D3s + 24 TS-E
Benn away for a while...
Finally got a D3s and managed to get the Nikon 24 mm TS-E shift and tilt lens. Using what I understand to be the method to get the nodal point I got the values of A=86 and B=115 (lower/upper rail). However taking a pano the other week at this setting, the join in PTGui was a little off. On top of that, I used the full shift up and down (10-11 mm).
A number of questions has arisen:
1. Is my figure of 115 correct for the upper rail. Does anyone have a different setting.
2. Is there anything particular to consider when using a TS-E lens, but only shifting in the vertical plane.
3. The figure of 115 was determined with no shift on the lens, in fact just like using a normal 24mm lens. Do I have to many different calculations for a chnage in the vertical shift.
4. Would PTGui have a problem stitching a pano if the 'film' plane is vertical (no lens distortion to really take into account).
If anyone has any experinece in using one of these lenses, and feedback you may have would be much appraeciated. I'll have another go at trying to get the upper rail figure a little later.
Thanks in advance,
Using a shift lens for panoramas is a very bad idea. I assume that you have taken 2 rows at zero pitch and shifted the lens vertically up and down.
What happens when you do this is that each row will have an asymmetric warping and they will not match at all.
I do not have a shift lens so I can not say how good it will work but you should in theory be able to stitch them by choosing individual shift for the 2 rows.
On the lens tab book all images in one row for individual shift.
Of course it is also necessary that you optimize with the advanced interface and book horizontal and vertical shift.
You will get some gigantic values for the vertical shift.
Taking the images with the camera tilted up and down is much better and you also get a larger vertical FOV. PTgui is fully capable of correcting for all changes in perspective and lens corrections for barrel distortion etc.
Light ray enters the lens at different incident angle when different shift is used. Shift can affect the entrance pupil (hence NPP) position, depending on its optical design.
Originally Posted by Makiru
use google for translation.
If no near object is in the scene, shift of NPP is not important.
PTGui should be able to cope. Like Hans, I haven't got a shift lens, but I have generated some sample images from a large equirectangular panorama to simulate a PTGui project for your lens. I have uploaded a 360 degree set of 2 rows of 8 images (resized down for convenience) together with a PTGui project file at: http://www.johnhpanos.com/ely-24mm-shift.zip (10MB). Note that since the shift parameters are in pixels, they will need to be scaled up appropriately for full size camera images. It helps the optimizer if you initialize the shift parameters to approximately the expected values.
Originally Posted by Makiru
Thank you very much for the replies. I think what Hans has written is the best option as I have tilted up and down before with a Nikon 14-24 and have got excellent results. I hadn't considered the angles by which light enters the entrance pupil. I've had a look at the file posted by John and this was the effect I was after. Even though PTGui can 'stitch' the images, it allows for the natural distortion of the lens. What I was after and which is in the ZIP file from John is no distortion, but a vertical image. I'll work on this to see what I can get. Very valuable info.
Brilliant, thanks again.