initial results are great. i tried optimizing with 'v' and the sample set of 12 images in my example above is stitched perfectly. i am now stitching the full set of 60 images to see how it turns out. if there are seams i will try your suggestions carl.
i'm going to try your suggestion of taking fewer shots, no horizon shots but rather tilted up and down. unfortunately, the furtherest i can tilt the camera up is about 70 degrees (due to the physical size of the lens), luckily 60 degrees is enough to get the zenith. i have noticed that i can exclude about half the shots when p=+60 and p=-60, however I haven't tried not including them for initial control point creation yet. thanks a lot for the help/advice! On May 26, 2:00 am, Carl von Einem <[email protected]> wrote: > Hi, > > >> On Mon, May 25, 2009 at 8:22 PM, slaterson wrote: > > >>> i was optimizing for the y, p & r. i have a, b & c parameters for my > >>> lens, so i wasn't optimizing for them. i will try this again, > >>> optimizing for 'v' as well. > > Make sure you have some vertical CPs in several areas of the room. I'd > try to optimize the horizontal row first and deselect all others. Maybe > in your case I would try to first of all only generate CPs for those > images that were shot with p=0. Only select those in the lens tab and > click the 'generate control points' button. As a first step I mostly > only optimize for y and v, after that I add p, then also r. From time to > time have a look at the CP table to identify bad CPs. Have a look at the > preview to see if some bad CPs lead to freaky results (you could always > reset the miscalculated values in the lens tab if something very strange > happens). As soon as your first row is sort of stable, add the other > frames one row after the other... > > While checking for bad CPs I often enough see areas that are worth > adding some CPs manually. > > Even if you have once calculated a, b and c for your lens: do it again > for this project. I mostly deselect y, p and r and then optimize first v > and b, after that a and c. Next, optimize also for shift (d and e, but > not together). Now again optimize for y and v, after that add p, then > also r. > > It's not really a simplified 'point and shoot' technique but it works > for me. > > >>> i'm using a canon > >>> 5d body, at 24mm, the hfov is 53 degrees, so i think i am ok with > >>> overlap. i initially tried taking shots at 20 degree steps, however > > 9 or 10 images in the center row should be enough to get a 25 to 30 % > overlap. This should suffice for rooms unless they consist of mostly > huge white walls. The upper and lower rows will surely need less frames, > just simulate that in hugin with dummy images (with a frame so you can > see the ca. amount of overlap in the preview). > > Just another idea: try a different matrix of shots that have two rows at > the horizon: one tilted up about 25-30 deg., the other tilted down the > same amount. That way you might need about 4 rows the full sphere: > 9x p=30 deg. > 9x p=-30 deg. > 6x p=80 deg. > 6x p=-80 deg. > -> it should work with just 30 images :-) > > Carl --~--~---------~--~----~------------~-------~--~----~ 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 -~----------~----~----~----~------~----~------~--~---
