Hand Held Landscape Question

[dwdallam]

If you hand held a landscape shot at say 1/200th, would it be sharp at distance?

I realize that a 1 degree move at the subject can equal hundreds of yards at a distant object, for example, aiming at the moon with a .00009 discrepancy could mean you miss the moon by hundreds or thousands of miles. I'm wondering if this would translate into the distant object moving at a high enough speed to blur it at higher, usually considered fast enough shutter speeds to stop action, such as 1/500th down to about 1/200th?

[BernardLanguillier]

If you hand held a landscape shot at say 1/200th, would it be sharp at distance?

I realize that a 1 degree move at the subject can equal hundreds of yards at a distant object, for example, aiming at the moon with a .00009 discrepancy could mean you miss the moon by hundreds or thousands of miles. I'm wondering if this would translate into the distant object moving at a high enough speed to blur it at higher, usually considered fast enough shutter speeds to stop action, such as 1/500th down to about 1/200th?

I guess that the impression of bluriness is probably indeed is a matter of characteristic angular lenght of the object vs camera angular movement.

An apple photographed 3 meters away with a slight blur will still look like an apple, while the windows of a building one km away will lose the appearance of windows...

Cheers,
Bernard

[Dick Roadnight]

If you hand held a landscape shot at say 1/200th, would it be sharp at distance?

The general rule used to be that the minimum shutter speed for hand-holding was one over the focal length = 1/125th of a sec for a 125 mm lens: with high res film or smaller pixels it gets more critical, but the rule held for all formats, so, with a larger format with longer lenses for the same angle of view, you have to use faster shutter speeds or a tripod.

You could read Harold Merklinger's "The ins and outs of focus" and work out what size of subject detail you can expect to resolve at what distance for your camera and lens (and confirm by experiment) ... and experiment to find out how much camera shake you need to allow for (in degrees) and work out (trigonometry) at what shutter speed the camera shake is likely to be the limiting factor.  

If you make the normal mistake of using wide angle lenses for landscapes, the rule would indicate that you should be OK at 1/200th if you are using anything smaller than 10*8".

... but I have a 640 mm Novoflex non-tele lens with two pistol grips and a shoulder support, and I could consistently hand-hold it at 1/125th (especially if I had my elbows on the ground), and I find that my Hasselblad will not produce sharp pictures @ 110 mm unless it is on a tripod with the mirror locked up!

I realize that a 1 degree move at the subject can equal hundreds of yards at a distant object, for example, aiming at the moon with a .00009 discrepancy could mean you miss the moon by hundreds or thousands of miles. I'm wondering if this would translate into the distant object moving at a high enough speed to blur it at higher, usually considered fast enough shutter speeds to stop action, such as 1/500th down to about 1/200th?

If an aeroplane is moving at 500 mph,  and you want it to look sharp so that you can see the windows, the blur (in feet or m at the subject) is not affected by the distance. If the plane is just a dot in the distance, you do not need to see the windows!

The beauty of digital is instant feedback, so you can suck-it-and-see, immediately, if you used a fast enough speed - the angle of movement is relevant, of course, and you might be able to pan: find a busy road, railway or airport and experiment.

If you are into photographing steam trains, and your subject only comes past once a decade, you need to think about it before hand and you might turn up the ISO if a cloud came past at the wrong moment... and you need to have some idea how fast the train will be traveling - steam trains pulling up hills or pulling out of stations move slower, and make more steam.

Of course, trains are a good argument for tilt/yaw, as you can position the plane of sharpest focus where the train is going to be (assume it will stay on the track!) and use a larger aperture and a faster shutter speed without increasing the ISO. (See Merklinger's "Focusing the View Camera".)

If you want a fast, distant subject sharp, and the foreground also sharp, you can take two exposures and combine them in PhotoShop.

[petermarrek]

I shoot a lot of images from a small boat, making a tripod almost useless, a monopod a bit better but not really friendly, so handheld is usually the preferred option. I feel that even though I miss the occasional shot to unsharpness, shooting a short burst will usually provide me with at least one useable frame, if not I will go back later if it was worthwhile and re-shoot. It's a huge benefit of digital that there is no film cost. I generally dislike this approach, but it does work. Peter

[petermarrek]

I shoot a lot of images from a small boat, making a tripod almost useless, a monopod a bit better but not really friendly, so handheld is usually the preferred option. I feel that even though I miss the occasional shot to unsharpness, shooting a short burst will usually provide me with at least one useable frame, if not I will go back later if it was worthwhile and re-shoot. It's a huge benefit of digital that there is no film cost. I generally dislike this approach, but it does work. Peter

[Bobtrips]

Perhaps someone more knowledgeable than I can wade in here....

I seem to recall that the 1/focal length rule works only for 35mm (full frame) by 'accident'.  If one want to use that guideline as a way to choose shutter speed then they need to convert to focal length equivalent.

I suspect the proof of what I'm half remembering is that one isn't able to get shake-free shots at 1/7th second on a small sensor compact with a (non-IS) 7mm lens.

[dalethorn]

I suspect the proof of what I'm half remembering is that one isn't able to get shake-free shots at 1/7th second on a small sensor compact with a (non-IS) 7mm lens.

That's for sure. With the Pana LX3, which is stabilized, I might take 10 shots at 1/8 second and get one that's sharp enough. Without IS, I can't imagine, unless you can brace the camera against something.

[situgrrl]

The rule served me well when I was learning with a 35mm SLR but I think it's pretty obsolete WRT digital.  With my E1 - I could hand hold approx 1 speed below 1/f, with my 30D, I struggled to hold anything below 1/60th except with my 10-20.  With my Leica 1/8 is easy and 1 sec just about possible.  With an Ixus thingy, 1/30 was impossible even with IS.

The Ixus, I think, was too small to be stable.  The E1 was possibly a mix of low pixel count and well damped mirror.  the 30D - higher pixel count perhaps?  Maybe it's down to ergonomics?  Then why can I hand hold the german brick with no grip?

I used to shoot a 1 series film at about 1/30 with an 85mm with approx 50/50 hit rate (maybe less) - but my subjects were close to me (live bands) and I was shooting film rather than 25 mpix.  Monopod?

[Dick Roadnight]

The rule served me well when I was learning with a 35mm SLR but I think it's pretty obsolete WRT digital.  With my E1 - I could hand hold approx 1 speed below 1/f, with my 30D, I struggled to hold anything below 1/60th except with my 10-20.  With my Leica 1/8 is easy and 1 sec just about possible.  With an Ixus thingy, 1/30 was impossible even with IS.

Mirror shake would not affect the Leica.

With a larger format (diagonal) you want proportionally more res, so the same shutter speed is appropriate for the same focal length, regardless of format.

For longer lenses doubling the focal length approximately halves the seen angle, so you need half the exposure time to have the angular movement (shake) during the exposure, giving the same blur as measured on the film or sensor, all things being equal.

For shorter focal lengths viewing angle does not change in proportion to focal length, so the rule ceases to apply.

[dwdallam]

What I'm talking about is a non moving subject shot at a mile or more. This is just to clarify what I mean.

Let's say you are shooting a laser at the moon, but first at a beer can 100 yards from you. Since the beer can is 3" wide, you can have an angle from the shooting position that equals 3" or less when it gets to the can and still hit the can. But, on the other hand, if you are shooting the moon, that same angle may become 1, 000s of miles off target because of geometry.

OK, this means that out at the moon distance, that laser may be moving from point to point, that is, covering 1000s of miles in the time it takes you to flinch using the shutter button. In other words, again, you can move the laser from the moon 180 degrees to the other side of our solar system, say Pluto, in a milisecond. That would be 100s of thousand of miles per hour. That is, it would be like your subject moving that fast.

One other example: If you have a flat earth and used the same laser, and it was perfect level with the flat ground, it would not climb or drop. If you move the laser away from it's perfectly level position "up" .00000000000000000000000000000000000000000001 degree given enough distance, it would soon be 1000s of feet in the air at some distant point, while remaining at ground level at its point of origin. Simple geometry.

Now back on earth, your three mile horizon line is the same deal, although since it's not as far as the moon, will "move" much slower.

Another way to put this is that every movement of the camera is multiplied the further away the subject. The rule about how fast the shutter needs to be according to how long the lens is will not, at some point, compensate for that distant movement--or will it? And if so, how far and what shutter speed will be a good rule of thumb to get acceptably sharp landscapes while hand holding given a specific distance to the horizon subjects (mountains, trees, etc.)? That's my question.

[dalethorn]

What I'm talking about is a non moving subject shot at a mile or more. This is just to clarify what I mean.
Let's say you are shooting a laser at the moon, but first at a beer can 100 yards from you. Since the beer can is 3" wide, you can have an angle from the shooting position that equals 3" or less when it gets to the can and still hit the can. But, on the other hand, if you are shooting the moon, that same angle may become 1, 000s of miles off target because of geometry.
OK, this means that out at the moon distance, that laser may be moving from point to point, that is, covering 1000s of miles in the time it takes you to flinch using the shutter button. In other words, again, you can move the laser from the moon 180 degrees to the other side of our solar system, say Pluto, in a milisecond. That would be 100s of thousand of miles per hour. That is, it would be like your subject moving that fast.
One other example: If you have a flat earth and used the same laser, and it was perfect level with the flat ground, it would not climb or drop. If you move the laser away from it's perfectly level position "up" .00000000000000000000000000000000000000000001 degree given enough distance, it would soon be 1000s of feet in the air at some distant point, while remaining at ground level at its point of origin. Simple geometry.
Now back on earth, your three mile horizon line is the same deal, although since it's not as far as the moon, will "move" much slower.
Another way to put this is that every movement of the camera is multiplied the further away the subject. The rule about how fast the shutter needs to be according to how long the lens is will not, at some point, compensate for that distant movement--or will it? And if so, how far and what shutter speed will be a good rule of thumb to get acceptably sharp landscapes while hand holding given a specific distance to the horizon subjects (mountains, trees, etc.)? That's my question.

Ignoring aesthetics or variations in lighting, it shouldn't matter, should it?  I mean, your camera is not so much moving through an image field in the distance as it is moving through its own pixel-image of the actual scene.  So if your image is 6,000 pixels wide and your horizontal shake-movement is 'x' pixels per second at the center of the movement (as opposed to the ends where movement stops and reverses), then you can calculate from there.

[dwdallam]

Ignoring aesthetics or variations in lighting, it shouldn't matter, should it?  I mean, your camera is not so much moving through an image field in the distance as it is moving through its own pixel-image of the actual scene.  So if your image is 6,000 pixels wide and your horizontal shake-movement is 'x' pixels per second at the center of the movement (as opposed to the ends where movement stops and reverses), then you can calculate from there.

I don't know. About the only thing I'm clear on is that even though the point of origin moves say .01 millimeter, that movement is compounded the further away it gets from that point of origin. It's like taking a long fishing pole and barely shaking the held end, but the tip is moving two feet each direction.

[Dick Roadnight]

Another way to put this is that every movement of the camera is multiplied the further away the subject.

This is not the case: the camera can move in six degrees of freedom, and only pan and tilt get magnified with distance, and these movements are angular (degrees) not millimeters. The effect of rotation increases with distance from the optical system axis.

[cmi]

In other words, again, you can move the laser from the moon 180 degrees to the other side of our solar system, say Pluto, in a milisecond. That would be 100s of thousand of miles per hour. That is, it would be like your subject moving that fast.

I understand what you are getting at. I will try to explain it very simple.

First lets discern between different kind of movements: Non angular (strictly parallel to the sensor) vs nodal angular (around any three axes).
To make it understandable lets discern between orthogonal perspective (not possible with camera) and our normal central perspective we are used to.

For this, suppose a wide, say 90 degree angle, sharp from 30cm - infinity with a Din A4 sized stone at 35cm (quite big) and detailed background at 4km.

Now, for parallel camera movement and orthogonal perspective, imagine how a single ray coming out of camera from would behave in parallel shake. In THIS case it is easy to see that distance traveled does NOT amplify in the far regardless how the ray is pointing. I move here 1 cm, the line coming from my finger moves parallel also 1cm. So in the theoretical case of parallel shake and orthogonal perspective, no matter how far an object is, blur due to camera shake is not dependend at all from distance.

In the case of parallel camera movement and central perspective this is different. What comes into play is distorsion caused by perspective. At a distance of 2 kilometres you got to see at a much wider distance all at once, lets say the view is 4 kilometres wide. But at the near, e.g. 35cm distance, the camera has a view that covers only e.g. only 50 cm. So if I move my camera parallel, as we all know, the foreground moves faster than the background (parallax).

So it is clear: Blur caused by parallel camera shake gets actually STRONGER the nearer an object is, because of the parallax effect. But, in practice this plays no role because if a movement ruins IQ, then its meaningless to ask if these where are variations caused by parallax. Either the picture is usable or not.

Now the case of rotational shake around a nodal point. This is simple. As we all know, we will get no parallax here so the scene will remains static as you freely look around your nodal point. So indeed a ray as it goes deeper into the picture, will gradually cover more and more distance as it shakes, but the increasing distance is of no consequence. Only the pixel-distance in the final image is important, and that depends on you. As long as you rotate around nodal, movements in the far and the near are the same speed.

So in theory there is indeed a factor wich introduces varying sharpness for camera shakes, and this is parallax. But in practice this is not important, at least as far as my experience goes.

Regards,

Christian (who spent way to much time writing this)

[BJL]

Perhaps someone more knowledgeable than I can wade in here....

I seem to recall that the 1/focal length rule works only for 35mm (full frame) by 'accident'.  If one want to use that guideline as a way to choose shutter speed then they need to convert to focal length equivalent.

Or at least someone more opinionated ...

I Agree: the shutter speed needed to reduce camera motion effects to a given level clearly depends on the angular field of view, not the actual focal length, so 1/(fake focal length) is better. It also depends on the desired display size of the final image (which the fake focal length takes care of so long as intended display size stays the same). Probably 1/f would not be safe with a 24MP 35mm format DSLR and the intent to print large ... unless you are unusually steady, but in that case 1/f is wrong in the opposite direction for 35mm film and more modest print size plans.

For the modern pursuit of "pixel level sharpness", the shutter speed needed varies I think about in proportion to the ratio of focal length to pixel spacing, or to be a bit fancier, in proportion to the desired angular resolution. There are other factors of course, like personal steadiness of hand and the moment of inertia of the camera (its resistance to twisting forces), mirror motions, shutter vibrations.

Maybe a useful guideline is
1. Experiment to find your minimum acceptable shutter speed with a particular camera and one reference focal length: 100mm would be convenient.
2. For other focal lengths, adjust the minimum shutter speed in proportion to focal length. (E.g. 300mm -> triple the speed; that is why a 100mm reference is convenient.)
3. Try to stay above about 1/50s anyway, to avoid additional complications with mirror/shutter effects.


And anyway, IS throws a lot of this out the window!

[dwdallam]

I'm still not getting my head around it. I can't make sense of this:

Using a laser beam again, and the laser beam is the camera, let's say you have it on a tripod and the tripod is outfitted with a panoramic device so it moves perfectly horizontal as you move it from left to right (rotate it right to left). Let's say that the motion is 1cm per hour. Now you have two subjects. One is 10 feet from the camera (laser) and one is 20 miles. I don't know exactly what the calculations would be in reality, but it would be something like this:

Shooting the closer subject would move 1cm across its surface in one hour. The laser point hitting the subject at 20 miles, and again I don't know how to calculate this, might move 500 yards in the same time. So the laser (camera) at its point of contact is moving so slowly at the near subject you could get a sharp picture at 1/100th or even much slower; the speed of the laser at its point of contact at 20 miles would be moving much faster, proven by how much space it travels in a given time (1cm an hour for the near subject and 500 meters in the same time for the far object).

This leads me to believe that the far objects are going to need a higher shutter speed to remain sharp. In fact, if you move the camera simply one degree in 1/2 a second (or much faster) at 20 miles it might be moving so fast that you would need a shutter speed of 1/8000 to stop it sharply. I don't know how to do the calculations though.

[agavephoto]

I'm still not getting my head around it. I can't make sense of this:

Using a laser beam again, and the laser beam is the camera, let's say you have it on a tripod and the tripod is outfitted with a panoramic device so it moves perfectly horizontal as you move it from left to right (rotate it right to left). Let's say that the motion is 1cm per hour. Now you have two subjects. One is 10 feet from the camera (laser) and one is 20 miles. I don't know exactly what the calculations would be in reality, but it would be something like this:

Shooting the closer subject would move 1cm across its surface in one hour. The laser point hitting the subject at 20 miles, and again I don't know how to calculate this, might move 500 yards in the same time. So the laser (camera) at its point of contact is moving so slowly at the near subject you could get a sharp picture at 1/100th or even much slower; the speed of the laser at its point of contact at 20 miles would be moving much faster, proven by how much space it travels in a given time (1cm an hour for the near subject and 500 meters in the same time for the far object).

This leads me to believe that the far objects are going to need a higher shutter speed to remain sharp. In fact, if you move the camera simply one degree in 1/2 a second (or much faster) at 20 miles it might be moving so fast that you would need a shutter speed of 1/8000 to stop it sharply. I don't know how to do the calculations though.

Maybe I can help some. In your example, you have a panoramic device, and you say it's moving horizontally; this means the laser is moving to, say, the left at 1cm/hr. Here, the beam moves to the left at the same pace, no matter how far away one is . . . because it's moving in one direction, and NOT rotating. Rotation is what would cause the greater speed at greater distances.

Allow me to add a couple of examples to put this into perspective.

Example 1: Imagine a CD spinning inside a CD player.  Draw a straight line on the CD from the center point outward to the edge (this line would cut the cd perfectly in half if extended on both sides). The CD is spinning at some rate (the exact RMP doesn't matte). Since the CD is a solid object, the whole disc rotates as a whole: meaning that it takes the same time for the edge of the CD to make one full rotation as it does for the inner part of the CD near the center. Now, look at the physical distance a point on the line must travel to make that rotation. A point near the outer part makes a larger circle than does a point near the inner edge of the CD (2*pi*r shows exactly how much larger as a function of the offset from the center of rotation). Now, since both points take the same TIME to make the rotation, the outer point MUST move faster through space than the inner point if it's going to cover the larger distance in the same time. Extend the line on the CD out to a far away object, and this would appear to make a very fast speed.

Example 2: Take a pencil and lay it on a table. Push the pencil in any direction from the side (perpendicular to the long axis, just don't rotate it!), and one should find that a points on opposite ends are moving through space at the same speed. Extend the pencil to some far away object, and it should be moving at the same speed.

Replace the CD and pencil with your camera, and the same should be true of a line from your camera out through the center of the field of view. NOW, you've likely seen this last one in principle yourself: from your car! Driving down a road, one sees the side of the road appear to move much faster than a distant mountain beyond the side of the road. As proof, one can simply take a photograph out the window of a moving car with a shutter speed not up in the 1/1000ths or higher. One should find that the farther away the object, the sharper it looks. Stop the car, get out and take another photograph of the same subject, but this time rotate the camera: the results should not look the same.

Added in the edit: One shouldn't need 1/8000 to get a landscape shot in focus hand held. As an example, I have linked to a photo I took through the window of a moving car: http://agavephoto.com/_MG_0010.jpg   Here, the moon and volcanoes are sharp and in focus, but the plants near the lower edge are not nearly as sharp. Exposure was 1/640 (and only that fast to get more foreground in better focus!).

Do these examples help? I can certainly try to explain them more, or differently, if they are confusing.

[dwdallam]

Maybe I can help some. In your example, you have a panoramic device, and you say it's moving horizontally; this means the laser is moving to, say, the left at 1cm/hr. Here, the beam moves to the left at the same pace, no matter how far away one is . . . because it's moving in one direction, and NOT rotating. Rotation is what would cause the greater speed at greater distances.

Allow me to add a couple of examples to put this into perspective.
Do these examples help? I can certainly try to explain them more, or differently, if they are confusing.

I'm talking about rotation.

So are you saying that given rotation the further out you go the faster the objects in the distance will move past the lens? If so, then sure, I understand that. I was wondering if I were correct in that assumption.

What I'm asking, if the above is true, is that when you hand hold your camera, if you rotate the camera not intentionally, but from releasing the shutter and just being a human, will that rotation be enough to blur a very distant background, and if so, how fast would the shutter need to be to get a sharp image, you know, a rule of thumb given average possible rotation hand holding for an average hand.

[Jack Flesher]

I think the answer is it depends... Your pixel's capture diameter subtends a capture "arc" based on the focal length of the lens mounted at the time of exposure.  So our system is only concerned with the percentage of effect the motion imparts on the subject, regardless of how far away the subject is -- meaning it is the the subject's total angular size AND texture compared to the percentage of motion that will be relevant to its perceived sharpness in the image.  IOW, if a coin at 1 meter is rendered using  500 pixels, and the moon is rendered using the same 500 pixels, then any camera motion will appear to have an identical negative affect on the edges of both subjects.  We tend to identify motion along texture (edges of relatively higher contrast) in our images, so any given camera motion will impart the same pixel displacement along all similarly-oriented edges regardless of their size or distance; so the rim of a small coin and a large coin at the same distance and the edge of the moon will all appear equally blurred (and blurred in the same direction) by a given amount of camera motion.  However, now we get into the frequency of detail (coarseness of texture) in our subject, and low frequency (widely-spaced or coarse) detail are less affected by motion than high frequency (closely-spaced or fine) detail, since there is more space between the edges.  So if our coins have a lot of fine detail in them, and our moon does not, then the coin's surface may appear more blurred than the moon's surface, even though the edges are identically blurred.  

Clear as mud now, right?

[Ben Rubinstein]

Of course the 1/FL may have applied in the film days given proper technique but stating it as a general rule for 25 megapixel cameras is laughable. I have found though that once you hit your personal limit, anything over that will either be sharp all the way through, almost tripod sharp, or not at all. I've never seen a graduation of sharpness as the shutter speed increases over your personal limit. Maybe it's just me.