Either the camera moved more than I thought, or the lookout I was standing on moved?!?

[ameerat42]

(I estimated) 2.5 mm of movement >(across the sensor)< over 80mm FL...
Sensor size does not matter.

But when your trying to figure out how much the camera moved by degrees, wouldn’t the sensor size be a factor in the calculation?


Sent from my iPhone using Tapatalk

Yep! sure does...

So, a disagreement. My argument is based on the simplest trig required to explain it.

Where I might be wrong is the estimate of the streak length. Nevertheless, that streak length
will not matter for a given focal length, whether the sensor is a crop on or not. Just the field
width is different.

Anyway, AK, I couldn't see in your reply where you showed that the sensor size does matter.

On the matter of the bridge, are you saying that it does not appear to have moved the same
distance as the other streaks? I think it has.

[tandeejay]

No, I think AK was saying the bridge is showing that the movement wasn't smooth...

- - - Updated - - -

or. the lights on the bridge were flashing?

[ameerat42]

...or. the lights on the bridge were flashing?

Ahh! Clearly, an ambulance was crossing at the time

[tandeejay]

Actually, the truth is, I was trying to get a photo of a pelican for the Panning weekly challenge


Pelican Brisbane Panning by John Blackburn, on Flickr

[arthurking83]

No, I think AK was saying the bridge is showing that the movement wasn't smooth...

..... [/COLOR]

or. the lights on the bridge were flashing?

Doh! didn't thunk of that possibility
Yeah, if there are concentrated spots of light, then either the light is variable in strength, or the movement over time wasn't smooth

Actually, the truth is, I was trying to get a photo of a pelican for the Panning weekly challenge ..

As Harry Hoo would say .... "two possibilities!"
I'm surprised you cloned the pelican out to begin with

So, a disagreement. My argument is based on the simplest trig required to explain it.

Where I might be wrong is the estimate of the streak length. Nevertheless, that streak length
will not matter for a given focal length, whether the sensor is a crop on or not. Just the field
width is different.

....

It should.
Like Tanning said, the length of the streak will depend on the FOV and the amount of movement.

So a camera with a 5° FOV with a 500pixel streak where the sensor is 5000pixels wide needs a lot less lateral movement than it needs for the same 500pixel streak with a lens that shows 50° FOV.

Think of it in terms of star trails, or blurry moon.
With start trails, using a 12mm lens, for a 30sec exposure may only give you a 5 pixel trail.
Using a 1000mm lens instead will render more like 100pixels of trailing.

For the purpose of explanation(ie. not real numbers) lets say the earth rotates 1° every 30sec, we know it's less, but for the sake of simplicity!

12mm lens gives a 120° FOV
1000mm lens shows a 1.2° FOV
the earth still only moves at 1° every 30sec.

12mm lens was only moved 1% of the frame, 1000mm lens moved by almost 100% of the original frame.

I think(from memory) it only takes about 2mins for the moon to start from one corner of the frame to the other corner of the frame when using an 800mm lens on a 135 format frame.
It may be 5 mins. I just can't remember, but I did the test a while back but it is very quick.


But in a technical sense, you're right Am ... a streak doesn't rely on the focal length or sensor alone.
But in John's situation here, we know it's an 80mm focal length set, and the sensor is a 1.5x crop .. and that it has 6000pixels on the long axis(with 500 pixels of movement).

If we didn't know the FOV value, then we couldn't calculate the amount of movement effected.
That is, we don't know the FOV in the image unless we know the sensor lens relationship.
Remember that the frame can be made the same whether a 10mm lens is used, or a 1000mm lens is used(you just need a lot of room to move about! )
The major difference will be perspective(compression/extension of the varying elements in the scenes).
But you can get the framing the same. What we can't work out just from the image is a value for the FOV, which then gives us a percentage amount of movement, and eventually how much the camera moved.
1.4° is very little movement, so hard to detect in some situations.

I can't imagine the VR mechanism would cause camera movement, and I din't think it moves by that much(ie. 10% of the frame).

One thing I reckon I can be sure of tho... I don't reckon you could replicate that same rendering if you wanted too using the same method.
You probably could do so, on a tripod tho(with the panning doodad).

Could have been something as simple as something (very small stone or twig or something) getting stuck between camera and railing, and causing movement whilst you had some pressure on the camera to keep it from falling into the abyss.
if there was any hint of wind too, then the force of the wind on you and your hand could have masked the movement too.. remember only a very small amount of movement there.


Anyhow, I reckon it was the (cloned out) pelicans fault. Is there any phosphate compound, possibly originating from a pelican, on the camera or lens!

[feathers]

I like the first image, as well as the flat lining in the second

[tandeejay]

In relation to the VR, I was not particularly thinking of the VR mechanism creating that much image movement, but of vibrations in the camera body caused by the VR mechanism attempting to compensate for a lack of camera movement that might have caused the camera body to move?

There must have been some breeze as the branches in the tree show movement that is definitely not linked to any camera movement.


Sent from my iPhone using Tapatalk

[tandeejay]

Also, if all you have is the photo can’t you figure out the FOV if you know the location of the camera and some of the landmarks In the image? Plot the points on the map, and then just measure the angle


Sent from my iPhone using Tapatalk

[ameerat42]

To continue this discussion, it is much easier to use text color to differentiate between what AK said
(or which I have left out as "...") and what I am saying...


It should.
I've forgotten what this relates to.

Like Tanning said, the length of the streak will depend on the FOV and the amount of movement.
Not disputed, but not exclusive.

Think of it in terms of star trails, or blurry moon.
With start trails, using a 12mm lens, for a 30sec exposure may only give you a 5 pixel trail.
Using a 1000mm lens instead will render more like 100pixels of trailing.
Utterly agree with this.

But in a technical sense, you're right Am ... a streak doesn't rely on the focal length or sensor alone.
No to the first bit, and yes to the second.

But in John's situation here, we know it's an 80mm focal length set, and the sensor is a 1.5x crop .. and that it has 6000pixels on the long axis(with 500 pixels of movement).
Not that anything but the FL really matters.

If we didn't know the FOV value, then we couldn't calculate the amount of movement effected.
Not necessarily, as long as we had other parameters, like FL and frame size, as we have here; and anyway, the assumption here is
that it's the whole frame shown. (Tands did not say otherwise and this is what I based the streak length on.)

That is, we don't know the FOV in the image unless we know the sensor lens relationship.
Yes!/Congruency!/At one!/Etc!

Remember that the frame can be made the same whether a 10mm lens is used, or a 1000mm lens is used(you just need a lot of room to move about! )
The major difference will be perspective(compression/extension of the varying elements in the scenes).
You must mean the scene in the frame. But the perspective will only change if you change the subject distance, not
the focal length. I assume you mean you are able to dramatically change the sensor size while changing the FL from
10mm to 1000mm.

...
What we can't work out just from the image is a value for the FOV, which then gives us a percentage amount of movement, and eventually how much the camera moved.
You can using sensor size and lens FL, using simply "similar triangles".

1.4° is very little movement, so hard to detect in some situations.
For situations of extremely wide-angle lenses, maybe, but 1.4° is ~ 3 full-moon diameters. From an F=80mm lens
(using similar triangles or sine or tangent for such a small angle) it would show as ~1.95mm on the frame - no matter
how big that frame.
...

To summarise: I'm just using the simplest approach that will explain it
(And now my throat is dry!)

[arthurking83]

So, I'm reading that we have both agreemement and disagreement in a few pockets of resistance.

You said:

.... Nevertheless, that streak length
will not matter for a given focal length, whether the sensor is a crop on or not. Just the field
width is different.

....

Are you referring to the streak, or the image itself(unclear)

I said:

...

It should.
Like Tanning said .....

the assumption is that there won't be a crop, and that the pixel dimensions will therefore be known.
Otherwise it's incalculable!(close to it).
FOV depends on the lens/sensor relationship. 80mm lens FOV is narrower than 80mm lens on full frame.
Back onto the crop point before this, if the sensor was 135 format, and the crop was definitely known(eg. to the common 1.5x APS-C format) then FOV can easily be calculated.

I wouldn't rely on estimations of FOV going by fixed points in a scene. For this instance, it may work with a degree of error .. but not guaranteed to work in all cases.
As this now turned into an extended discussion, and not only for specific assistance to the original post .. then we should try to keep the topics applicable to more situations.

Perspective doesn't change with sensor size, only with lens focal length.
**Many modern lenses change focal length with focusing, so focusing can also affect perspective, but an ideal lens(that doesn't change focal length with focusing) won't change perspective either.

What wouldn't matter (in this topic of movement vs FOV) in terms of sensor 'size' is pixel density.
Had the camera been a 6Mp(2000x3000) type, then what would have been different would have been the number of pixels from one end of the steak(s) to the other.
ie. half again(from 500 down to 250).
The movement was simply that the camera move X number of degrees, or a percentage of the FOV.


as a hypothetical: lets say Tandee created those 500 pixel streaks in photoshop, to simulate movement, then on the 6000pixle frame that equated to about 8% lateral rotation.
Had the sensor been the 6Mp 3000pixel across type, and using the same 500pixel streak effect, then the percentage of 'movement' would then amount to double that lateral movement(ie. 16%).
When I referred to the 10mm vs 1000mm scenario I meant actually moving the camera/lens to simulate the save FOV at infinity.
So I suppose in theory we could determine FOV without any idea as to focal length and sensor size, and rely solely on some reference points in an image, but if you know the coordinates of those reference points. Not easy to do in every situation .. but doable(I suppose).
An arduous task, at best! with most of my images which involve many flat featureless outback scenes tho!

.... For situations of extremely wide-angle lenses, maybe, but 1.4° is ~ 3 full-moon diameters. From an F=80mm lens
(using similar triangles or sine or tangent for such a small angle) it would show as ~1.95mm on the frame - no matter
how big that frame.

Can't work.
Imagine two extremes:

1. 80mm lens on a smartphone sensor. Would equate to a 3000mm lens on a full frame camera. Point that at the moon and you won't get the moon edge to edge(horizon to horizon).
2. 80mm lens on an 8x10 view camera(if you could possibly ever locate one .. I think non existent). Would make that lens an UWA type lens on that 'sensor size' if such a lens was possible. I'd guess at least a 120° FOV.

So while an 80mm lens on <whatever sensor> may allow 3 moon diameters within 1.4°, on a smartphone it'll be more like 0.5 moon diameters for the entire frame, and possibly 10 moons in that same 1.4° of angle on the view camera sensor.

This is why the topic of focal length and sensor size 'cropped up'(pardon the pun) .. just an easy way to determine FOV, then having worked out percentage of movement of the camera(via the streak length) .. it was easy to show that the camera ... 'barely moved' in relation to the operator, but much more obviously in relation to the rendered image.

[ameerat42]

Phew! Let's call a break. There's enuff info here for people to wade through and conclude for 'emselves.

Maybe in a day or two I'll put up a diagram to illustrate an otherwise 1000 words. I've only persisted in the discourse
so far because it's not something that can be just left aside.

As Walter Concrete may have said, "Cementics can get in the whey!"

Are you happy to take a breaver?

[tandeejay]

Wow! I never imagined my image would generate so much discussion!

Might need to win the Creative processing challenge so I can put it in there

[ameerat42]

Wow! I never imagined my image would generate so much discussion!

Might need to win the Creative processing challenge so I can put it in there

Ah, so that was it?!! You had it in the wrong section

[Mark L]