Hi Robert,
On 14/11/2023 21:00, Robert Mahar wrote:
Yeah, thats whats forcing me to stick with enblend and "jiggle the
handle" till the black tiles go away. This 240 tile project gives 1 - 4
black areas every stitch if the images are in order ( right to left /
top to bottom ). I was going to dig into this, but I've always had
issues building enblend from source owing to vigra, but I guess I need
to figure that out to look under the hood. FWIW I do see this issue
when there is a regular overlap in the shooting pattern as there is in
this project. Unfortunately the tool I'm working on as a substitute for
cpfind needs the order of the images to be preserved at the moment -
once thats fixed somehow then I can permute the image order and sidestep
this.
I am not sure how helpful my suggestions here are going to be, but from
my experience over the past month (doing the same but for photos
microfiche), I can maybe suggest a few things for this:
1. For blending, give multiblend [1] at try if you haven't already. I've
had some problems with enblend producing some odd artifacts at time. I
am not confident enough to say it was really an enblend problem (and not
my project variables / control points at the time), but I appreciate the
speed and reproducibility of multiblend.
2. For control point finding, you said you use a custom one. I also
ended up writing my custom control pointer (ORB based) finder and I was
having problems with visible seams until I realized my control points
weren't actually fitted properly with RANSAC - I wasn't accounting for
rotation in between my images (I expected there to be none). I'm not
saying the problems you're having are because you use a custom control
point finder, but it might be worth ruling that out by just using cpfind
with --multirow
2b. Related to control point finding, I don't know what is the reason
for not using cpfind, but if the reason is performance, then this might
help. If you make a template and apply it using pto_template, you can
use cpfind --prealigned and have cpfind 'do the right thing', so it
won't do excessive matching that you might be trying to avoid. At least,
I've understood this should do the trick at being more efficient.
(Please let me know if you figure this out (it's probably not too hard,
but I didn't try yet)
3. Maybe just open the project file in Hugin and see if the control
points are really all correct/align for the individual images.
Regards,
Merlijn
[1] https://horman.net/multiblend/
On Monday, November 13, 2023 at 10:21:02 AM UTC-5 [email protected] wrote:
The black shadows are an enblend bug, they don't appear if you use
the built-in Hugin blender (which is not as sophisticated as
enblend, but is more stable) - Bruno
--
A list of frequently asked questions is available at:
http://wiki.panotools.org/Hugin_FAQ <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]
<mailto:[email protected]>.
To view this discussion on the web visit
https://groups.google.com/d/msgid/hugin-ptx/430b6057-4fb3-4f29-acb3-9cb3030a1713n%40googlegroups.com <https://groups.google.com/d/msgid/hugin-ptx/430b6057-4fb3-4f29-acb3-9cb3030a1713n%40googlegroups.com?utm_medium=email&utm_source=footer>.
--
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/dc6a45b6-5abf-3ae5-b30a-359c0160f096%40archive.org.
