Hi guys,
whenever I perform alignment it doesn’t finish the whole task in the first pass. I rather get a pretty good result for only a part of the images. If I rerun the procedure a couple of times it usually finishes the rest. Am I doing something wrong here? Is there a way to finish in the first pass?
Cheers,
Jo
Hi Jo
It is mostly related to proper image overlap among images. try getting more overlap into your scene and then you can align it on the first run…
Hi Wishgranter,
thanks for the answer. I guess I missformulated my question. What I really wanted to know: What happens in a second pass that didn’t already happen in the first one, if I don’t change any parameters? Also, for some reason, sometimes the procedure just stops without even considering (i.e. extracting key points, matching) all images.
Cheers,
Jo
Hi Jo
It depends on many factors, can you post your ALIGNMENT settings ?
Hi Milos,
I pretty much keep the default settings but usually set the image overlapp to high (as I have a pretty high overlap) and sometimes provide a prior focal lenght estimation. The problem seems to be that at one point it just stops extracting tie points for the images (and therefore also skips the rest of the procedure). The images where the tie points are detected are consecutive, but not necessarily at the beginning or the end of the imput images.
Cheers,
Jo
P.S.: In the second run of “Align Images” it usually does all of them!
Hi Jo
Change it to LOW overlap and run it from scratch 
LOW -> will try to use the whole image
MEDIUM -> medium…
HIGH -> only the “central” part of an image is used
These settings are here because it is not always good to use the whole image space because in most cases the “outer” part of images is of lower quality ( vignetting, blur, etc. ) that is why we use MEDIUM as the default… it is about experience to know when to set different settings as the default ones…
Hi Milos,
unfortunately this doesn’t help with the issue, it took 3 runs to include all images. I don’t mind running the procedure multiple times, I just don’t understand why no key points are detected at all for the remaining images.
Cheers,
Jo
P.S.: In general, I really like your tool!
Hi again,
could you maybe elaborate what exactly happens when alignment is performed multiple times. Because even if there is already a result that used all images, a successive run yields different (sometimes better, sometimes worse) results.
Cheers,
Jo