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.
i realize this isn't easy, especially after serveral attempts, including taking many sets of photos. :) i had a feeling it would be a little easier, though, to be honest. after looking at several sets of photos i have taken with this setup, i think i've got it dialed in parallax wise within a few pixels between shots, not sure how to get it better. i am also using the longest lens i have available, i will likely rent a 15mm fisheye soon and try with that. i forgot to add that i am not using a cropped body. 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 this produced so many images it was difficult to work with and obviously very time consuming. how much overlap should there be to get a good set of 'automatic' control points as a starting point? thanks! slate On May 25, 6:04 pm, Gerry Patterson <[email protected]> wrote: > Hello, > > This may seem a silly question, but what parameters are you optimizing? > Since you have a full 360 around, you should also be able to optimize 'v' as > well, which might help. > > Keep in mind, what you are doing is hard. :-) Taking a full spherical pano > inside a room with that long of a lens means you have to be dead-on with > your measurements to avoid paralax. Much more difficult that stitching a > pano of objects far, far away. (I had problems like this too, it just took > me a while to understand what I needed to try to fix it. Once I got the > hang of it, it wasn't bad). > > Based on what you have said in your e-mail and my assumption that you are > working with a cropped body, your field of view in portrait mode (you are > using a nodal ninja which mounts the camera vertical) is about 34 degrees. > 12 images around doesn't give you much overlap (14%). I may be wrong here > but am guessing all of your control points are near the edges of the > images. Can you switch to a different detent disc to give you more images > and increase the overlap? This way you can have a more widely spaced set of > control points to pass to the optimzer. It has been said here that it isn't > the quantity of the control points, its the quality. > > For the time being I would also consider dropping down to 8bit tiffs until > you get this sorted out. > > Best Regards, > > - Gerry > > On Mon, May 25, 2009 at 7:37 PM, slaterson > <[email protected]>wrote: > > > > > i recently got a pano head (a nodal ninja) and i'm having all sorts of > > trouble with seams in my panos. i have simplified my image set in an > > attempt to isolate the problem, not that i think i have the problem > > identified i'm not sure how to fix it. > > > i took a set of photos to make a 360x180 of my living room. the > > photos were taken at a focal length of 24mm, 30 degree steps around > > the vertical axis and 30 degrees around the horizontal, for a total of > > 60 images. when stitching the entire set i was getting a lot of > > misalignment and seams in the full 360x180 pano. i tried to fix this > > by working with the control points, but ended up going in circles > > (quite literally!). i then decided to look at only the 12 images that > > make up the 'level' cylindrical pano. when adding control points, i > > noticed i was having the same problem, but less pronounced. my next > > step was to delete all control points between the first and last > > images in the 12 image set. this produced a perfectly aligned set of > > 12 images, however the first and last images were obviously not > > aligned and major errors were present between those two images. if > > control points are created between these two images, the seams re- > > appear between previously 'perfectly' aligned images. the seams have > > proven very difficult to minimize, let alone remove entirely. > > > i have carefully found the nodal point for the lens i am using and > > can't see any visible parallax across the images. i am using 16-bit > > tiffs from ufraw, they are pretty big (about 73 megs each, 810 megs > > for all 12), so its might not be feasible to post them. i can re- > > process them as 8-bit (about 400 megs for all 12). > > > i'm using hugin 0.8 and autopano-sift-c from svn. > > > is this normal? am i doing something wrong or missing a step? > > > thanks for any help! --~--~---------~--~----~------------~-------~--~----~ 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 -~----------~----~----~----~------~----~------~--~---
