On Sunday, 14 August 2011 at 15:11:17 -0400, Yuval Levy wrote: Sorry, just got round to reading this thread.
> On August 14, 2011 05:47:43 am kfj wrote: > >> If, of course, we'd find that in certain use scenarios one definitely >> outshines the other, this would be helpful. Not that I expect this to >> happen > > I also do not expect this to happen, but if we don't collect > information we won't find out. And it is often such borderline > cases that help the next iteration of development that eventually > brings about a third generation tool that is superior to both the > first and second generation. FWIW, I have a scenario where I find that the control point generators generate very different control points from the ones I choose myself when I'm doing it manually. Every week I make a number of panoramas, the most complicated of which is at http://tinyurl.com/3mcw2bb This is a two-row 360° panorama made out of 24 images, each tone-mapped from three images. The subject has both straight lines and corners (timber) and soft transitions (leaves and other vegetation). The leaves can move both between individual tone-mapped images and also between the background images for each tone-mapped images. The timber stays where it is. But all the CP generators I have tried go for the leaves. That's a particular problem when it's windy. When I do it manually, I choose a spot or a corner on the timber, and things work much better. Clearly there's a reason why the CP generators make this kind of choice, but I don't understand it. Are there some technical reasons why it's done this way? Or are there generators that do choose hard surfaces? Greg -- Sent from my desktop computer Finger g...@freebsd.org for PGP public key. See complete headers for address and phone numbers. This message is digitally signed. If your Microsoft MUA reports problems, please read http://tinyurl.com/broken-mua
pgpPojzvrlToX.pgp
Description: PGP signature