So how are panoramas of whole streetlines (or streetviews, street facades, street architecture etc) created? Unlike normal panoramas that are shot from one viewpoint, our streetlines are normally shot from many different viewpoints, along a street, hence we create linear or multi-viewpoint panoramas (alos called multi-perspective). This means, no programm can automatically put together and stitch these panoramas, as there are a lot of parallax errors. For example if a tree is covering a part of the street or a high building is standing behind a low building, then there are problems when you try to stitch the facade together. You will miss parts of the tree, but cannot remove it completely. Or the background building will be seen again and again from different viewpoints, making it visible several times or as a much wider building than it actually is. A lot more small problems come up in the details of the buildings and the architecture. Hence the process of correcting the perspective of the source images, scaling them to fit together and stitching them is mainly manual, which makes the process rather time consuming.
In our blog we wrote a summary about the differences between classic 360° panoramas and linear panoramas. Also there is a (german) example shown on our Leipziger Ring project page.
However, in rare cases you have enough space to shoot a street from one viewpoint, like on large squares. Then the stitching is comparably easy by using a program with an automated solution.
