On Friday, April 18, 2014 9:10:02 PM UTC+1, Donald Johnston wrote: > > I have two overlapping images of a building that I'm creating an > architectural rending from. I auto generated the control pints and then > added vertical and horizontal line control points. In the overlapping area > I put the same vertical line in both images. It did a great job of squaring > up the building. > My question. Did I have to put the same vertical line on each image or > would it have been better to put it on only one and let all the other CPs > straighten the other image to match? >
Levelling a panorama is usually accomplished with t1 (vertical line) control points. Provided the images have been accurately aligned by the optimizer and the lens parameters are good, only two t1 control point pairs (hereafter referred to simply as two t1 points) are needed to level the panorama perfectly. I.e. two horizontally separated vertical features need to be marked with a t1 point on each. The two vertical features must not be separated by exactly 180 degrees of yaw otherwise the horizon will not necessarily be levelled even though the marked verticals will be correctly brought into the upright position. Assigning three t1 points on three vertical features is guaranteed to work since they cannot all be separated by 180 degrees of yaw. Assigning more than the minimum required number of t1 points can be useful to assist in the evaluation of the lens parameters. For example, to correct barrel distortion you can apply one t1 point on the top half of a bowed vertical edge and a second t1 point on the lower half (or the second one on the full extremities of the bowed edge). Marking several vertical features can also be useful to obtain an averaging effect when there is some doubt about which vertical features are actually vertical (if any). Lamp posts are somewhat variable in this respect. When doing this, it's probably best to exclude the already optimized lens parameters from the levelling optimization in order to avoid the lens parameters being skewed in an unreasonable attempt to get all the not quite vertical features simultaneously vertical. As to the question about which image to use when a vertical feature appears in two overlapping images: it doesn't really matter provided that the images have already been nicely aligned with normal control points. Also, it's ok to assign a single t1 point on the top and bottom of a single vertical feature that extends over two or more images in different rows. I.e. a single t1 control point pair need not be confined to a single image. You assign the t1 point to the vertical feature with little regard for which image(s) it happens to appear in. John -- 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/76957af9-040a-40f8-ad5f-a9920f277ce4%40googlegroups.com. For more options, visit https://groups.google.com/d/optout.
