One of the many technical problems in the Firehouse pictures is the control of reflections, specifically those of the camera and tripod. There aren’t any of me because I’m using a self-timer and getting out of range before it goes off. I can deal with the reflections in the highly convex shiny bits in Photoshop, but the diffuse reflections from flat surfaces are more problematical. I have considered building a blind, and I still may go there, but I’m resisting the hassle in setting it up, and the greater hassle in positioning the camera while behind it.
There’s an old view camera technique that is a possible solution, that’s usually illustrated by a seemingly straight-on photograph of a mirror with no camera in sight. The trick is to shift the lens to the right or left, so that the camera is not directly in front of the part of the mirror in the picture, but off to the side. It’s possible to do that with a tilt/shift lens on the a7R. However, those lenses are thin on the ground, and there are none that I know of that are of the quality of the Otus 55, or even of the Zony 55. And then there’s the fact that, once you’ve made a big shift, you’re no longer using the best part of the lens’ imaging circle. While contemplating those things, It occurred to me that shooting from the side with a regular lens and correcting the perspective in post might give me better quality.
A traditional advantage of shifting the lens vs fixing the image later is that you can get the plane of focus to be parallel to the subject with a shift lens. However, my subject plane is too bumpy to get it in focus in one shot, and I’m having to do focus-stacking to deal with that. Since I’m focus-stacking anyway, why not make a few extra exposures and use the focus-stacking software to sort out the effects of my plane of focus being skewed with respect to the nominal subject plane?
Sounds like a plan, or at least a plan worth testing. I’ll still have to deal with the loss in quality caused by Photoshop’s interpolation of the image as it does the perspective correction, but I suspect I’ll be better off than using a poorer-quality tilt/shift lens at the edges of its image circle.
[Added at 12:30]
I’ve completed a preliminary test that indicates to me that the image quality loss from perspective correction in post, at least for the modest corrections required to keep the camera out of the reflection with a 55 mm lens, are modest enough to be difficult to reliably measure.