Hey everyone, Thanks for the input. Looks like I'll have to go with manually adding control points. Since there are 100s of photos to overlay, every click counts and I'll be trying to really reduce the time for each photo by streamlining the process as much as possible. Originally I had thought a large amount of control points would be required for each pair, in which case I wasn't sure the process would become much more efficient (the photoshop process is very action oriented at it is). All I really want to do though is shift, rotate, and scale though, so like Bruno said, only 2 points are needed.
Thanks, Chris PS btw David, we are generally interested in just the images pairs, rather than a blended version, but I love the gif you put together (although I can't look at it for more than 10 seconds ;) ). On Jan 26, 10:20 am, David Haberthür <[email protected]> wrote: > On 26.01.2010, at 16:01, Tom Sharpless wrote: > > > > >>> I am involved in a project that repeats history survey images in the > >>> Canadian Rockies (www.mountainlegacy.ca). > >>> part of the overlay > >>> process is to scale the historic image down, center (which is usually > >>> off), and rotate so the images match as much as possible. Our current > >>> workflow involves manually executing these procedures in photoshop for > >>> each pair. However, I believe that Hugin/panotools has some real > >>> potential in automating this process. The key though is automatic > >>> control point detection. I've struggled to get this to work between > >>> the images. > > >> I haven't run any tests with your photos, but I suspect that the > >> existing automatic control point generators are extremely sensitive to > >> lighting and texture variations between shots. > > >> Have you tried manually adding control points? This is very fast and > >> you only need two control points to get an equivalent to your existing > >> Photoshop workflow. With more control points you can correct for lens > >> distortions and even characterise the original equipment - which would > >> be useful if records are incomplete. > > > I would agree. Setting a few control points by hand is easy -- far > > easier than matching rotations in Photoshop -- and will likely work > > much better since you can select reliable features, along the ridge > > lines, for example. The automatic CP finders are very likely to miss > > those, and give you large numbers of spurious matches too. > > Hey Chris et al. > Indeed, with a little fiddling with the lens parameters I was able to stitch > the two pictures in less than 10 minutes from start to end. > You can find my results inhttp://habi.gna.ch/tmp/mountain/For size-reasons > I've converted all .tiffs to .jpgs. Nonetheless the images are in the > multi-MB range, so keep an eye on that when you click the links below... > > The .pto file shows how I did it (BTW using Harrys 2010.1.0-svn4933 build), > setting some manual control points and optimizing for position and view. > > The resulting panorama is not too impressive > (http://habi.gna.ch/tmp/mountain/Mountain_fused.jpg), so in the end I > averaged the resulting layers > (http://habi.gna.ch/tmp/mountain/mountain0000.jpgandhttp://habi.gna.ch/tmp/mountain/mountain0001.jpg) > using ImageJ, which gavehttp://habi.gna.ch/tmp/mountain/Mountain.jpgas a > result. I think this should be not too bad to work with. > > For fun (and for a nauseating effect) I also saved the results as an animated > gif, which you can see herehttp://habi.gna.ch/tmp/mountain/Mountain.gif > > So as Tom and Bruno mentioned, use manual control points, and then you should > be able to semi-automate your process fairly easy. > > Cheers > Habi -- You received this message because you are subscribed to the Google Groups "hugin and other free panoramic software" group. A list of frequently asked questions is available at: http://wiki.panotools.org/Hugin_FAQ To post to this group, send email to [email protected] To unsubscribe from this group, send email to [email protected] For more options, visit this group at http://groups.google.com/group/hugin-ptx
