You are not logged in.

> The forum rules have been updated. Please take a moment to read them.


#26 2013-02-16 16:01:59

klausesser
Member
From: Düsseldorf, Germany
Registered: 2006-05-22
Posts: 6600
Website

Re: Deformation Problems

lumelix wrote:

Hi Leonardo
You used a lens with a fixed focal distance, but not with a fixed focus wink
And if you change the focus while shooting, you also change the nodal point position in the lens!
This results in more distortions over your pano. So don't change the focus.

@ Klaus
For normal stitching, shift or PC-Nikkor lenses aren't nessessary.
But for special use, they are really interesting: If you shift them, you get some images that are shoot exactly in a plane and can be stitched together very precisely.
So you get something like a virtual bigger sensor area by this shifting. Actually PC-Nikkor lenses have a shift about +/– 11mm. So you get a stitched sensor size of 46 x 58mm, or an equivalent of 111 MPixel on a D800 !

No question - i do it also: stitching by SHIFTING.

But we talked about stitching around thge NPP - and here you´ll most likely run into massive problems using a shift-lens because you´re shooting off-axis when you shift.

wink

best, Klaus


If you want something you´ve never had,
then you´ve got to do something you´ve never done.

Offline

 

#27 2013-02-16 16:11:25

klausesser
Member
From: Düsseldorf, Germany
Registered: 2006-05-22
Posts: 6600
Website

Re: Deformation Problems

lumelix wrote:

And if you change the focus while shooting, you also change the nodal point position in the lens!
This results in more distortions over your pano. So don't change the focus.

Focusing is not the cause for this kind of distortions. The cause is non-symmetrical shooting.

Again: imagine a large-format camera or a shift lens on your dslr. You NEED to level the camera in the way the lens-axis is parallel to the ground.
Then you shift the lens parallel to the wall.

The moment you TILT the camera things change.

So: instead of levelling the camera so that the lens-axis is parellel to the ground you need to shoot identical angles up and down for getting the same result as a lens-axis parallel to the ground.

best, Klaus

Last edited by klausesser (2013-02-16 16:11:50)


If you want something you´ve never had,
then you´ve got to do something you´ve never done.

Offline

 

#28 2013-02-17 03:20:38

lumelix
Member
From: Basel Switzerland
Registered: 2010-10-25
Posts: 409

Re: Deformation Problems

Hi Klaus
I think part of the distortion is probably caused by an inexact or moved nodal point and possibly also by a change of focus setting in the upper pano - we don't know so far. When stitching more than 200 such images together with APG, this error will be much more visible.
I only wanted to point out this, in case Leonardo photographed the wall again. Then he can avoid distortions cause by this point wink
If he don't need the high resolution of a 200mm-pano, shift lenses from Nikon (PC-E) are a good choice, because they have less distortion than other wide lenses and can be shiftet to preserve the vertical lines.
I think a this way created 110 MPixel-image of this wall should be enough for every use. It's a real millimeter-resolution in original size. More resolution will not bring more advantages.
If I had to make this picture, I would try it with a PC-E lens and a higher position. The distortion correction should be made ​​correctly with the Photoshop lens-module.


Regards
Martin

Offline

 

#29 2013-02-17 13:18:51

klausesser
Member
From: Düsseldorf, Germany
Registered: 2006-05-22
Posts: 6600
Website

Re: Deformation Problems

lumelix wrote:

Hi Klaus
I think part of the distortion is probably caused by an inexact or moved nodal point and possibly also by a change of focus setting in the upper pano - we don't know so far. When stitching more than 200 such images together with APG, this error will be much more visible.
I only wanted to point out this, in case Leonardo photographed the wall again. Then he can avoid distortions cause by this point wink
If he don't need the high resolution of a 200mm-pano, shift lenses from Nikon (PC-E) are a good choice, because they have less distortion than other wide lenses and can be shiftet to preserve the vertical lines.
I think a this way created 110 MPixel-image of this wall should be enough for every use. It's a real millimeter-resolution in original size. More resolution will not bring more advantages.
If I had to make this picture, I would try it with a PC-E lens and a higher position. The distortion correction should be made ​​correctly with the Photoshop lens-module.

Hi Martin!

A NPP-mismatch results in stitch-errors. Not in distortions. Again: if he would have shot symmetrical - as an equivalent to parallel shift - this kind of distortions wouldn´t have occurred. He tilted the camera upwards and this way got a non-symmetrical POV. You can´t fully compensate it!
If he had shot symmetrical - as an equivalent to levelling the optical axis parallel to the ground - he would not have get this kind of distortions.

I use to shoot this way very often when i shoot hires-architechture. THIS kind of distortions usually can be compensated very easily . . . . IF the shooting is correct.

But in the actual case the "warp" tool in Photoshop can do it also.

best, Klaus

Last edited by klausesser (2013-02-17 13:41:58)


If you want something you´ve never had,
then you´ve got to do something you´ve never done.

Offline

 

#30 2013-02-17 13:44:43

lumelix
Member
From: Basel Switzerland
Registered: 2010-10-25
Posts: 409

Re: Deformation Problems

klausesser wrote:

A NPP-mismatch results in stitch-errors. Not in distortions.

Hi Klaus
Again: I only want to point at this so he can avoid eventually errors when shoot again wink

A wrong setting for the nodal point will result in paralax errors between the images, and this in stitch errors - I agree.
In this case we have a near flat object, so paralax errors are small and should not result in stitching errors.
But I think, it will also result in global distortions, if you swing the lens upwards and then try to stitch these images together.
It's because the rotation is not around the optical center/nodal point of the system. If the rotation isn't around the NP the POV
will change while rotate, and so the perspective. Any deviation from the optimum condition will cause in errors.

To produce an possible undistorted image/pano a precise setting of the nodal point and fixation of the Focus adjustment is inevitable.
The optimization in APG is also to complex to understand what's going on in detail. It's for stitch images, not for avoid distortions wink
For the creation of undistorted orthophotos we have a totally other kind of software.

Last edited by lumelix (2013-02-17 13:58:12)


Regards
Martin

Offline

 

#31 2013-02-17 13:53:44

klausesser
Member
From: Düsseldorf, Germany
Registered: 2006-05-22
Posts: 6600
Website

Re: Deformation Problems

lumelix wrote:

klausesser wrote:

A NPP-mismatch results in stitch-errors. Not in distortions.

Hi Klaus
Again: I only want to point at this so he can avoid eventually errors when shoot again wink

Hi Martin!

Fine. But he will encounter the very same issue as long as the shooting doesn´t match the very basic optical correlations.

best, Klaus


If you want something you´ve never had,
then you´ve got to do something you´ve never done.

Offline

 

#32 2013-02-17 14:00:06

lumelix
Member
From: Basel Switzerland
Registered: 2010-10-25
Posts: 409

Re: Deformation Problems

klausesser wrote:

Fine. But he will encounter the very same issue as long as the shooting doesn´t match the very basic optical correlations.

Yes, I totally agree!


Regards
Martin

Offline

 

#33 2013-02-17 23:39:49

con
Member
From: Dublin, Ireland
Registered: 2011-06-13
Posts: 69
Website

Re: Deformation Problems

klausesser wrote:

Yes. Ideal geometry iexists in theory only. You can´t photograph theories . . . . wink

Nice smile

Last edited by con (2013-02-17 23:41:35)

Offline

 

Board footer

Powered by PunBB
© Copyright 2002–2005 Rickard Andersson