I recently started using a Canon Digital Rebel XT (350D) for KAP. I am still doing AutoKAP using a CAMremote to control the pan and tilt servos and trigger the camera via infrared.
I have the system programmed to take a shot every 7 seconds. I noticed, however, that I was getting many fewer images than expected, so I tried to figure out why this was occurring. I believe what's happening is that the 350D, when in autofocus mode, is trying to be very precise in focusing, and, when the rig is in motion, is sometimes having a difficult time locking on and therefore skipping shots. This especially occurs in lower light conditions.
On my next outing, before sending the rig aloft, I locked focus on a distant subject and switched to manual focus mode. This did the trick. I got most of the shots I expected,missing a few possibly due to having trouble metering during camera motion.
Yesterday, I again focused on near infinity and put the camera in manual focus mode. I must have touched the focus ring by accident, though, because when I got home I found that all of my photos were a little out of focus.
I've been searching through the camera's manual, and it seems that there is no infinity setting. I'm wondering how others are focusing their dSLRs, or other cameras, if that will be relevant to my situation. If you think that manual focus is the way to go, any suggestions on how to prevent focusing errors would be much appreciated.
If you buy any of the bunch vegetables at the store that come with the ultra-wide, kinda fat rubber band, these do a great job of locking the focus rings on Canon lenses. If the ring doesn't move when you're in auto-focus, you can slip the rubber band over the gap between the focus ring and the lens barrel, locking it in place. Then switch to AF and focus on a distant object, then flip back to manual, just like you did. The rubber band should keep focus from slipping. I've done this when doing astrophotography, and it works pretty well.
On lenses where the focus ring moves when they're in AF mode (a few of the older Canon lenses do), you have to roll the rubber band out of the way first, and VERY carefully roll it back once you've set the focus. More tedious, but it still works.
Michael, what shutter speeds are you using on the Canon Digital Rebel XT (350D)?
If less than 1/500 of a second this could contribute to less than sharp images.
I agree with Tom's comment. Use the rubber band trick to lock the lens at infinity. If you are still getting soft images try to shoot at higher shutter speeds.
I have been flying a Canon XTi (400D) using a radio-controlled rig for well over a year now and have logged 59 KAP sessions with it.
It is my general sense that the camera takes a longer to focus when the rig is moving and I am aware that, on occasion, I do not get a shot that I remember taking. I only activate the shutter relay for 1-2 seconds and the camera probably fails to achieve focus on these occasions. It really chaps my hide when this causes one image in a panorama sequence to go missing.
One of the advantages to using radio control for the shutter is that I can time my shots for when the rig is relatively still. And, I have learned to wait a couple of seconds for the rig to calm down after a pan rotation before shooting. While higher shutter speeds are always welcome I think my Canon Rebel rig does quite well with exposures slower than 1/500 second. If the winds are calm I get useful shots below 1/100.
Regarding focus, I leave the camera in autofocus with all focus points enabled and this seems to work for me. Every now and then I get to play with a foreground object.
Thanks for your replies, Tom, Jim, Cris, and Brooks. Excellent suggestions.
I will probably try Cris' autofocus option first, since it is the easiest to implement. Until now I have been only using the center focus point, so I'm hopeful that using all focus points will improve matters. BTW, I have my CAMremote programmed to wait 4 seconds after servo movements before taking a shot. Increasing the interval might help also.
I had already been using Tom's rubber band trick, but only to keep the zoom at its widest setting, not to fix the focus ring. Maybe a 2 rubber band system is called for?
Very useful to know the optimum apertures for digital cameras, Brooks. I've been shooting mostly between f/4 and f/5.6 unless there's very low light, so it's good to know I'm right in the sweet spot.
If you switch off autofocus there's no need to set the distance to infinity. What we're aiming for here is the Depth of Field (DOF). This is the distance in front of the cam which is sharp. I'm using a 20mm lens, if I set the distance at 5 metres, the sharp area is from 1,5m till infinity.
About Aperture: every lense has the best optical setting if you 'close' the A 2 stops. So if the lense has a f2.8 value, set it at 6-7. More than f10 has no effect on quality.
ps: on a Nikon you can switch the lens to manual without any rubber stuff ;-)
Brooks, that blog entry is a good 'un. And it's just as true for film as it is for digital. One other data point to keep in mind, though: At the wide end, optics tend to go a little soft unless they're really REALLY well made. When lenses are ground, the edges of the glass tend to have more issues than the center of the glass. So if you use a small(ish) aperture, you're only using the center-most part of the glass, and are less prone to aberrations. Open up, and more of the glass is used and aberrations present in the outer edges of the glass start to dominate. An unfortunate case in point is Canon's 50mm f/1 lens. It's actually really well corrected out to about f/2, but if you open it up to f/1, it goes soft. Which is a pity, since its wide aperture is its real claim to fame.
What this amounts to is that every lens/camera combination has a sweet spot where you strike a balance between the aberrations in the glass at the wide end and diffraction off the iris at the narrow end. You can find this by focusing on a target and running a series of exposures at every aperture you can. Compare them, and pick the one that has the sharpest focus at the point of best focus. Because not every lens has the same glass characteristics, even in a given lens model, it's worth doing this with each lens. The sweet spots will typically vary a little, lens to lens. For my 100mm macro on my Canon 20D, it's about f/16. For my 50mm f/1.4 it's f/11. For my 150mm on my 4x5 camera it's f/22. I haven't really tested my KAP camera, but given this whole discussion, I really should!!
One nice point about all this, though: In discussions of lens focal lengths on digital cameras, we don't tend to quote the actual focal length of the camera, but rather the 35mm equivalent that would give the same field of view. This is great when calculating field of view, but can be very misleading when looking at depth of field. Here's a case in point:
Take a 35mm camera with a 35mm lens on it, opened to f/5. Your DOF when focused at 10 feet should approximately be 8.6 feet (7.3 to 16.0).
Now take my A650. The 35mm equivalent is 7.4mm of actual focal length. Opened to f/5 and focused at 10 feet, I should get acceptable focus from 3.7 feet out to infinity. With the smaller detector size, you can use a wider aperture to get similar depth of field.
I'd still take all this with a grain of salt. Or a grain of film, anyway. What constitutes "acceptable focus" depends a lot on what you're using to measure it. With film, it was what you could perceive in your film grain. I'd argue that someone shooting Tech Pan 25 or Kodachrome 64 would be able to make the point that their definition of "acceptable focus" is a little different from someone shooting TMax 3200. With digital detectors, I'd argue that the smaller pixels on the compact digital cameras compared to the larger ones on DSLRs probably offset the shorter focal lengths to some extent, and that a 1:1 comparison probably isn't fair or even accurate.
The nice thing with cameras is that it all really boils down to what the photographer determines to be good. If in doubt, run tests of your own. Find out what depth of field you really want or need. Find out what the sweet spot in your optics really is. And by all means keep taking pictures!
As an old teacher of photography, my students would frequently ask me for the formula for taking great pictures. My reply was to set the camera at F8 and "be there".
I experienced similar problems with DSLR and manual infinity focus, and it depends on the lens, even among AF lenses. On the nikkor 10.5 fisheye, no problem, the focus ring end point is infinity, you only have to maintain it with a piece of tape. On the nikkor 18-55 kit lens, and on the sigma 10-20, the focus ring end point is just a little bit BEYOND infinity (!). That is if you place the ring here, everything is out of focus on the picture (tested). I imagine this is a side effect caused by constraints on AF design. And probably the same on cheap Canon lenses ? This means you have to make precise manual adjustments before putting the tape or rubber band. Not very handy in the field ... at all.
emiel's point on DOF and hyperfocal distance is a good one : it is probably better to focus a little bit before infinity, rather than trying to find the exact position for it (with the above-mentioned risk).
Putting AF back on is also the solution I use today, but I am not very satisfied with it: as for Cris this is a source of missed frames.
I think the ability to go past infinite focus is a feature, rather than a bug. (Though I curse at it fluently and often when it messes me up.) A friend who does astrophotography said it's there for a couple of reasons:
1 - Thermal: Especially with plastic-bodied lenses, when they get cold they shrink. Quite a bit, actually. It's enough to skew focus, especially if you're using it wide open (as you would doing astrophotography). So by giving the lens a little more range, even if the lens barrel contracts from thermal effects, you can still reach infinity.
2 - Variability in Manufacture - No two lenses are the same. No two pieces of glass are ever the same. So when a camera lens is assembled, the lens mounts and barrel assembly are designed to give each element in the lens a fair bit of "wiggle room". The individual spacings can be adjusted on the bench to give the best optical quality as possible out of that lens design. But this also means that no two lens assemblies will be identical, so the values on the focus ring, and even the physical stop at either end, have to be taken with a grain of salt. Rather than assemble a lens that can't reach infinite focus at all, manufacturers err on the side of caution and put the physical stop at the far end of focus a little past where it should theoretically fall.
Which goes back to manually focusing on the ground and locking focus before sending the camera aloft.
What we really want to have (for max sharpness) is the hyperfocal distance. If you focus on this distance, everything from half that setting to infinity is sharp. Read all about it here: http://www.dofmaster.com/hyperfocal.html
Bottom-line is that you never want to focus on infinity if you'll need sharpness on the maximum DOF.
I have spent the last several weeks testing many different settings on my Canon 350D. I tried Cris' suggestion of Autofocus with all focus point enabled, and increasing the waiting time after pan servo motions, and this improved things, but not enough to satisfy me. After eating some delicious broccoli, I went back to manual focus using a broccoli band, as Tom suggested. This resulted in many fewer skipped shots. However, one day I again (slightly) missed proper manual focus during a KAP session, so I really wanted to make autofocus work somehow. I went to my 350D manual and looked up the focus options. I found that Canon recommends AI Servo autofocus mode for moving subjects. Switching the camera to AI Servo made a big difference in the cameras ability to focus during camera movement. I first tested this on the ground, moving the camera around while successfully snapping shots. I've had a couple of KAP sessions since, and it seems to skip many fewer shots in the air as well. I have also switched the metering mode from evaluative to center-weighted average, and believe that this might be helping somewhat too.
Just wanted to report back in hopes that this info might benefit others using Canon dSLRs, and possibly other brands too.
I recently tested a new Canon XSi for KAP. I'm happy to report that the camera seems to have significantly better autofocus than the Rebel XT, taking virtually all images under the AI servo autofocus setting. This seems to hold true even when the camera is moving quite a bit, and also under low light conditions.
I was testing one of those in the store and was impressed with the auto focus performance. It seemed much faster than my Nikon D40x. The D40X will skip from time to time due to focusing issues as well.
This old thread might help me out. I want to take some shots of a golf course, I am using a Canon A560, normally I just go out there and set it on auto and use wind watchers script to let it go, but I would really like to improve my game using the existing equipment. Does anyone have any suggestions to set the camera using the manual settings that may improve my shots?
This Luminous Landscape article explains in a bit more detail why hyperfocal distance calculations are far from being an exact science. You can't just ignore the circle of confusion issue if you are intersted in making large prints. It is really a matter of personal taste though.
The following piece of advice from the above article is very useful:
"......... I always use the lens' DOF setting for one stop smaller than indicated. This ensures that even if I make a very large print, I'll still have adequate DOF. In essence what I'm doing is doubling the effective COF."
Perfect lens (such thing does not exist, but, for a sake of simplicity, let's assume we've got one) when focused on a point (again non existent 0-dimensional theoretical thing) projects that point to a point on projections surface (film, sensor, etc). Now add another point to that scene, but now at different distance from the lens. This point wouldn't be focused perfectly of course, so on the projection surface it will be imaged as a circle, and diameter of that circle depends how much out of focus that other point is. But our eyes have finite resolution, our photographs have finite size, etc, thus if the circle is small enough it is indistinguishable from perfectly focused point on a real produced (printed or displayed) photographical image. The maksimum circle indistinguishable from a perfect point is "circle of confusion" (COF).
1. For standard 35mm film COF is usually defined as 0.035mm or 0.03mm for general use (5x7" enlargements, standard quality prints), and 0.02mm for professional use (8x10" or bigger[*] enlargements, fine quality prints, slides, etc). For APS-C cameras (majority of KAPable DSLR's except 4/3 system) COF is 0.02mm for general use and 0.015mm for pro use. In case of 4/3 cameras, those are 0.015mm for general use and 0.1mm for pro use. And for typical 1/2.5 sensor cameras KAPers most frequently use, COF is 0.006-0.007mm. And here we rather do not care for large enlargements, those cameras generally image to poor to allow smaller COF to make a difference. On such small cameras with high quality lens (I don't know if even single such camera exists, as it requires lens with 250 line pairs per mm resolution, and such resolution occurs only in the better of prime (non zoom) lenses) it'd be around 0.004mm i.e. 4microns.
2. There is also limitation on how small diaphragm (iris) could be set - due to diffraction limits (light diffracts while passing through a hole, and from smaller holes diffraction is stronger). For Standard 35mm general use limit is f/22-f/27, for high enlargements (pro) it's generally f/16. For APS-C it's f/14-f/18 for general use and f/11 for quality use. For 4/3 system it's f/11-f/14 for general use, and f/8 for quality use. For small 1/2.5" cameras it's f/5.6 for general use and f/4.0 might make a difference on cameras with high quality lens(if it's at all available, as in case of some zooms at long end it's not available at all).
So. If some scene required f/11 to get appropriate DOF, it will need f/8 on APS-C, f/5.6 on 4/3 and f/2.8 on 1/2.5" point'n'shoot. But if someone wanted to get shallow DOF and nice bokeh available on 35mm at f/4 or f/2.8 one would need f/2.5 or f/1.8 on APS-C, f/2 or f/1.4 on 4/3 but on 1/2.5" point'n'shoot it's impossible to attain (as f/0.8 or f/0.56 would bee needed - both impossible[**])
[*] Enlargements bigger than 8x10" typically do not require smaller COF, as they are viewed from somewhat larger distance, not from 10". If there are special cases when even higher resolution is needed then COF could be reduced accordingly -- some high quality lens are capable of rendering images with over 300 line pairs/mm i.e. COF 10 smaller than normal. [**] While theoretically it's possible to make f/0.5 lens and fastest (biggest aperture) usable lens ever made was f/0.55 (which looks like a ball of glass and could not be fit to anything resembling normal camera, this was made for some special purposes), and Zeiss f/0.7 50mm lens could be fitted to a camera and was even used while making some movie (AFAIR one of Stalney Kurbick's) where pictures were taken in a dark Scottish castle room lit only by a few candles, one should not count that such type of lens would be fit onto some small compact camera. f/1.8 is a maximum any could even hope, and in real life existant cameras it's rather 1/2.5 or 1/2.6 and only and wide end of a zoom.
> "Enlargements bigger than 8x10" typically do not require smaller COF, as they are viewed from somewhat larger distance, not from 10"."
Nice summary Sebastian. I don't think the above quote is necessarily true for KAP though. I have a small exhibition of KAP images here and notice that almost everyone walks in to look as close as they can to pick out the detail. Sometimes they even take them off the wall. Hence my interest in the subject.
Brooks, I think you understand that I'm really asking why it isn't COC. I don't think the question has been answered but I do have a better understanding of the concept of "circle of confusion". Thanks, all.
That 0.02mm COF has one more important aspect. 0.02mm is a focusing precision autofocus sensors of the kind used in 35mm SLR's are supposed to focus with. I.E. if you use auto-focus don't expect the point you focused on will be sharper than that 0.02mm (50 line pairs per mm). It might be, but is is not required nor enforced. So If you want "crispiper" images use manual focus, or ensure other conditions make it so (for example close down your aperture by 1 full step -- as SLRs focus with lens wide open, with one full step aperture closedown, it will reduce COF at the focus point be a factor of 2 (more-or-less), given that lens has good enough resolution).
With full frame DSLR such resolution like 0.01mm COF is attainable given good quality lens (most plastic zoom lens sold as "kit" lens with cameras are not, but who'd attach such el cheapo lens to full frame DSLR?). On 35mm film camera, besides good lens (here el cheapo kit lens problem really applies) one also needs a good, fine grain film (Fuji slide films will do (except ISO 200 Sensia, strangely ISO 400 Sensia and Provia will do) , Ilford and Kodak ISO 100 B&W will do as well, all ISO 50 and slower films will do too, Fuji ISO 100 colour negatives should do as well). On APS-C one needs really good lens (only some best profi-zooms or simply prime lens). On 4/3 pretty much prime lens is required and not a poor one. And remember that little point'n'shoots with 1/2.5" (1cm) sensor are simply not capable of imaging with such resolution at all.
Some equations for you to get your teeth into in the above site Broox!
Sebastian - I actually have a KAP outfit that follows your advice. I have kept the lightest variants of Pentax 24, 28 and 50 mm manual focus lenses. Fortunately sometimes the slower, lighter lenses were the better quality lenses. I 'use' these (not very often) on a light Ricoh body with a motorized winder. I can set the focus where I wish. The film choice is exactly as you suggest, Velvia 50 or more often Provia 400.
These lenses all work on the Pentax DSLR bodies (APS-C), but unfortunately they then become 36, 42 and 75 mm. So the 24 mm which is also very small and light is most used.
In practice I tend to use this more for pole photography. My carbon pole is set up for a Ricoh GRDII at 10 m, for the above mentioned setup at about 8 m and a full frame DSLR or light medium format camera at about 5 m.
Thanks, Tony, for that lead. I had seen the Free Dictionary before but forgotten about it. Now I have its toolbar! Anything worth doing is worth overdoing, as my son often reminds me. Now back to the subject.....
Back to subject... One more thing for manual focus setting at hyperfocal distance -- even if someone wants to use "amateur" COF/COC setting one should still know, that what seems to be OK for close subjects is could look too soft for distnant or very detailed subjects (i.e. many of KAP subjects). This is a psychosensory effect, that people want to get as much sharpness for distant subjects (people task their vision more and when looking at distant subjects and expect similar performance also on photographed images), so while non razor-sharp image of close thing is acceptable (ore even goes unnoticed) it gets noticeable when image is full of small distant subjects. So use smaller COF/COC that you typically accept in grounded photography for your hyperfocal distance calculation when KAPing.
Now wrt time setting. If you want avoid blur due to camera movement (primary blur cause in is camera rotation) that camera during an exposure should not rotate more than about 1/2000 of image width (for horizontal images, use height in case of vertical cadres) for "non demanding amateur" quality and 1/3000 for "pro" quality. Thus if you use moderately wide (35mm on 35mm film equivalent) lens and want sharp image, your camera must not rotate more than about 1' (one arc minute) for pro quality or 1.5' for "non demanding amateur". If you have rotation speed of 10 degree per second (thus 36s for full revolution) you want about 1/400s exposure time for "non demanding amateur" and about 1/600 for pro. For 1 rpm those times could be doubled. When zooming in to moderate tele (70mm equivalent) use times twice as short.
And that Wikipedia link from 2 posts above (by Tony) is pretty good and full of useful formulas.
