National War Memorial of Canada Pano - revisited

[Manfred M]

Tonight was the last night of the Christmas lights in downtown Ottawa, but the weather forecast called for snow all night long. Yesterday, on the other hand was the last night of clear skies during a cold snap, so being in the area for another photo shot, I dragged my pano gear along. The temperature was around -26°C / -15°F at the time I was shooting, so the scene was extremely clear. I was rather cold by the time I got back to the car...

This shot is a 7-image pano taken with the Nikkor 24mm f/3.5 PC-E lens. I did a fair bit of colour cleanup as well as some dodging and burning to do this shot. I'm still not 100% happy with the camera / lens rotation. I recalculated the no-parallax point for this camera / lens combination, but it's still not quite right.

[Tronhard]

Manfred even if you are not completely satisfied I think that is absolutely gorgeous!

[Shadowman]

Nicely captured.

[Artshot]

I really don't know enough about pano's to say anything other than very nice. But you really have set the bar as far as getting out there under adverse conditions. It inspires me much more than brilliant photoshop work.

[Rick50]

Yes, this came out very well.

[george013]

Tonight was the last night of the Christmas lights in downtown Ottawa, but the weather forecast called for snow all night long. Yesterday, on the other hand was the last night of clear skies during a cold snap, so being in the area for another photo shot, I dragged my pano gear along. The temperature was around -26°C / -15°F at the time I was shooting, so the scene was extremely clear. I was rather cold by the time I got back to the car...

This shot is a 7-image pano taken with the Nikkor 24mm f/3.5 PC-E lens. I did a fair bit of colour cleanup as well as some dodging and burning to do this shot. I'm still not 100% happy with the camera / lens rotation. I recalculated the no-parallax point for this camera / lens combination, but it's still not quite right.

How did you calculate the non-parallax point and why is this now not quite right. I know it's a small image but I don't see any fault in it.

George

[Manfred M]

How did you calculate the non-parallax point and why is this now not quite right. I know it's a small image but I don't see any fault in it.

George

The fault I see is quite small and is clearly visible when I view the image at 100% magnification.

The setup to take a panorama I use is shown here; the camera is held vertically using an L-bracket. This in turn sits on an indexing head which attaches the camera to a long plate that allows the camera to slide back and forth along the plate.


(Note: the indexing head is not shown in this image)

Finding the no-parallax point involves putting two thin targets that line up and pointing the camera at them. Using the indexing head, the camera is rotated from side to side. If there is a shift between the two targets, the camera is slid up (or down) the long focus plate until this no longer happens. This is the no parallax point where the camera rotates around the entrance pupil. At this point, there should be no distortion when shooting the pano. The long focus plate has markings in 1/2 cm increments, so I can get a good estimate of the no-parallax point.

There are a number of references on how to do this; this is one of them:

http://www.johnhpanos.com/epcalib.htm

When I look at the shot qt 100%, this some minor bowing visible. Those lines should be straight and parallel.



There are three likely explanations:

1. I did not accurately measure the no-parallax point when I did my tests;

2. I did not set up the camera correctly when I took the shot (always a possibility when working in harsh weather); and

3. The shot was taken with a shift / tilt lens and there are no correction profiles available for these lenses in the software, so this could just be lens distortion. The lens was definitely shifted upwards to take this series of images.

[Tronhard]

One thing that is important when doing the calibration is to absolutely clear that what you are shooting against is level and straight. A fellow photographer spent ages getting frustrated because he did a similar thing to Manfred but always got a slight dip in his image. On inspection the dip was actually in the surface of the ground for drainage!

[Manfred M]

One thing that is important when doing the calibration is to absolutely clear that what you are shooting against is level and straight. A fellow photographer spent ages getting frustrated because he did a similar thing to Manfred but always got a slight dip in his image. On inspection the dip was actually in the surface of the ground for drainage!

Totally agree. I use a hot shoe mounted 2-axis bubble level when setting up the shot. These are far more precise than the simple bubble levels built into my tripod head.

I am 100% certain that the tripod was level for both the test and the shot.

[george013]

Totally agree. I use a hot shoe mounted 2-axis bubble level when setting up the shot. These are far more precise than the simple bubble levels built into my tripod head.

I am 100% certain that the tripod was level for both the test and the shot.

The bubble level on your camera is not relevant. The turning plate must be level. And the camera with the 2 horizontal axes parallel to that plate. When they're not the uncropped pano will look as a saw.
Isn't it that even when everything is in the right line,only the middle line will be horizontal?

I see it's a Great War Memorial.

George

[Manfred M]

The bubble level on your camera is not relevant.

In theory, I would agree. In practice I have found this is not the case.

I can level indexing head and clamp down the camera. This partially obstructs the view I have on the level located on the quick release bracket. Adding the camera bubble level lets me confirm that the camera is level throughout the swing and attaching the camera has not caused any deflection / sag or tripod movement. If the hot shoe bubble level indicates a problem, I have to remove the camera and re-level the indexing head.

[george013]

In theory, I would agree. In practice I have found this is not the case.

I can level indexing head and clamp down the camera. This partially obstructs the view I have on the level located on the quick release bracket. Adding the camera bubble level lets me confirm that the camera is level throughout the swing and attaching the camera has not caused any deflection / sag or tripod movement. If the hot shoe bubble level indicates a problem, I have to remove the camera and re-level the indexing head.

I've made only a few panos, most of them just out of the hand.
I don't see any fault in your image. It's small. But the only line that should be horizontal should be the middle horizontal line. Line above bend up to the sides, lines down bend down to the sides. If you want them horizontal than that's something for the pre-processing, not the camera.

I don't know what the influence is of a tilt/shift in a pano.

George

[Manfred M]

I've made only a few panos, most of them just out of the hand.
I don't see any fault in your image. It's small. But the only line that should be horizontal should be the middle horizontal line. Line above bend up to the sides, lines down bend down to the sides. If you want them horizontal than that's something for the pre-processing, not the camera.

George

It's the curvature of the horizontal lines that should not be there George; that's the issue with that shot and yes it is quite minor.

I've done quite a few hand-held panos too. As long as there as the foreground is far away, the parallax issues are not noticeable. The moment they are close, the shooting gets trickier and the use of no-parallax point rotation is important in the shot.

[george013]

It's the curvature of the horizontal lines that should not be there George; that's the issue with that shot and yes it is quite minor.

I've done quite a few hand-held panos too. As long as there as the foreground is far away, the parallax issues are not noticeable. The moment they are close, the shooting gets trickier and the use of no-parallax point rotation is important in the shot.

But the straight horizontal lines are a correction of the curved line made in pre-processing, not a fault of your setup. As soon you rotate the camera the same subject will have another magnification, the distance to the sensor changes. A kind of fish eye effect.
Just a simple theory of me.

George

[Manfred M]

But the straight horizontal lines are a correction of the curved line made in pre-processing, not a fault of your setup. As soon you rotate the camera the same subject will have another magnification, the distance to the sensor changes. A kind of fish eye effect.
Just a simple theory of me.

George

Yes and no. As these shots were taken with a 24mm rectilinear wide angle lens, this issue generally goes away for subjects a few meters away from the lens, so long as the rotation is around the no-parallax axis. At the distance, I shouldn't be seeing that.

Here is a pano I took a few years ago with a different lens at the same focal length (24mm) rotating around the no-parallax point. Here those lines are parallel; but unfortunately the image is not as sharp as I would like, so I know it can be done. That lack of sharpness is why I went back to redo the scene this year.

[george013]

Yes and no. As these shots were taken with a 24mm rectilinear wide angle lens, this issue generally goes away for subjects a few meters away from the lens, so long as the rotation is around the no-parallax axis. At the distance, I shouldn't be seeing that.

Here is a pano I took a few years ago with a different lens at the same focal length (24mm) rotating around the no-parallax point. Here those lines are parallel; but unfortunately the image is not as sharp as I would like, so I know it can be done. That lack of sharpness is why I went back to redo the scene this year.

Especial with wide angle, short focal length, you have this behaviour. M=f/(d_object-f). The smaller f the bigger the differences.
Of course it can be done. But I don't think it's wise to use a tilt/shift with a pano. Making all kind of effort to get that camera level and then using a tilt\shift?

I've some tests in my mind. I'll show them when I made them.

George

[Manfred M]

But I don't think it's wise to use a tilt/shift with a pano.

Why would you say that? I know that some photographers use them in vertical orientation to get a double height pano. One pano is shot with the lens shifted to be near the bottom of the travel range and the other shot is taken with the lens shifted to shoot at the top of the travel range. This is something I have not tried yet, mostly because the scenes I have been shooting don't need that much vertical coverage.

[george013]

Why would you say that? I know that some photographers use them in vertical orientation to get a double height pano. One pano is shot with the lens shifted to be near the bottom of the travel range and the other shot is taken with the lens shifted to shoot at the top of the travel range. This is something I have not tried yet, mostly because the scenes I have been shooting don't need that much vertical coverage.

The main point about leveling is to get the optical ax parallel to the plane of rotation. A tilt/shift is disturbing that. Unless I missed something.
And the shorter the distance/focal length, the bigger the influence.

George

[Manfred M]

The main point about leveling is to get the optical ax parallel to the plane of rotation. A tilt/shift is disturbing that. Unless I missed something.
And the shorter the distance/focal length, the bigger the influence.

George

The shift is along the optical axis, so there is no impact on the axis of rotation. The result is very much the same as if I had raised the camera on the tripod column.

George - have you ever shot with a shift / tilt lens?

[george013]

No, I haven't and I've noticed I miss something.

George