On Mon, 2021-12-27 at 22:56 +0000, Bruno Postle wrote: > On Mon, 27 Dec 2021 at 15:22, Jon Schewe <[email protected]> wrote: > > > With a small experiment I managed to get into this state by > > > telling Hugin to optimize X, Y, Z only for the images. It ran the > > > optimize and told me that the average distance between all > > > control points was zero. At this point the project appears to be > > > in an unusable state. > > > Maybe attach a PTO project (no need for the images), it may be > > > obvious what the problem is. > > > > Attached. > > > > > > I needed to change the output projection from cylindrical to > rectilinear, the panorama angle of view to 120 degrees, and the > canvas size to a more reasonable 10000 pixels. The angle of view of > 69 degrees for the input photos seemed credible. > > As already noted, for some reason all the photos were assigned to the > same stack, so I gave them all a separate stack in the photos tab > (I've never seen this problem in a project before). > This seems to happen when I reset the positions and then try and optimize positions. I have been telling Hugin to optimize X, Y, and Z.
> I deleted all the horizontal and vertical control points, these > shouldn't be necessary for a mosaic unless there are a lack of other > features to use for alignment. You may want to reintroduce a handful, > just to straighten things up a bit. > > > I set optimise -> custom parameters in the photos tab, and then in > the optimiser tab optimised just X and Y (except the first anchor > image) to get an initial layout, then I added in the Z parameter for > all images (except the anchor image) and optimised again. At this > point I had a maximum error of about 200 pixels which shows that this > initial layout is more or less ok. > > Then I added in yaw, pitch and roll for all images (I didn't optimise > roll for the anchor, otherwise the whole panorama might spin without > a horizontal or vertical control point). This got the maximum error > down to 120 pixels, so better but not great. > > Then I optimised the angle of view of your lens, which reduced from > 69 degrees to 20 degrees - this is a big difference, if Hugin > detected 69 degrees in the first place then this would be a bug. I never manually set the angle of view, so Hugin detected it this way. > I deleted some obviously bad control points: when a pair of images > has a good spread of control points, but one of them has a much > higher error distance, this is a sign that this point needs deleting. That's interesting. I hadn't been looking at the distances. I had been looking at the images and trying to be very careful to put all of the control points in the right places. > So after all this, the average error is 10 pixels and the maximum > error is 45, this may be the best you can get with this panorama. > > Attached, resulting PTO project. > Thank you for all of that. I just rendered it and it looks beautiful. -- A list of frequently asked questions is available at: http://wiki.panotools.org/Hugin_FAQ --- You received this message because you are subscribed to the Google Groups "hugin and other free panoramic software" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/hugin-ptx/6206b28d2af4c4c364c80f50acc3e49a56735c75.camel%40mtu.net.
