5
Replies
From your screenshot I conclude that Upright has applied the strongest correction it is designed to, bur this has not quire removed the convergence altogether. I don't consider the function "useless" just because there's a sensible limit on how strong a correction it will apply.
As far as I can tell, the Upright tool is intended for quite moderate corrections that will not cause very severe distortions of proportion and of shape.
LrC can achieve a stronger correction by combining with the Manual transform sliders (below the Upright options in the transform panel), also by using a Guided upright transform instead. The attached screenshot shows one of my own photos where I have forced a severe verticals correction, and illustrates I hope how aesthetically odd the results can be..
A different photo would have needed to be taken that did not require such extreme treatment, or else it should be accepted that in this case verticals are not sensibly going to show parallel.
Strong vertical convergence is part of the character and drama of some photos, and in any case consitutes their genuine perspective. This happens inherently when they were taken with the camera aimed significantly upward or downward. The way to avoid it is to take the photo from a different vantage point or at least framed differently whihc may require a wider angle capture. The oddness doesn't go away when a physical shift-lens or pano stitching method is used, if the geometry of the representation is still unchanged.
Eliminating that verticals convergence artificially, IOW 'pretending' that the photo was taken from a different viewpoint than it really was, is not guaranteed to be pictorially successful. At the least, cues within the photo such as subject proportions or even which parts of the scene are visible / not visible, will make this manipulation very apparent.
this is what Upright did on its own with my image - without the additional manual correction needed to achieve parallel verticals. Looks to me as if the angle of the white corners is about the same as in your own first example. I believe that in both cases Upright has calculated a correction that proved larger than the maximum it was designed to automatically apply. So your only option would be to manually intervene. And looking at even this less extreme correction, you can still see adverse effects: in the squatness of proportion generally, and in the shape changes that increase noticeably towards the image corners.
So, I don't think your LrC is behaving unusually. Perhaps you could show an example with less 'necessity' for correction and see how it does there. As I say: in cases such as my own example - either distinctive and IMO ugly pictorial effects are inevitable, or else something of a natural converging perspective should be accepted.
AdChoices
