Monthly Archives: September 2009

More on autofocus errors

While reading Thom Hogan’s detailed review of the Nikon D300 today, I came across this interesting little bit in his discussion of the autofocus system on that camera:

The camera detects the current focus situation, calculates where focus should be (yes, it knows not only how much out of focus, but which direction, the error is when you initiate a focus operation), then tells the lens to move a set amount. To prevent hunting, if the new error after the focus is performed is within a reasonable tolerance, focus isn’t attempted again. Tolerances are the bane of accuracy…

In other words, what I was trying to explain last week, in my entry on using a superzoom lens as a focusing tool in wide angle shots where deep depth of field is needed, appears to be correct: If the camera thinks it has turned the lens far enough, i.e. “within a reasonable tolerance”, then that’s where it stays, rather than trying to find the absolute optimal position. And this would be why I was sometimes getting shots where the farthest background or closest foreground details were out of focus, even though the depth-of-field at my chosen aperture should have been adequate to prevent that from happening.

And what I secretly love about this is that I managed to reason it out just from using the camera. Heheh. ;)

Leave a comment

Filed under Nikon, photography

So, what lens should I get? :)

I’ve been itching to upgrade my lens collection lately, but of course it’s a challenge for a person in my position to figure out which way to go. By “my position”, I mean “not having a lot of money”, and also “not having much in the way of lenses to begin with.” That means almost anything would be an improvement.

Here’s the current situation: My camera body is a Nikon D40. For lenses, I have the Nikon 18-200mm VR, and a non-functional 18-55mm, where the autofocus doesn’t work. For those not familiar with the 18-55, the manual focusing on that lens is especially tricky, in that you turn the lens only about 1/8th of a revolution to go all the way from closest focus to infinity. Combined with the difficulty of using the D40 viewfinder for focusing, it makes the lens virtually unusable in many situations. I have been meaning to send it in to get it fixed, but haven’t gotten around to it yet. If I got it fixed, it would serve as my “macro” lens—it doesn’t focus to 1:1, but I don’t need to focus that close anyway. I find that at 55mm, it focuses close enough for my purposes. Here is an example:

False Sunflower

(Actually, that pic was taken after the autofocus failed, so it is clearly possible to get good results on manual focus. But it took a bit of extremely careful, minute rotations of the lens barrel to get that shot into good focus. Autofocus probably would have been easier.)

The image quality of the 18-55 is also better, in general, than that of the 18-200 VR. Even the barrel distortion at 18mm isn’t as bad, although it is certainly obvious enough to be seen, if there are straight lines in the shot.

So, given my current situation, I’ve been thinking about a few specific expansion options. There are four that come in under $1000 (I would not feel comfortable spending a huge amount of money right now—in fact, I probably am not going to go with any of these, since I really shouldn’t be spending anything right now, heheheh). They are:

10-24mm f/3.5-4.5G ED AF-S DX – A good DX landscape lens, which would give me some wider angle options than what I currently have, and better quality overall. A possible drawback to this choice is that I am not sure whether I would actually need to go this wide. The widest lens I have used on a regular basis was on my old Konica SLR, a nice little 24mm prime. In general I found it to be a bit wider than I needed it to be, except for in a few limited circumstances. It should be noted, however, that in the years I used this lens (roughly 1994 through 2005), I didn’t have as much idea of how to use a wide lens as I do now. I guess the best thing to do would be to actually try out a superwide lens before committing to buying one.

16-85mm f/3.5-5.6G ED VR AF-S DX – A general purpose replacement for the 18-200, to be used primarily for landscape shots. It would give me a bit more reach on the wide end, would force me to narrow my mental focus while in the field (by depriving me of instant access to the longer focal lengths), and would probably give me better image quality than the 18-200. One question with this lens is whether, by getting it, I would be making myself even less likely to ever try a superwide lens. The 16mm wide end would put me at an equivalent angle of view to the 24mm lens on my old camera.

Another question relates to the fact that it is a DX lens. It can be argued that the best Nikon camera body for landscapes is currently the 24 megapixel D3x, which happens to be an FX camera. In that case, the 24-70mm f/2.8 would be a better option. That, however, is getting pretty speculative, given the great expense of that equipment.

70-300mm f/4.5-5.6G IF-ED VR AF-S – This would extend the reach and overall quality on the long end. I’ve been really interested in bird shots lately, and let’s face it: The 18-200 is not a very adequate lens for that purpose. It’s not long enough, and the quality at 200mm is not the best. I’ve gradually learned, through trial and error, that my best birding configuration with this lens is to set it to 200mm and f/6.3 (opening up all the way to f/5.6 seems to result in lower quality), and hope for the best. At that aperture, I find that I often need to bump the ISO up to 400 (or even 800) to get an acceptable shutter speed. I also find that autofocusing doesn’t always lead to the best results, although I think this is more due to trying to focus on something that’s too small in the frame. When a bird only fills part of the center focus point, the camera will often focus on what’s immediately behind the bird, instead of the bird itself. I’ve learned to try and get the bird positioned right at the top center part of the marked focus area in the viewfinder, since that seems to be where the focus point looks, but even with that, I often find I’ll get better results if I go through the trouble of manually adjusting the focus. I haven’t been able to detect any clear focusing error in the lens itself, luckily—it seems to be entirely due to just pushing the lens beyond what it’s designed to handle. Being able to zoom in more would certainly help.

The reason I hesitate with the 70-300VR is that it’s not really the 300mm option that I want. What I really want is the 300mm f/4D ED-IF AF-S, along with a 1.4x teleconverter, which would take me up to 420mm f/5.6. This would be the same as having a 630mm lens on a film SLR, which is more telephoto power than I thought would ever be within reach for me. I practically drool at the possibility. :) The problem is that this combination would cost me close to $2000, and it would not give me VR, which I find very helpful during focusing and framing. Assuming Nikon updates this lens with a VR “G” version at some point, I’d expect the price would go up by at least a couple hundred dollars, plus there are rumors of further upcoming adjustments in pricing thanks to the excessively high value of the Japanese Yen. So it’s unknown whether getting that combination will be a likely possibility over the next few years, thus I am considering the 70-300 VR as a “best I can do for the foreseeable future” option. (I still need to look at 3rd party offerings, too, admittedly.)

35mm f/1.8G DX – This lens is tempting for two reasons. One is that it’s just cheap, compared to Nikon’s other lenses. The other is that it’s a prime lens, roughly in the “normal” range (I actually consider it on the long side of normal, since the “normal” lens on my old camera was a 40mm, and this 35mm comes in at about 53mm equivalent, when adjusting for the D40’s crop factor). Using a lens like this would be an interesting creative challenge. Part of it would involve leaving the other lenses at home, which would force me to work with just the one, single focal length. I wonder what I could come up with? I wonder if I would hate it? It’s been a long time since I was limited to just one focal length.

I also think this would be a handy lens to have for trips to museums and the like. It’s much smaller than the 18-200, which means security people would be less likely to look askance at it. :) (If my 18-55 worked properly, this reason would go away.)

Drawbacks to this lens? Other than the concern that using a prime lens would drive me absolutely nuts, the main issue would be the lack of a distance scale. Probably I would just have to learn to make do with the native autofocus capabilities of the camera, inadequate as they are, and fancier stuff like what I was talking about in this previous entry, simply wouldn’t be possible. On the plus side, since it is a 35mm lens, rather than an 18mm, I’m thinking that would be less of a big deal. The other problem with this lens? Nobody has it in stock!

So, there they are: my current options. Comments appreciated, especially if you happen to have switched from the 18-200 to one of the lenses listed here, or have opinions on third-party telephoto alternatives.

Leave a comment

Filed under brainstorming, Nikon, photography

Coolness of the Day

Check this out: “Liu Bolin is the 36-year old Chinese ‘invisible’ artist. He paints his clothes and body to blend in almost perfectly with his surroundings.” Intriguing! I imagine getting the paint just right on his body and clothing must be quite a challenge, plus I’m thinking he must have to anticipate the final lighting conditions…huh.

Not the sort of thing I would tend to get into myself, but definitely worth a look.

Leave a comment

Filed under links, photography

Quote of the Day

“You should never need to have to pay someone (or be that computer whiz) to keep your computer running perfectly.” – Ken Rockwell

That pretty much sums it up. Computers should just work.

(As for the malware thing, please refer to the concept of “enumerating badness” for a refreshing take on the issue.)

Leave a comment

Filed under annoyances, links

Using a superzoom lens to get around imprecise autofocusing

I noticed a while back that a lot of the landscape shots I was taking had problems with background or foreground objects being out of focus, in spite of shooting at 18mm with a fairly small aperture (f/11, typically). Worse, the effect wasn’t consistent, which made it doubly hard to figure out what was going on. Sometimes a picture shot at f/8 would be just fine, other times I’d have problems even though I was using a smaller aperture than that. It was getting pretty frustrating.

I think I’ve got it figured out now, though, along with a possible solution.

I’m guessing the problem is that, at 18mm, the hyperfocal zone of the lens is so broad and gradual that it’s impossible for the camera to distinguish the sharpest distance from distances which are substantially in front or behind—from the camera’s viewpoint, all of these distances will appear to be equally “in focus,” so the camera will autofocus on a distance that is technically not optimal. This has been borne out by experimentation: When I autofocus on a moderately close object, and then re-autofocus on a moderately distant object, often the lens will not move very much at all, even though it should have, based on comparing the distances of the objects to the distance scale markings on the lens. In the viewfinder there is no visible difference in focus between the two objects, and apparently the camera isn’t able to tell the difference either.

So, if the camera just happens to autofocus the lens in a way where the “most in-focus” object is too far away, then the foreground will end up blurred in the final picture. Likewise, if that object is too close, then there will likely be some softness in the background. In either case, I’m pointing the focus sensor at an object at roughly the correct distance (I hope), but since the depth of field is so wide, the camera isn’t able to precisely center the focal zone on that object, it merely adjusts the lens until the result is “good enough.” Except that it’s not.

(Part of the problem is admittedly that I am using a fairly slow lens—its widest aperture is f/3.5 at 18mm, which is apparently small enough to cause this problem. I wonder how much better things would be if I had an f/2.8 lens? Or an f/1.8 lens?)

Clearly what needs to happen is to get the lens focus set so that the optimal point of the hyperfocal range is resting on an object that’s a little close, but not too close. (Choosing an object that’s exactly halfway between the foreground and the background wouldn’t be the best choice, because at short focal lengths, the hyperfocal range of a lens tends to be asymmetrical, skewed towards the camera.) Since the autofocus mechanism can’t be relied upon to do that by itself, I am left with manually focusing the lens.

But here’s where the second half of the problem crops up: The Nikon D40 is really not at all good for manually focusing on a wide angle lens. In fact, just using the naked eye at 18mm, I would guess I’m even less capable of picking the right setting than the camera is! So how can this problem be solved?

Well, the solution is a bit tedious, but in my initial tests it seems to be working out fairly well. I’ve been using an 18-200mm zoom lens, and while focusing at 18mm is pretty much hit-and-miss, if I zoom the lens out to 100mm or more, it’s quite easy to see if something is in focus or not.  At those focal lengths, the hyperfocal range is much more narrow and easy to spot in the viewfinder. It’s also easy to see if the camera is focusing where I need it to, which means I can leave autofocus turned on for the first part of this.

So here’s what the procedure boils down to:

  • After making sure my exposure settings are good, I’ll start by composing a rough framing of the scene, then I’ll take note of some close elements in the frame, and some distant elements.
  • I’ll then zoom the lens to a telephoto setting, and focus on one of the close elements I noted. Autofocus works fine for this, so that’s what I use.
  • I’ll then take a look at the distance scale on the lens, to see where it is. Typically it’s somewhere around the 3 meter mark, but it varies of course.
  • Then I’ll repeat this zoom-and-focus procedure with one of the distant elements in the original composition, and again check the focus scale on the lens. Typically it will be closer to the infinity mark, but again there is some variability.
  • Then I’ll switch the lens over to manual focus and zoom back to 18mm, or whatever length I was using in the original composition.
  • I’ll then manually rotate the focus ring on the lens to a position moderately closer than halfway between the distance from the nearer focus point to the farther one—again, not choosing the exact halfway point is due to the fact that the in-focus zone on a wide angle lens will be skewed closer to the camera rather than being perfectly symmetrical.
  • Then I’ll reacquire my original composition in the viewfinder, make sure it’s just the way I like it, and take the shot. Hopefully my original exposure value will still be valid—if it’s not, though, the beauty of a manual focus setting is that it stays where it is, if I need to make any further adjustments (as long as I don’t accidentally bump the focus ring!). It’s also perfectly reasonable to “bracket” the focus a little bit, on multiple shots.

The whole procedure is a bit of a pain in the neck, but far less annoying than getting back to the computer only to discover that my background or foreground is fuzzy.

The one big question in all of this is, which aperture is needed? Since Nikon, in
its infinite wisdom, doesn’t bother putting any sort of depth-of-field markings on lenses anymore, I simply resort to guessing (admittedly, it would be better to utilize an actual depth-of-field chart, so maybe I’ll have to get myself one of those). Depth-of-field is typically pretty wide at 18mm, but it’s not infinite by any means, so I generally won’t bother with apertures wider than f/8. I tend to stick around f/10 or f/11 for the most part, although I honestly don’t have enough practice at this yet to definitely have a feel for it.

The other complication is that the aperture choice becomes more critical at higher focal lengths. Sometimes I’ll zoom the lens all the way up to 35mm or so, for “wide angle” landscapes, and obviously increasing the focal length will decrease the depth of field. But, on the other hand, the original problem becomes less of a problem as depth-of-field decreases, so eventually I can just fall back onto regular autofocus as I zoom in on something.

In the long run, I’m thinking a better solution might be to upgrade my camera body.  I’ve heard the D300s has pretty good autofocus—I wonder if D300s owners have these problems? :)


Filed under brainstorming, photography

Who needs a time machine when you have cranes?

I came across an interesting little fact this morning, while surfing the interwebs.

But first, a little context.

A couple of years ago, during an evening hike at Kettle Moraine State Forest, as the sun was getting close to the horizon and the wooded areas of the trail were starting to get distinctly dusky, I heard something really, really bizarre coming from above.  To my ears, it sounded a lot like a pterodactyl flying overhead, although it obviously could not have been one of those. The source of the racket turned out to be two fairly large birds flying overhead, calling out as they flew.

I later realized that the birds were Sandhill Cranes, which I had seen before, but I had never seen them in flight, nor had I heard their call prior to that evening. What an experience it was, too—almost as if, for a moment, I had been transported back in time to when the dinosaurs still roamed the earth. It’s really a challenge to describe how otherworldly a sound it was, although people who have heard that call themselves can probably understand.

Since then I’ve become more familiar with Sandhill Cranes, having seen them quite often at Horicon National Wildlife Refuge, where they are pretty common. They are odd birds, looking distinctly more primitive than many other species, I’ve always thought (with the possible exception of storks).

Well, it turns out that perhaps they really are more primitive, in one sense. This morning I was browsing through the website of the International Crane Foundation, specifically their field guide to the various crane species of the world. On their Sandhill Crane page, they state the following:

A Miocene crane fossil, thought to be about ten million years old, was found in Nebraska and is structurally identical to the modern Sandhill crane, making it the oldest known bird species still surviving!

So Sandhill Cranes have not only been around for at least ten million years, but no other bird species is known to have been extant for that long? That is just cool. Sort of like time travel, almost. :)

Leave a comment

Filed under birds

You know you’re a real photographer when…

…your camera setup is worth more than your car. ;)

However, since my car is pretty much worthless, does that apply to me?

(Seriously, my car would be worth more if I broke it down and sold the parts than if I tried to offload it fully assembled.)

Anyway, I had this thought after several hours of lusting after this lovely beast. :) By the way, I’m pretty sure that’s a D90 it’s mounted on, and I wouldn’t mind one of those either. :)

Now, how much would that lovely thing set me back? At today’s prices, about US$1500 for the lens. A 1.4x teleconverter would also be needed, for another US$450 or so. I’m guessing I’d have to defer purchase of the teleconverter until later. ;)

Obviously I need to think about something else for a while….heheheheh.

Leave a comment

Filed under photography