Hello,

I'm trying to make hugin align quite a challenging set of pictures. I've 
taken handheld photos from a number of locations on most of the days during 
last year. The idea is to create a video of how the locations change over 
the course of the year.

I'm using hugin 2020.0.0 on Ubuntu.

I first tried using ffmpeg and vidstabdetect/transform to stabilise the 
video with not-that-great results. Then I turned to hugin and found this 
excellent example of a way to do it: 
https://photo.stackexchange.com/a/108146
I followed the steps there up to using nona, after which I again used 
ffmpeg to create the final video.

When it works, the results are excellent. Here's one example, unprocessed 
images on the left, hugin on the right:
https://www.youtube.com/watch?v=1M4SNpkFLCo

I have about 300 photos from five different locations taken around the same 
time of day with an iphone8.

Now to the problem:

With the other four locations the process most often halts (or slows down 
to a crawl) in the pano_modify step. I've added some progress output 
printouts to pano_modify to see what's taking it such a long time (eating 
100% of CPU all the time).

I can now see that it's busy in 
src/hugin_base/algorithms/basic/CalculateOptimalROI.cpp
autocrop() method.  More specifically it keeps running the nonrecheck() 
method over and over again. I've stopped the run a number of times after 
testRects list has been well over 200k and the progress has been slowed 
significantly. I'm not sure if pto_modify will produce something passable 
or even finish at all if it goes that far.

Most often if pano_modify works it finishes in a few seconds.

Do you have any advice on what could I do to have a better chance of 
getting pano_modify to do its thing?

Is there a minimum number of matches found by cpfind that I should aim for? 
Sometimes it just can't find any matches between some images. In those 
cases I've excluded the problematic images from the series.

Is there a minium Average (rms) distance between Controlpoints that could 
lead to better results? In my current run the autooptimizer ended up with a 
value of 0.474933066833906. After that pano_modify has been running now for 
16 hrs already and has 150k+ testRects in the list :-)

Any advise is welcome!

Best regards,
 Juha


-- 
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/83000a9c-daa2-4cb3-a431-15b922bfd987n%40googlegroups.com.
  • [hugin-ptx] Ho... Juha Kallioinen

Reply via email to