Getting eyes sharp on full-length portraits: VR-assist?

[studio460]

<p>I was just testing my AF-S Nikkor 85mm f/1.4G to determine approximate working distances for full-length shots, and discovered some shocking news. At about 15', my success rate was less than one in ten, less than 10%. At 5 feet, my hit rate was much better, about 90%.</p>

<p>In marked contrast, performing the same tests with my Sigma 150mm f/2.8 OS, I had nearly a 100% success rate of in-focus shots. Could this be a result of the Sigma's OS (optical stabilization) assisting in ensuring that my AF target's aim was actually over the subject's eyes at the time of shutter-release?</p>

[studio460]

<p><img src="http://studio460.com/studio460/150v-2.jpg" alt="" /><br /> Sigma 150mm f/2.8 OS @ f/3.0; handheld, OS-active; eyes are sharp at 100% [full-frame].</p>

<p><img src="http://studio460.com/studio460/150v-1.jpg" alt="" width="467" height="700" /><br /> Sigma 150mm f/2.8 OS @ f/3.0; handheld, OS-active; eyes are sharp at 100% [full-frame].</p>

[studio460]

<p><img src="http://studio460.com/studio460/85f3.jpg" alt="" /><br /> AF-S Nikkor 85mm f/1.4G @ f/1.4; handheld; eyes sharp at 100% [full-frame].</p>

<p><img src="http://studio460.com/studio460/85f-4.jpg" alt="" /><br /> AF-S Nikkor 85mm f/1.4G @ f/1.4; handheld; focus is soft at 100% [full-frame].</p>

[studio460]

<p>[Note: Both set-ups were lit with a single Nikon SB-800 at half-power (through a 5' octa), so the flash duration and effective shutter speed is approximately 1/800th.]</p>

[lornesunley]

<p>A couple of questions<br>

1) Why are the pictures using the 85 1.4 shot at 50 ISO?<br>

2) Are you using FOCUS priority and shooting with AF-S and single point set?</p>

<p>You depth of field with the 85 1.4 lens is about 0.8 feet (8 to 9 inches) so you should not have any issues with depth of field limitations unless you are recomposing after focus is achieved.</p>

[studio460]

<p>Lorne said:</p>

<blockquote>

<p>1) Why are the pictures using the 85 1.4 shot at 50 ISO?</p>

</blockquote>

<p>I was specifically trying to test the lens' ability to focus at f/1.4. I only had a minute to shoot this test, and it was a quick way to change my exposure.</p>

<blockquote>

<p>2) Are you using FOCUS priority and shooting with AF-S and single point set?</p>

</blockquote>

<p>Focus-priority, single-point, AF-C, AF-ON.</p>

<blockquote>

<p>You depth of field with the 85 1.4 lens is about 0.8 feet (8 to 9 inches) so you should not have any issues with depth of field limitations unless you are recomposing after focus is achieved.</p>

</blockquote>

<p>Yes, that's what my depth-of-field calculators tell me, but it sure doesn't appear to be that much. I am not recomposing in the 85mm shots--they are all fired with the AF point hovering over the subject's eye.</p>

[studio460]

<p>Just tested the 85mm again using focus-priority, single-point, AF-S, AF-ON, and changed my AF-fine-tune value from -12 to -13. But I noticed that the AF just isn't that accurate under low light. I aimed an LED flashlight onto the target, and the in-focus percentages increased. What I find curious is that the Sigma 150mm had a significantly higher in-focus percentage (with no AF-fine tune value active), regardless.<br /> <br /> Also, am I even using the focus mode on the D800E correctly? I have in the past been using focus-priority, single-point, AF-C, AF-ON, and manually selecting a specific focus point in the viewfinder with my Nikon D3s, and now, on my D800E as well. I'm using AF-C so that any minor movement by either me or the model, after initial AF acquisition, will be continuously adjusted. Focus-tracking with lock-off is set to "off."<br /> <br /> However, the top of the D800E's LCD display reads: "d51" although I am still selecting specific focus points in the viewfinder. Does the "d51" mean that dynamic focus mode is active? If so, is there a way to turn off the "dynamic" part so that the camera doesn't use any of the adjacent AF points to "assist" the selected focus point? I just read on another forum that the D800 operates in this mode differently from the D700/D3/D3s-series.</p>

[studio460]

<p>I think I found my problem. I have the AF area mode set to "d51" instead of 'S' for <em>single</em> on my Nikon D800E (setting the D800/E to "AF-S" does <em>not</em> mean you're necessarily also shooting in single-point <em>area mode)</em>. To select single-point area mode, press the AF mode button, then turn the sub-command dial until it reads 'S.' This is the same function handled instead, by the hard selector switch on the back of the Nikon D3s, where a more obvious-looking "point" icon sets the camera explicitly to "single-point."</p>

[studio460]

<p>Yup! That was it! My Nikon D800E was never really in single-point focus area mode (though, I thought I was). Now that it's properly set to 'S,' the 85mm f/1.4 is focusing as it should at f/1.4.</p>

[ilkka_nissila]

A fair comparison would be at the same aperture and magnification on both lenses.

 

You can expect the 85mm to require a slightly different fine tune setting as you move to really long distances but this

range should be ok with one setting. Low light does introduce scatter into the focus results.

[ilkka_nissila]

<p>I do think that VR can help with focusing on full body portraits, this has been my experience when using the 70-200II; the focus point and viewfinder image are more stable and so it is easier to hold them precisely on the eye. But for shorter lenses such as the 85mm I don't have problems due to the lack of VR. With the 200/2 II I find myself no longer using VR in normal use - my arms are used to the weight after some years and the weight itself stabilizes the image sufficiently. With high resolution sensors I've come to trust high shutter speeds (i.e. 1/1000s or faster with the 200mm) more than VR. When the light gets really low (i.e. when I have to shoot at slower speeds) I do turn on VR.</p>

[studio460]

<p>Thanks for your comments, Ilkka. If nothing else, I'm learning quite a bit about the lens inventory I already do have. I still can't believe I had my D800E set to the "wrong" AF area setting all this time. Everything is looking a lot better now that I've set that correctly.</p>

[Andrew Garrard]

<p>VR certainly helps framing (this is one reason I went with the OS 150-500 when I got one). I'm sure noise doesn't help, and nor will the shallow depth of field at f/1.4 (and there probably is more spherical aberration left even in the 85 f/1.4 AF-S wide open than with the f/2.8 lens), but I had no idea that d51 would affect the accuracy of a single focus point that isn't moving. I'm going to adjust my settings next time I shoot - I'd always assumed it was safe to stay in d51 in case I or the subject swayed a bit. Thanks for the heads-up!</p>

[studio460]

<p>Ilkka said:</p>

 

<blockquote>

<p>A fair comparison would be at the same aperture and magnification on both lenses.</p>

</blockquote>

<p>Yes, but I wasn't necessarily attempting a "fair" comparison. I just wanted to see how difficult it was to focus a full-length subject at each lens' maximum aperture. Thankfully, after correctly setting my area mode to 'S,' both lenses focused accurately.</p>

[studio460]

<p>Andrew said:</p>

<blockquote>

<p>. . . but I had no idea that d51 would affect the accuracy of a single focus point that isn't moving.</p>

</blockquote>

<p>Not to put too fine a point on it, but I believe the "d51" setting completely f****d up my focus accuracy. Recall that these are all handheld tests (the way I prefer to shoot portraits), so there is some inadvertent movement while framing. I didn't realize I was in "dynamic" mode in this setting--I thought it just meant that I could choose from 51 AF points. In this mode, I believe the "adjacent" AF points were simply "helping" too much, leading to other, close-proximity focus targets resolving (e.g., eyebrow, some strands of hair, etc.), rather than the intended (superimposed) target.</p>

[studio460]

<p>One more thing about accurate focus . . .the mathematically derived depth-of-field calculators seem to render very liberal results. Even with a circle-of-confusion set to 0.03mm, I shot some tests with the 85mm at about 14' where the calculated depth-of-field of "acceptable sharpness" is reported at 8" at an aperture of f/1.4. To my eye, I had at most, four inches of acceptable sharpness at this distance.</p>

<p>Also, I have a question. Where, ideally should you set your AF-fine tune adjustment? In the middle of the range, or one-third of the way in? Or, is this simply preference? Say, you want to acquire focus on the eyeball, but also want all eyelashes in focus as well--would I then fine tune for a two-thirds deep focus point (so that the range extends two-thirds of the way closer to camera)?</p>

[Peter_in_PA]

<p>Shooting handheld at f1.4 is going to be problematic, imho.</p>

<p>That said, if you get good results eventually... awesome!</p>

[studio460]

<p>Peter said:</p>

<blockquote>

<p>Shooting handheld at f1.4 is going to be problematic . . .</p>

</blockquote>

<p>Yes, it's definitely a challenge. Happily, on a static target, once I set my D800E's area mode correctly, I was able to nail a well-lit target about 14 feet away about 90% of the time, handheld.</p>

<blockquote>

<p>That said, if you get good results eventually . . . awesome!</p>

</blockquote>

<p>Yes! Executing this technique perfectly isn't easy, but the payoff is often pretty sweet!</p>

[Peter_in_PA]

<p>Ralph</p>

<p>Human beings (most of them) aren't static, however. And yes, I would probably enjoy shooting like that, too. I was jus' sayin'...</p>

[studio460]

<p>Peter said:</p>

<blockquote>

<p>Human beings (most of them) aren't static, however . . .</p>

</blockquote>

<p>Yes, that's why I typically shoot under AF-C, rather than AF-S. I recall my focus accuracy increasing once I switched to shooting that way with my D3s. I haven't shot any "real" shoots with my D800E yet. Another "practice" modeling session should prove this out with the D800E, now that it's correctly set to area-mode 'S.'</p>

[pete_s.]

<p>Ralph, regarding the DOF calculations.<br>

A CoC of 0.03mm is based on a 8x10 print from 1 feet distance and normal vision. If you are looking at your image and see that you have 4" DOF what <strong>you</strong> consider sharp you just need to pick a smaller CoC that gives you 4" DOF and then use that CoC when calculating DOF.</p>

<p>When I shot a lot of DX I would print the images at the intended size and then judge from the print what I thought was sharp according to my standard and reverse calculate what CoC gave me the right DOF. I don't remember the exact number but it was smaller than the standard. I also noticed that sharpening technique and post processing would (of course) affect the DOF.</p>

<p> </p>

[nsfbr]

Something else to consider, although you seem to have found your issue, AF on dSLRs is not done at f/1.4 for an f/1.4

lens. This is due, at least this is my understanding, to limitations on the size of the mirror. I think I read somewhere that it

is limited to around f/2.2. I have always assumed therefore that I need to assume some kind of fallout when shooting

wide open with fast lenses even in a best case.

 

I'm pretty sure someone will correct me if I'm wrong here. ;-)

[Andrew Garrard]

<p>I don't trust any of the DoF calculators as far as I can throw the computer I'm viewing them on. They're far too prone to making assumptions about viewing distance, print size, diffraction effects, and on occasion not allowing for the focal length of the lens. The maths involved is not that complicated (represent the lens, focal plane and sensor as parallel lines and draw a perpendicular line through them to represent the optical axis; then start drawing triangles). It would do many people who argue about CoC on the internet a lot of good to do this rather than to quote numbers from a web site, whether or not the web site has it "right". Excuse my bitterness, I've been involved in a lot of these <strike>arguments</strike> full and frank exchanges of views.<br />

<br />

I always assumed that dynamic 51 mode would work with one sensor at a time, but notice when the subject appeared to have moved laterally to a different focal point. I'd no idea it would make any one of them less accurate. (This is distinct from area "trust the camera to pick where to focus" mode.) I'll believe you if you tell me it does, though, since I've had accuracy problems with the D800. I'll try myself when I next get the chance.<br />

<br />

Joel: The f/2.5 limit that you're referring to is due to a fresnel lens in the finder screen, provided to give additional brightness when viewing through slow lenses. It has the result that, in the finder, you won't see depth of field change between ~f/2.5 and anything faster. However, it's nothing to do with the phase-detect AF sensor module.</p>

[studio460]

<p>Pete said:</p>

<blockquote>

<p>A CoC of 0.03mm is based on a 8x10 print from 1 feet distance and normal vision. If you are looking at your image and see that you have 4" DOF what <strong>you</strong> consider sharp you just need to pick a smaller CoC that gives you 4" DOF and then use that CoC when calculating DOF.</p>

</blockquote>

<p>Thanks! That makes sense. I suppose looking at a 36MP file at 100% requires an adjustment to the CoC specification in my depth-of-field calculator. I always thought those calculators were wildly liberal in their results.</p>

[studio460]

<p>Andrew said:</p>

 

<blockquote>

<p>I'll believe you if you tell me it does, though, since I've had accuracy problems with the D800. I'll try myself when I next get the chance.</p>

</blockquote>

<p>Well, don't quote me on any of this, since I'm just figuring this stuff out myself. If these series of threads have taught me anything, it's to test for yourself, and analyze the data you get with your own gear. Proper testing methodologies aside, at least you can get some primary data which isn't colored by bias or opinion (just by erroneous testing methodology, of which I am definitely guilty).</p>