Hi everyone,
I've been working on stitching a lattice/grid of 4x4 images together
with Hugin. I'm running into problems where parts of the final image
gets cut off and I am hoping to find some advice here to prevent that
from happening.
A few quick details: the images are made at a high magnification and
high resolution (151PM), a thumbnail of the layout can be found here
[1]. The stitching is part of an effort to digitise microfiche cards and
the stitching is a fully automated process - so people aren't supposed
to say launch Hugin and fix some problem(s). I have written a custom
control point finder using ORB which utilises the fact that is knows
which images overlap and in what regions they overlap. The output images
are about ~1.3 gigapixel grayscale images.
The Python code that I have written can be found here [2]. Note that the
custom control point finder seems to work well and shouldn't be the
cause of my issues here - and neither should multiblend, as I can see
the fov causing a bit to be cut off in Hugin too.
So my problem is that in some cases, parts of my images get "cut off" in
the final blended image, which I have come to understand is a function
of the 'field of view' that can be set using pano_modify (and in Hugin
as well).
In the past I relied on pano_modify to find the field of view by itself
(using --fov=AUTO) but this caused the same problem as above: where a
part of the final image is cut off. So currently I set the fov to a
'known good' value for my setup - but this still causes some of the
images to be cut off. An additional problem is that if I raise the fov
much further, the pano_modify tool will hang (run forever) or crash when
trying to automatically calculate the canvas and/or crop size using
--canvas=AUTO and --crop=AUTO. Additionally, --canvas=AUTO seems to
create a canvas that "feels" like it is way too large and I would be
able to fit my stitched images on it more than a few times, but I don't
know enough about how this works to understand how this could go wrong.
Summarising:
1. --fov=AUTO doesn't work for me in some cases
2. Setting a (too) large fov breaks --canvas=AUTO and --crop=AUTO
3. --canvas=AUTO seems to usually create a unwieldy and overly large canvas
Does anyone have any tips on how to deal with this? I feel like there's
something that I am missing.
Ideally I'd just tell the software to include as much from all my input
images as possible without having to specify the field of view or canvas
size, or maybe I could write some custom code to try to set the right
values?
Regards,
Merlijn
[1] https://archive.org/~merlijn/micro_IA40385005_1500_itemimage.png
[2] https://archive.org/~merlijn/microstitch-hugin.py
--
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/c6ef09fa-4f6d-45f1-8fdd-ea9fa628c3cf%40archive.org.
