Page 1 of 1

RMS sesems quite high and other weirdness

PostPosted: Wed Aug 15, 2012 5:01 am
by inMotion
The global RMS is 3.93 for a 15 gigapixel image (1000 photos) shot with a Merlin/Panogear with pappywizard import. While this is not "high", I've shot several ~10 gigapixel panoramas without using a rotating head and got RMS as low as 2 before. I would think that with a rotating head and xml import that it would be fairly close to 1 even. What gives?

It also seems that each image is rotated a bit...some images are rotated quite a bit actually...more than they should be (uneven horizon). There also is an issue with the images not blending together seamlessly, especially in the sky and horizon (see image).

PostPosted: Thu Aug 16, 2012 1:06 am
by inMotion
Can you run neutral-hazer on a 15 gig file or do you have to do each file individually? i've already edited a bunch in photoshop, and don't want to start over.

PostPosted: Thu Aug 16, 2012 1:28 am
by inMotion
also, any idea why the images are spaced unevenly - they were taken with the robotic head, so they should all be of equal space between one anohter.

also, why would they need to be rotated at all in autopano? the rotation seems to be arbitrary.

PostPosted: Mon Aug 20, 2012 12:21 pm
by AlexandreJ
Hi InMotion,
For me, it looks like that there might be some arbitrary rotation on location where no control point are present. We modified the algorithm responsible for them in a way that can handle also cases without exif data. These location are not always hard location, the picture needs to move around if a real image with real cp does move in the neightboorhood. So there is a slight influence of moving picture ( meaning, thare are really optimized ) on image without control points.
Would that explain your case ?