On Jan 13, 6:55 am, Harry van der Wolf <hvdw...@gmail.com> wrote: > All, > > After Pablo's announcement[0] (3 January) of the almost patent free CP > detector I built it on OSX, but due to other priorities I had set myself, I > did not have the time to do tests (sorry Pablo [1]). > > This is only a mail to share some first test results. I need to do more > tests, so don't expect graphical and statistical test results. > == This mail is also to "invite" other users to test as well == > (but I have to admit that I did not create a portable plugin binary yet to > be delivered with Hugin for OSX). > > (My) General conclusion: Based on this prelimiary results and even as it is > now (but that was Pablo's own comment in his mail[0]), I think it does a > good job and I think it is suitable for use with Hugin but in combination > with celeste and cpclean. It needs some further tweaking. > > Note: I only have partial panorama's based on a lineair lens. I compared 6 > linear lense partial panos. > I have only one 360x180 fish-eye pano set. Yuval "released" a > "hugin_6aroundtilted_testcase" somewhere end of 2007/early 2008. This is my > only but always fisheye 360x180 testcase for hugin builds, including now > with the new CPdetector. > For now I only tested with Pablo's suggested parameters: --grad --sieve1size > 100 --sieve2size 3 -o %o %i > > Results/Remarks: > - The new CPdetector detects/generates a huge amount of Control Points. > Pano's having approx. 30-40 CPs now have approx. 150-180 CPs. Pano's with > approx. 60-70 CP's now have 220-230 CP's. Yuv's pano now generates 354 CP's > (forgot to note the original number for this mail). After running Celeste > and cpclean generally 25-50% of CP's are removed, which still means that you > have a lot of CP's but the optimizer can easily deal with that. > - The spread of CP's over the overlapping areas is OK, but this might also > be due to the huge amount of CPs. Now is this good, having a good spread of > CP's, or is this bad, having so many CPs? > - Both celeste and cpclean should always be run on the result IMO. Most of > the time I run cpclean 2-3 times (but I also do that when using the original > panomatic and/or autopano-sift-c). > - Speed seems to be comparable with the original panomatic, but I didn't do > stopwatch measurements as I don't care about optimal speed right now. It is > definitely perfectly acceptable :-) > > So far, so good. > I'm very interested in test results from others. > Apart from the fact that panomatic and autopano-sift-c do a good job, the > license restrictions are by now a real pain. The sooner we have a > patent-free CPdetector the better. > Hoi, > Harry > > [0]:http://groups.google.com/group/hugin-ptx/browse_thread/thread/d9c0558... > [1]:http://www.mail-archive.com/hugin-ptx@googlegroups.com/msg07614.html
Can someone post the right instructions for downloading the branch? I installed bazaar using MacPorts, and the command: baz branch lp:~pablo.dangelo/hugin/panomatic-lib results in an error: branch: could not determine source revision from directory: /Users/sgaede/development Thanks, --skip
-- 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 hugin-ptx@googlegroups.com To unsubscribe from this group, send email to hugin-ptx+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/hugin-ptx
