Nikon 8-15mm @15 mm in very narrow environments

[Manuel Dahmann (kubische-panoramen-de)]

Hi Joost,

with my Samyang 12mm, PTgui generates an average distance of 0.6 points immediately after stitching in my test room – perfect. I am currently trying to achieve similar results with my Nikon 8-15 mm @15mm. First with a DIY head, second with a Nodal Ninja. The nodal point is now perfect, but PTgui #11 and #13.3 both produce very poor results. 3.2 average distance in the test room. This was identical with both heads.
I can work around it by activating “individual parameters” for all (!) images to get 1.4 point average distance. With “Generic Fisheye” it gets even worse.

My guess is that the Nikon 8-15 mm profile stored in PTGui is not yet perfect - or my 8-15 mm lens has a flaw.
Would you take a look at it? I just don't want to post the 9 test photos publicly here.

Thanks in advanve

[Matthew Rogers]

Ivd stitched thousands of panoramas from 20+ different canon 8-15mm lenses and never had any issues. 

if you share the images i'll take a look. 

Matthew Rogers

On 23 Nov 2025, at 09:57, Manuel Dahmann (kubische-panoramen-de) <ma...@malum.de> wrote:

Hi Joost,

--
You received this message because you are subscribed to the Google Groups "PTGui Support" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ptgui+un...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/ptgui/a865ed73-5cee-429c-ad77-550b50b42fabn%40googlegroups.com.

[PTGui Support]

Hi Manuel,

Feel free to share your pictures with me. But I assume I'll get the same
results as you do.

I can think of two possible reasons: the no-parralax-point actually is a
circle. So even at the best panohead setting there will still be
parallax. Inside a room this effect would be more prominent than outside.

Or the lens suffers from a distortion that is not perfectly corrected by
the a/b/c lens model in PTGui. But that can only be concluded after
ruling out all parallax.

The nikon is a complex zoom lens, so I can imagine it deviates more from
the perfect lens model.

Kind regards,

Joost Nieuwenhuijse
www.ptgui.com

> --
> You received this message because you are subscribed to the Google
> Groups "PTGui Support" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to ptgui+un...@googlegroups.com

> <mailto:ptgui+un...@googlegroups.com>.

> To view this discussion visit https://groups.google.com/d/msgid/ptgui/

> a865ed73-5cee-429c-ad77-550b50b42fabn%40googlegroups.com <https://
> groups.google.com/d/msgid/ptgui/a865ed73-5cee-429c-
> ad77-550b50b42fabn%40googlegroups.com?utm_medium=email&utm_source=footer>.

[John Houghton]

Manuel, Exactly what tests have you performed in setting up the head to eliminate parallax issues?

John

[PTGui Support]

Hi Manuel,

Actually I see parallax in your images. Not sever but it is in the order
of 3 pixels. See screenshot.

Kind regards,

Joost Nieuwenhuijse
www.ptgui.com

On 11/23/25 10:57, Manuel Dahmann (kubische-panoramen-de) wrote:

[Manuel Dahmann (kubische-panoramen-de)]

Hi John,
Thanks for your answer.
I shifted the camera in 0,5 cm increments from 9.9 cm (distance from screw) to 11 cm. 10.2 is recommended for Nikon 780 and 8-15mm, I used 10.3.
I took 15 panoramas all in all. The diffferences around 10.2 cm were very small, average distance errors did not change much.Visible stitch errors did never disappear in the upper and lower regions. 

[Manuel Dahmann (kubische-panoramen-de)]

Sorry, it was 0,5 mm shifts of course, not cm, from 9.9 cm (distance from screw) to 11 cm.

[Manuel Dahmann (kubische-panoramen-de)]

One addition: For best parallax in the center, 10.2 cm could be better. For best parallax for zenith/nadir, 10.4 or 10.5 cm would be better. 10.3 is something in the middle.
So, how to handle that? 

[John Houghton]

Manuel, Can I suggest you additionally perform the following test:  rotate the camera down to point directly at the nadir.  Focus on the floor with a small aperture setting for good depth of focus.  Take two shots, panning the camera around through 180 degrees between them.  Align the shots in PTGui with control points only on the floor.  Output a layered psd file for examination in Photoshop. With the the top layer at 50% opacity, you should see something like this:

You want to see the centres of the head perfectly aligned (or the best you can get!).  Any horizontal displacement here can be corrected by a lower rail adjustment of half the visible displacement (as measured via the scale markings).  Any vertical displacement (in this view) might be corrected by a camera mount twist/shift if possible.

John

[Manuel Dahmann (kubische-panoramen-de)]

Hi John,

thanks a lot - very interesting. But not possible to get this result with this lens. Tried a lot.
The lens appears to be decalibrated and has an internal problem. I already suspected this when I was unable to build a DIY head with good results. I have built many heads, and they have always worked with an average distance of less than 1 point. This lens can only be used in such an narrow environment with “individual parameters.” That brings you to 1.4 points average distance, optimized it's even less, and I can no longer detect any stitching errors.

I understand that the use of “individual parameters” cannot be specified either by default or by template, neither in PTGui 11 nor in 13.3. Or am I overlooking something?

[John Houghton]

On Sunday, November 23, 2025 at 4:46:33 PM UTC Manuel Dahmann (kubische-panoramen-de) wrote:

I understand that the use of “individual parameters” cannot be specified either by default or by template, neither in PTGui 11 nor in 13.3. Or am I overlooking something?

Manuel,  I am not aware of any such restriction.  I just tried applying a template with individual lens parameters specified and PTGui (V13.3) did what was expected and was happy to use individual lens parameters .

John

[Manuel Dahmann (kubische-panoramen-de)]

Hi John,

thanks - you are right. I´ve tried that before as a template  with random images (8-15 @15mm of course), individual settings for all images and got bad results. Something else muste have been wrong. Now it did work, thanks.

Regarding the nadir test you described, I want to make sure I understand it correctly.
The value for the offset in the horizontal direction can be measured. This is the distance between the camera base and the center of the lens. For the D780, as with the D750, this is 44 mm.
When I point the 8-15 mm downwards on the head and use the crosshairs to precisely position the NN star in the center of the image, the measured distance is exactly 44 mm then. There is no deviation. When testing with 43 mm or 45 mm, the average distance was even worse.

You can't set anything other than 44 mm with the D780, can you?

[John Houghton]

Manuel,  The technique of adjusting the lower rail setting by positioning the centre point of the viewfinder on the centre point of the head is usually good enough but not necessarily accurate.  The camera might not be pointing down exactly vertically,  due maybe to sag in the supporting arms.  What you are attempting to do is position the entrance pupil of the lens exactly on all three axes of rotation.  So, as you pan the camera round, horizontally and vertically, the entrance pupil remains fixed in position so cannot cause any parallax effects.  By taking two down shots separated by 180 degrees rotation, you can prove whether or not the tripod head is shifting relative to the background (floor) - i.e. exhibiting parallax shift.  It's a simple test and quick to perform, so worth doing.  Perhaps you could show the result you get (like my example).

John

[John Houghton]

Manuel,  BTW - is the camera you are using DSLR or mirrorless?

[Manuel Dahmann (kubische-panoramen-de)]

The Nikon D780 is a DLSR.
Other way around: When I position the NN horizontal position to (measured) 44mm, the crosshair points exactly to the NN star.
There is no bending of the head, the camera is exactly vertical orientated. I´ve tested and measured it all.
Just:  
The ground moves around when I perform your test. The upper and lower halves of the lens are obviously distorted differently. That's what I mean with lens problems. This can be partially compensated with “individual parameters.”

So thanks for your highly valued support!

[John Houghton]

Manuel,  Ah! you mentioned the camera (D780) earlier, so it's a DSLR.  The reason I asked was to see if my novel tutorial for finding the no parallax point might be of interest:  FINDING THE NO-PARALLAX POINT .  But all of this is a bit academic.  As Joost has mentioned, the entrance pupil for a fisheye lens is not precisely located at a single point, but moves about according to the changing direction of light rays entering the lens.  So parallax cannot generally be completely eliminated and is always going to be a problem when shooting in very small spaces.  But can you show the result of taking the two shots down and aligning them on the ground like my example.  Or supply the shots.   I would be really interested to see that.

John

[Manuel Dahmann (kubische-panoramen-de)]

Here is the result. The camera ist exacty vertical on the head, no bending, the horizontal position is 44 mm, the NN star ist exactly in the crosshair center.
But the floor has shifted. There was no way to fix this without shifting the axis of rotation from the center. If you do that, you will logically end up with massive stitching errors.

[John Houghton]

Manuel, No.  You must align the images on the background floor.  That will reveal the parallax movement of the tripod head due to the entrance pupil changing position.

John

[Manuel Dahmann (kubische-panoramen-de)]

But to do what after? The camera is exactly (measured and by crosshair) in the center at 44 mm. I tried 43 and 45 mm, much worse results. Here is the stitched version of the two images within the panorama. This requires individual parameters applied. Quite good now, 1.1 points average distance. No obvious stitch errors.

[John Houghton]

Thanks.  Well, that does show minor parallax effects that would normally be of little consequence.  The lower rail setting might be off by half a mm or so.  You are probably getting the best result you can expect if the lens has these weird distortions .   Incidentally, it's best to have all three legs seated on a hard surface when doing these tests - not on a carpet/rug.

John

[Manuel Dahmann (kubische-panoramen-de)]

I´ve done so many test right now with the lens, in different , but narrow surroundings, 3 different pano heads - never reached a "very good" otimize result without "individual parameter". 

One more question, if you still have time:
I'm a little concerned because I'll soon have a job where I'll need to use the lens. Are there any disadvantages to setting “individual lens parameters” that I should be aware of? So far, I've only used this for really different lenses in a panorama—and for the many linear panoramas I've been doing for a while.

[John Houghton]

On Sunday, November 23, 2025 at 9:12:53 PM UTC Manuel Dahmann (kubische-panoramen-de) wrote:

I´ve done so many test right now with the lens, in different , but narrow surroundings, 3 different pano heads - never reached a "very good" otimize result without "individual parameter". 

One more question, if you still have time:
I'm a little concerned because I'll soon have a job where I'll need to use the lens. Are there any disadvantages to setting “individual lens parameters” that I should be aware of? So far, I've only used this for really different lenses in a panorama—and for the many linear panoramas I've been doing for a while.

Manuel, Using the optimizer to "correct" stitching glitches arising from parallax isn't usually recommended, as it's likely to introduce alignment/distortion problems elsewhere.  Much depends on the distribution of control points.  But if you can get acceptable results without noticeable side effects, then why not?  Good luck.

John

[Manuel Dahmann (kubische-panoramen-de)]

Thanks to your nadir parallax test I could verify the center position with my Samyang 12mm - nearly perfect. about 0,2 mm, after exchanging the lens on the head while maintaining the exact camera position. Nikon 8-15mm deviation is more than 2 mm. So this is a lens-dependent parallax problem. 
It's great that PTGui can fix lens problems like this almost perfectly.