I agree that straight line CPs are extremely important. When I stitched a large mosaic (135 images with an 8mm fisheye, ~1800 manual CPs), I used a pre-calibrated lens and optimized position and translation after every image [1]. At first, I started off without using straight line CPs, and it did not turn out well; adding straight line CPs, however, did wonders. After, I still had to do a lot of tweaking and masking to get better alignment, but the straight line CPs were crucial in maintaining the overall shape of the mosaic.
Matthew [1] http://www.mpetroff.net/files/openzoom/taft-class-of-2011/ On Jan 14, 12:50 pm, Tom Sharpless <[email protected]> wrote: > Hi Kay > > In my experience the most important things are to place lots of > straight line CPs, and to rigorously exclude both off-plane and bogus > CPs. Hugins CPFind can generate a lot of false matches when matching > 'all vs all', so I try to organize my shots so that 'multirow' mode > works. But in any case I check every CP visually. > > Straight line CPs are the 'backbone' that keeps the whole pano > square. They only work when the target projection is rectilinear, and > at first the optimizer wants to generate yaw angles that are too big. > So I have to reset and reoptimize from scratch a lot -- just positions > and viewpoint, with nominal lens parameters. Eventually when the CP > set is good, the pano snaps into alignment and I can then tweak lens > parameters if it seems necessary -- though with a calibrated lens it > rarely does. > > Regards, Tom > > On Jan 13, 5:05 am, kfj <[email protected]> wrote: > > > > > > > > > On 12 Jan., 18:48, Tom Sharpless <[email protected]> wrote: > > > > I've been happily making high resolution mosaics of wall paintings > > > with Hugin. > > > > It's a bit chancy getting the very wide ones to align right, but with > > > plenty of straight line CPs and care to avoid CPs not on the plane of > > > the mural, I've had good results. In some cases I've had to add one > > > image at a time to the optimized pool, presetting y,p,r,x,y,z to > > > reasonably close values. > > > I've done a fair amount of similar stitches (mainly maps and > > signposts), and I found that trying to optimize y,p,r,X,Y and Z all at > > once introduces to many degrees of freedom, so the whole thing totally > > fails. But you can still work on the whole image set at once if you > > only only do a few parameters at a time. I often start out with just X > > and Y and then take it from there, optimizing the others in turn. Once > > everything has 'gelled' nicely, you can optimize more parameters > > simultanously. Doing it this way is probably faster than introducing > > the images one by one and estimating parameters. The process also > > benefits from lots of CPs (like, 100-200 per pair). > > > This makes me wonder if there couldn't be a specific mosaic > > optimization mode which follows such a pattern, thus avoiding the mess > > that usually happens if you try optimizing everything at once. > > > Kay -- 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
