Re: [hugin-ptx] cpfind - max matches?
On August 30, 2011 05:29:41 AM Jeffrey Martin wrote: > Sorry Yuv but this is a completely useless strategy, either you get too > many points or not enough. Sorry Jeffrey. 1 - do not shoot the messenger and 2 - no need to apologize for the obvious. You asked if there is a way to limit the number of CP's in cpfind and I simply reported what cpfind has to offer. > there is no way to specify that you want 50 points but not 5000? There are ways - added on top of cpfind. Bruno mentioned the python script to reduce quantity. There are other tools to prune CPs such as celeste and cpclean. The overall mix / strategy is difficult to use, I agree with you. But I would not qualify it as "completely useless". It is the glue between the tools that can be improved; and the python scripting is a major opportunity in that sense. The real question is: Which points do you want to keep? Others have chimed into the discussion and eventually it may lead somewhere. There is something IMHO fundamentally wrong in the assumption underlying recent development of Hugin, but there is light at the end of the tunnel: On September 3, 2011 02:48:05 AM Andreas Metzler wrote: > Bruno Postle wrote: > > Though my question is, why do you want to reduce the number of > > control points? If it is to workaround some bug in other tools, > > then the problem isn't really with cpfind. > > > cpfind is not perfect. Manually filtering out bad control points is > often necessary for me. I rather search for the 10 out of 100 bad > control points than 50 out of 500. Nailed! Yuv signature.asc Description: This is a digitally signed message part.
Re: [hugin-ptx] cpfind - max matches?
On Tue 30-Aug-2011 at 02:29 -0700, Jeffrey Martin wrote: Sorry Yuv but this is a completely useless strategy, either you get too many points or not enough. there is no way to specify that you want 50 points but not 5000? That seems like a significant limitation. Recent Hugin snapshots with python enabled have a 'action' menu item called 'Keep 5 control points per image pair', this seems to do exactly what you expect. Editing it to change 5 to 50 is very easy. Though my question is, why do you want to reduce the number of control points? If it is to workaround some bug in other tools, then the problem isn't really with cpfind. -- Bruno -- 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
Re: [hugin-ptx] cpfind - max matches?
I think these CP strategies are interesting. APSC tended to give me 5 points in a corner a bit too often. So CPfind seems to be the way forward even if it will need some more options. It should be fairly straghtforward to limit the number of CP's per image pair by some more or less clever strategy to throw away CP. My question on this discussion is: Are the buckets here referring to "feature points" in each image, or the actually paired feature points that become Control Points between images? To me it seems like the bucket-model is a good way to distribute feature-points, but actual Controll Point distribution depend utimately on actual overlap. And I would like the CPfind to find thousands of feature-points in order to eventually match tens of control points. Cheers /O 2011/8/30 Jeffrey Martin <360cit...@gmail.com> > Sorry Yuv but this is a completely useless strategy, either you get too > many points or not enough. there is no way to specify that you want 50 > points but not 5000? That seems like a significant limitation. I wish I > could do something besides point out this problem :( > > > > > On Monday, August 29, 2011 2:52:55 AM UTC+2, Yuv wrote: >> >> On August 22, 2011 11:55:17 AM Jeffrey Martin wrote: >> > Is there a way to limit the number of CP's as a-s-c does ? >> >> First, you can limit the max number of points per bucket with --sieve1size >> >> CPfind partitions the images into a grid of buckets with --sieve1width and >> >> --sieve1height. Both have a default value of 10, which means that each >> image is sliced into 10 rows and 10 columns, yielding 100 buckets. >> >> --sieve1size is the maximum number of points to be added for each bucket. >> >> Of course no points will be added to a bucket if it is not in an overlap >> area. >> >> So there is no way to limit the number of CP's in absolute terms as APSC >> does, but modulating the three parameters should help. >> >> If after reducing --sieve1size there are still too many points, try >> decreasing --sieve1width and --sieve1height >> >> >> High values of --sieve1width and --sieve1height are recommended for small >> overlaps (because more buckets fall into the overlap area). >> >> HTH >> Yuv >> >> -- > 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 > -- /O -- 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
Re: [hugin-ptx] cpfind - max matches?
Sorry Yuv but this is a completely useless strategy, either you get too many points or not enough. there is no way to specify that you want 50 points but not 5000? That seems like a significant limitation. I wish I could do something besides point out this problem :( On Monday, August 29, 2011 2:52:55 AM UTC+2, Yuv wrote: > > On August 22, 2011 11:55:17 AM Jeffrey Martin wrote: > > Is there a way to limit the number of CP's as a-s-c does ? > > First, you can limit the max number of points per bucket with --sieve1size > > CPfind partitions the images into a grid of buckets with --sieve1width and > --sieve1height. Both have a default value of 10, which means that each > image is sliced into 10 rows and 10 columns, yielding 100 buckets. > > --sieve1size is the maximum number of points to be added for each bucket. > Of course no points will be added to a bucket if it is not in an overlap > area. > > So there is no way to limit the number of CP's in absolute terms as APSC > does, but modulating the three parameters should help. > > If after reducing --sieve1size there are still too many points, try > decreasing --sieve1width and --sieve1height > > > High values of --sieve1width and --sieve1height are recommended for small > overlaps (because more buckets fall into the overlap area). > > HTH > Yuv > > -- 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
Re: [hugin-ptx] cpfind - max matches?
On August 22, 2011 11:55:17 AM Jeffrey Martin wrote: > Is there a way to limit the number of CP's as a-s-c does ? First, you can limit the max number of points per bucket with --sieve1size CPfind partitions the images into a grid of buckets with --sieve1width and --sieve1height. Both have a default value of 10, which means that each image is sliced into 10 rows and 10 columns, yielding 100 buckets. --sieve1size is the maximum number of points to be added for each bucket. Of course no points will be added to a bucket if it is not in an overlap area. So there is no way to limit the number of CP's in absolute terms as APSC does, but modulating the three parameters should help. If after reducing --sieve1size there are still too many points, try decreasing --sieve1width and --sieve1height High values of --sieve1width and --sieve1height are recommended for small overlaps (because more buckets fall into the overlap area). HTH Yuv signature.asc Description: This is a digitally signed message part.