to a wide cloice of apertures, based on the ISO limits imposed by the camera and photographer.
>Just >because a lens is fast, e.g., f/1.4, doesn't mran it always >has tonbe used wide open. Rather, all lens can be stopped >down to a wide cloice of apertures, based on the ISO limits >imposed by the camera and photographer.
Yes, but that doesn't mean that the "fast" problems end there. There are both lens compromise issues that show up (glare for example), but more importantly no matter how stopped down for the shot, the focus decisions by the AF system are done wide open. That's why focus shift is a recurring theme in fast lenses, where you focus at F1.4 but shoot at F4 -- and find it's moved where you focused. Then cross your fingers and hope the DOF still covers what you wanted in focus.
With regard to the D800, I've never read anything definitive, but I think its prevalence with fast lenses is exacerbated by the shallow DOF, but that the problem showing more strongly there originates from the wide aperture itself. Anyone know for sure?
AF systems are sensitive to aperture, some topping out at F5.6, some like the D800 at F8. People often think that speaks to how well they focus with low light, but that's not it (if it was, they would talk about EV not F stop). It is about the angles at which the light cone is striking the sensor from the lens and what that does to the phase detection.
So some issues may remain if you shoot a very fast lens stopped down.
I'd be curious to see (say) a 50/F2 compared to a 50/1.4 on the left focus test, shooting both at F2 (but focusing at F1.4). I wonder if the left focus comes out identically with them at F2 for both.
>AF systems are sensitive to aperture, some topping out at >F5.6, some like the D800 at F8. People often think that >speaks to how well they focus with low light, but that's not >it (if it was, they would talk about EV not F stop).
In fact, I think "they" are correct. AF performance depends on the amount of light reaching the AF sensors - which in turn is governed by the f/stop. EV encompasses aperture AND shutter speed, but shutter speed is not relevant to AF performance.
>>AF systems are sensitive to aperture, some topping out >at >>F5.6, some like the D800 at F8. People often think that >>speaks to how well they focus with low light, but that's >not >>it (if it was, they would talk about EV not F stop). > >In fact, I think "they" are correct. AF performance >depends on the amount of light reaching the AF sensors - which >in turn is governed by the f/stop. EV encompasses aperture >AND shutter speed, but shutter speed is not relevant to AF >performance.
Well, during exposure.
During AF processing, EV depends on aperture and ambient light.
If you look at the Nikon specs, you will see they SEPARTELY give specifications for AF, for example the D4 is said to work up to F8, but it also says:
>> Detection Range -2 to 19 EV (ISO 100, 68°F/20°C)
This is obviously about available ambient illumination, not (just) aperture. This would appear to indicate that you could focus with an F8 lens or an F1.4 lens at EV=-2, which is a HUGE range (if you were taking an image it would be 240 vs. 8 seconds.
Anyone who has tried a narrow aperture with TC (say the 200-400F4 with a 2x) outdoors in bright light on an older camera knows it is not completely reliable. Yet the EV there is well into the middle of that range.
At the limits does it work better with more light? Yes. But the limit is not a hard, fast, line -- above Nikon's limit it starts to be unreliable. Regardless of being in bright light.
But there are two separate limits -- the F-stop relates to the way the light strikes the AF sensor for its phase detection. The EV limit is about how much light is required.
It's not either or, it is both. But they are separate limits for different purposes, I think.
You may be right on this Linwood - it clearly is related to the amount of light. But focus always occurs wide open for the mounted lens, so stopping down does not matter for the purpose of AF.
The key in Phase Detect AF is there needs to be enough light for the camera AF system to recognize a difference between slightly out of focused elements and better focused elements. Recognizing out of focus is easier with more light.