On 19 August 2019 08:27:29 CEST, Vegard Brenna wrote:
>
>3. My source images are HDR, produced by a different app. The sizes may
>vary by a pixel or two. Each time the size is different, Hugin assigns
>a new lens. Hugin computes HFOV, which varies between 42° and 44°.
>Really? That's more than 4%. 1/3887 = 0.025%
Hugin makes an initial FoV calculation based on EXIF data in the photos (if
any). This is not very accurate, so it will recalculate FoV as part of the
optimisation/fitting process.
Note that these optimised FoV numbers are those that give the 'best fit' in the
panorama, so unless you are careful to only pick control points on objects very
far away, the calculated FoV will vary every time.
>5. I'm unable to find an explanation anywhere of what Distance means.
>What is Optimal? I settle with the fine-tuned results, which are not
>distances, but correlations. But I have no idea what it is the
>correlation between.
The error distances are measured in pixels in the output panorama (so they will
change if you resize the canvas area). For example, a large distance of 10
pixels corresponds to a 10 pixel dislocation in the panorama, which may be very
visible.
The exception is immediately after you run 'fine tune all points', here the
list of distances is temporarily overwritten by the correlation between the
local image around each half of the control point. In this case small numbers
are a _bad_ thing, as zero represents no correlation and one represents perfect
correlation - this is not the best bit of UX in Hugin, but it does the job if
you know to expect it.
>6. What does fine-tune do? Run through all the CPs and assign them to a
>common, consistent mapping universe?
Cpfind identifies lots of features in the scene (using corner detection) and
describes them, it is these descriptions that are compared to create the
initial set of control points. Often these features are not readily visible to
the human eye, and the provided point coordinates may be one or two pixels off
where a human might place them.
The fine-tune function does something else altogether. Here Hugin is directly
comparing the image around each half of the control point - you can visualise
this as subtracting one image fragment from the other and seeing if any
features remain. This is much more like what you do when manually adjusting a
control point, and the process gets very different results to cpfind.
>10. After Align, I may or may not get a well assembled picture.
>Sometimes it's beautifully seamless, sometimes it's gross, with partial
>images not matching.
Yes, this indicates that the automated process isn't working and you need to
manually inspect and fix the set of control points before reoptimising.
>18. Stitch! does not pass any arguments to PTBatcherGUI...
This is supposed to work...
>19. ... so I open the PTO file manually from within PTBatcherGUI and
>click Play. No big deal, compared to the rest.
>PTBatcherGUI then randomly crashes with
> enblend: failed to open "image0000.tif": No such file or directory
This could be a problem with file permissions, does Hugin have permission to
write in the output folder? In the past Hugin has had many problems with
special control characters or untested language characters in filenames and
foldernames, so it could be something like this.
>I have no web page, and attaching files to posts is frowned upon, so
>where do I go from here?
Please reply and attach a PTO project that demonstrates the problem, there is
no need to attach photos at this stage.
Hope this helps.
--
Bruno
--
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/F7AD81D5-29A8-4DE9-B7DA-E086945328C7%40postle.net.
