Am Freitag, 20. April 2012 um 11:52:01, schrieb kfj <[email protected]> > On 19 Apr., 22:36, Bruno Postle <[email protected]> wrote: > > > I recently added a new tool called 'ptomorph' to Panotools::Script > > that helps with stitching Hugin panoramas with parallax errors. > > ... > > I'd like to hear some feedback, do you > > think something like this would be useful in Hugin? > > Bruno, this is very good! > > I tried it today with a scan of a map. I had 8 individual images, and > because the map had been folded and I couldn't get it totally flat, I > couldn't get the stitch right. There were jumps by up to about 10 > pixels, and the result wasn't really good enough. So I thought that > since I actually had flat images which were merely distorted and not > parallax errors your tool might be perfect for the purpose. It was. > The stitch from the pto your tool generated came out near perfect, I > could only see 1-2 pixel errors in areas which weren't covered by CPs, > close to the edge, on the black frame lines of the map; inside the map > image I couldn't detect a single fault. > > I think 'useful in Hugin' is a gross understatement. I haven't yet > unleashed it on a 'real' panorama, but I sure will soon and report on > that as well, and it looks like the others who have tested it have had > good results, too. > > Thanks > Kay >
So have I. The only thing I found, was a crash of nona due to
some "errors?" in resulting EXIF data in morphed images.
(It does not happen on every project, so I do not wonder that nobody else had
seen it)
In my work-flow I had to add a call to exiftool "-all= " on each image to
overcome the problem.
And yes, I am impressed too, especially when using together with "Warped
Overlap Analysis".
Kornel
signature.asc
Description: This is a digitally signed message part.
