On Wed, 06 Aug 2014 08:24:25 +1000, Terry Duell wrote: > Hello Robert, > > On Tue, 05 Aug 2014 22:07:49 +1000, Robert Krawitz <[email protected]> > wrote: > >> On Tue, 05 Aug 2014 12:34:31 +1000, Terry Duell wrote: > > [snip] > >>> OK. Can you provide a .pto from a very early stage, i.e. prior to doing >>> any refinement, say just after generating control points for all images >>> and no optimisation steps performed? >> >> Here it is. I presume that control point generation is deterministic. > > Using this .pto (almanor-16-bad-1.pto), if I deselect images > 1,3,5,7,9,11,12 I can then optimise everything without all hell breaking > loose. I optimised in steps, positions, positions and view, positions and > barrel, positions view and barrel, then everything. Are you able to > reproduce this behaviour?
It still eventually happens, after enough iterations of removing bad points, reoptimizing everything but translation, and fine tuning all points. In this case, when I got the max error under 2 (average .6, max 1.7), the problem showed up again. This is about when I'd typically expect it to show. I've attached it at the point it shows up. > Whilst this might not be an acceptable workaround for you, the mean error > is 3.1 pix, max 12.4, it is still considered a good fit. > It doesn't solve the basic problem, but maybe adds a bit more info to help > track down the cause. It would be helpful if there were a way to trace every operation I perform. My description isn't very specific, but it's very hard to be precise when there are thousands of control points and all that. I don't consider max 12.4 to be a good fit at all. At that level, there's usually some visible discontinuities in the background, where they're most objectionable (IMO). Maybe I'm too finicky, but I like to get it under 1. -- Robert Krawitz <[email protected]> MIT VI-3 1987 - Congrats MIT Engineers 6 straight men's hoops tourney Tall Clubs International -- http://www.tall.org/ or 1-888-IM-TALL-2 Member of the League for Programming Freedom -- http://ProgFree.org Project lead for Gutenprint -- http://gimp-print.sourceforge.net "Linux doesn't dictate how I work, I dictate how Linux works." --Eric Crampton -- 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/201408052249.s75MngRt003128%40dsl092-065-009.bos1.dsl.speakeasy.net. For more options, visit https://groups.google.com/d/optout.
almanor-16-bad-2.pto
Description: Binary data
