Blur: Modified by Distance?

[EricGullicksen]

Been trying to work this out in my head for a few weeks, unfortunately the search terms are so vague I can only think to ask it here:

There is a rock ten feet in front of you.  There is a mountain one mile in front of you.  You take a picture with both in the frame, handheld, and you yaw the camera while doing so.

Will the rock or the mountain be more blurry (or will the detail in each be blurred by the same amount)?

I'm not talking about apparent loss of detail but rather pixel by pixel/grain shift in each. 


What spurred this thought is that when you're shooting a gun, inaccuracy caused by the slightest amount of yaw or pitch in the rifle is amplified by your distance from the target.  However, when shooting a camera, the sensor or film is the target, and the light can perhaps be thought of as the bullets. 

This is further complicated by the fact that a bullet striking a target is (for all intents and purposes during a competition) an instantaneous event, whereas the time in which an exposure takes place is highly relevant (ironically, given that light is many times faster than any projectile).  Perhaps camera yaw can be thought of as a rifle target moving very quickly, causing the bullet to not only punch a hole but also tear sideways a bit through it? 

Or, if distance is irrelevant to yaw blur, as a piece of paper with a rod through it extending to infinity at every point.  Yaw will cause the paper to tear, but the distance of the object has no effect.

Please help.

[Roman Racela]

What are you trying to accomplish?

If you're trying to shoot handheld 2 objects at 2 different distances and 2 different perspectives and want them to be sharp when combining them in photoshop, then the answer is yes photographers already do that...including me. Take your camera off autofocus (especially if you're shooting into the Sun) and shoot the 2 scenes. Make sure you manually focus on your foreground scene and take a photo, then shift your camera's angle and manually focus on your background scene and take a photo. Now you have 2 sharp images (1 foreground and 1 background). That's the easy part. The hard part is manually doing a perspective blend in Photoshop.

[pcgpcg]

When you cause blur by moving the camera while the shutter is open you are essentially just shifting the image (projected by a lens) that falls on the sensor. The sensor has no knowledge of where the light came from, i.e. it can't tell if it came from a foot away or a million miles away. So if, while the shutter is open, you move the camera in a manner that results in a rock ten feet away being moved by six pixels, you will also shift the edge of a mountain in the background by six pixels. The result will be a captured image that looks equally blurry everywhere.

[EricGullicksen]

When you cause blur by moving the camera while the shutter is open you are essentially just shifting the image (projected by a lens) that falls on the sensor. The sensor has no knowledge of where the light came from, i.e. it can't tell if it came from a foot away or a million miles away. So if, while the shutter is open, you move the camera in a manner that results in a rock ten feet away being moved by six pixels, you will also shift the edge of a mountain in the background by six pixels. The result will be a captured image that looks equally blurry everywhere.

Thank you, that was exactly the answer I was looking for.

[luxborealis]

When you cause blur by moving the camera while the shutter is open you are essentially just shifting the image (projected by a lens) that falls on the sensor. The sensor has no knowledge of where the light came from, i.e. it can't tell if it came from a foot away or a million miles away. So if, while the shutter is open, you move the camera in a manner that results in a rock ten feet away being moved by six pixels, you will also shift the edge of a mountain in the background by six pixels. The result will be a captured image that looks equally blurry everywhere.

True, to a point. If you take a photo from a moving car, the foreground will be far blurrier than the background as its speed relative to the camera/sensor is much greater. See attached.

[Iluvmycam]

OP...Jesus...with all that stuff going through your head no wonder you guys are landscapers and not street photos.

Here is the textbook example of motion blur in relation to camera distance...but it is for street photogs. 

 
https://danielteolijr.wordpress.com/2015/04/25/an-example-of-motion-blur-in-relation-to-camera-distance/

[pcgpcg]

True, to a point. If you take a photo from a moving car, the forground will be far blurrier than the background as its speed relative to the camera/sensor is much greater. See attached.

For pure translation that is correct, but not for yaw, as in the example stated by the OP. I think, however, that the answer is not as simple as I had previously stated. Assuming the camera is being yawed around the sensor, the farther the nodal point of the lens is from the sensor the more potential there is for some parallax shift to show up as shifting pixels during exposure, though it will be very small and likely less than a pixel in many cases so not noticeable.

[NancyP]

Also figure in distance blur from atmospheric effects (heat waves, smog, etc) - a separate issue, I know, but often a limiting one.

[EricGullicksen]

True, to a point. If you take a photo from a moving car, the foreground will be far blurrier than the background as its speed relative to the camera/sensor is much greater. See attached.

Wait, because in this example, everything in frame is moving at the same speed relative to the sensor: but only in a certain way. Stick a long rod out of a car.  The car moves perpendicular to a mountain and a nearby mailbox that are on the same axis relative to the road the car is on.  The rod will strike the mountain and mailbox at the same time...

However, let's say the photographer can yaw the camera (or rod, whatever).  The speed at which a photographer would have to yaw the camera to keep the mailbox from blurring (or rod to keep it pointed at the mailbox) is far greater than the speed at which he would have to yaw to keep the mountain in focus.  

What pcgpcg said is still valid; I just think that if we are talking about YAW in particular, you are simply not going to be shifting the image of the rock/mailbox and the image of the mountain by the same amount of pixels.

Now I need someone to explain why what I just said makes sense, if it does at all  

[EricV]

Motion blur by camera yaw shifts the image field of view by the yaw *angle*, displacing all pixels on the sensor equally, regardless of object distance.  Motion blur by camera translation (for example, holding camera at fixed angle while in moving car) shifts the image field of view by the translation *distance*.  This must be converted to an angle, which means dividing by the object distance, to figure out the object displacement on the image in pixels.  Distant objects will be displaced by the same physical distance as nearby objects, but this represents a smaller angle and hence a smaller displacement in pixels in the image. 

In a photograph of star trails at night, the blur is caused by the Earth's angular rotation (360 degrees per 24 hours), not by the Earth's lateral translation (30km/sec orbital speed around the sun).

[EricGullicksen]

Motion blur by camera yaw shifts the image field of view by the yaw *angle*, displacing all pixels on the sensor equally, regardless of object distance.  Motion blur by camera translation (for example, holding camera at fixed angle while in moving car) shifts the image field of view by the translation *distance*.  This must be converted to an angle, which means dividing by the object distance, to figure out the object displacement on the image in pixels.  Distant objects will be displaced by the same physical distance as nearby objects, but this represents a smaller angle and hence a smaller displacement in pixels in the image. 

In a photograph of star trails at night, the blur is caused by the Earth's angular rotation (360 degrees per 24 hours), not by the Earth's lateral translation (30km/sec orbital speed around the sun).

Okay, that does make a great deal of sense.  Now just to pick nits, wouldn't pixels at the edges of a sensor be more affected by yaw, as they are physically moving more than the center?

[pcgpcg]

An easy way to visualize that the pixel shift resulting from yaw is the same for background and foreground is to conduct a thought experiment as follows:

Draw a circle and then draw two lines out from the center to the edge that represent the field of view of the lens. This pie shaped piece contains everything in your image. The starry sky is basically on the arc at the edge of the circle. If you draw another arc across this piece of pie that is closer to the center of the circle that can represent where rocks in the foreground are.

Now, say your pie section is 50 degrees. If you rotate it (yaw the camera) 50 degrees in either direction you have completed shifted the background by one frame and also the foreground by one frame. If your frame is 5472 pixels wide, then you have shifted both foreground and background by 5472 pixels. If there is a mountain in the background that fits exactly on the outer arc, and a rock in the foreground that fits exactly on the inner arc, they are both shifted 5472 pixels.

...wouldn't pixels at the edges of a sensor be more affected by yaw, as they are physically moving more than the center?

No, all points on each arc move together as the arcs are rotated about the center of the circle.

[EricGullicksen]

Right.  That's what had me going cuckoo: I was using the wrong criterion - the outside world - rather than the sensor itself.

The rifle analogy actually made it more difficult to imagine, as well.

Since the outside world doesn't change, here is another illustration:  draw a rock in the foreground and a mountain in the background.  Draw a line through them hitting a "sensor."  Now simulate yawing the sensor.  The important thing to see in your minds eye is that the LINE doesn't change.  What it draws on the sensor does, but the line going through the outside world does not shift.