You need not be so pessimistic. Typical alignment deviations are in the
3 to 5 pixels ballpark at the edge, and can go up to 15 pixels in the
corner for a particular "bad" lens (meaning the hugin abc model cannot
cope with that lens), all using the barrel distortion parameter b. Now
do the maths for an image 4000 pixels wide with 40 degrees field of
view. 10 pixels still give you 1/10 degrees.
Another issue are EXIF-provided lens parameters. If hugin takes them as
gospel (view not optimised) in the way it does, one may be wrong in
focal length by a few percent. The way to calibrate this is to take a
360-degrees panoramic and let hugin determine the parameter v, and
possibly the full set of parameters for your individual camera lens.
On 30.03.20 15:37, 'ChameleonScales' via hugin and other free panoramic
software wrote:
Ok. It's fine if it doesn't get that precise. I can work with around 1 degree
if the software doesn't allow for less.
The point is that having a scaling feature would allow me to fine-tune the
scale based on what I know.
Since I superimpose my panorama to imported terrain and map data in Blender, I
can see when a building or hill should be slightly more to the left or right on
the photo, then adjust the FOV in Blender using the slider I made, and I know
this FOV to be correct. I just don't know an easy enough way to give this back
to Hugin so it exports a corrected version.
‐‐‐‐‐‐‐ Original Message ‐‐‐‐‐‐‐
On Monday, March 30, 2020 10:35 AM, Klaus Foehl <[email protected]> wrote:
As you are talking sub-degree precision, there is an inherent limitation
in the hugin lens model or abc parametrisation. From the Brown-Conradi
model, a mathematically sound distortion description, hugin implements
only one of the non-trivial parameters, which is parameter b.
Parameters a and c are not in Brown-Conrady, for polar coordinates they
are mathematically not sound, and in practice their use does not lead to
the quantitative alignment improvement an extra good parameter would
provide.
To add to it, I have seen situations where the use of a and c parameters
have made things observably bad. If parameters b and the offset
parameters d and e provide you with enough precision, then fine, then
hugin is a really good tool for you.
On 30.03.20 09:37, 'ChameleonScales' via hugin and other free panoramic
software wrote:
Unless I'm missing some geometrical effect, it doesn't seem to me that you
would have to re-optimize it given that the transformation should precisely
preserve the panorama's sewing just like in the animation above, so as I
understand it, control point distances should only change proportionally to the
scale you apply. But correct me if I'm wrong.
As for getting it right in the first place, in my use case, I have to
superimpose a panorama to its virtual 3D environment in Blender and I need
sub-degree precision on the HFOV.
I don't think any software could make such a precise guess with the photos I
get from my clients.
--
A list of frequently asked questions is available at:
http://wiki.panotools.org/Hugin_FAQ
------------------------------------------------------------------------------------------
You received this message because you are subscribed to the Google Groups "hugin and
other free panoramic software" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To view this discussion on the web visit
https://groups.google.com/d/msgid/hugin-ptx/282135e3-7846-f179-5382-410c95b68c4e%40gmail.com.
--
A list of frequently asked questions is available at:
http://wiki.panotools.org/Hugin_FAQ
---
You received this message because you are subscribed to the Google Groups "hugin and other free panoramic software" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To view this discussion on the web visit
https://groups.google.com/d/msgid/hugin-ptx/445e41d9-4ead-1722-025c-3df0538fb19f%40gmail.com.
