A practical approach is to estimate the original camera’s ground position first, then refine by flying.

Use fixed landmarks that appear in both the old photo and a current map. On the old photo, draw straight lines through pairs of landmarks that should line up vertically from the camera’s viewpoint. Transfer those same landmark alignments onto the map; where several such lines intersect gives an estimate of the camera’s optical axis/position over the ground. Using more landmark pairs improves accuracy.

This method helps you find horizontal position, but not altitude. For height, you’ll still need trial and error: raise the drone until the relative size and spacing of features starts to match the old image.

To truly match perspective, you must also match field of view. If the original was taken from higher up than your drone can practically match with its lens, the image may not look identical straight out of camera. In that case, shoot from the best matching position and crop the drone image to mimic the older photo’s framing.

So: estimate location with landmark line intersections, then fine-tune altitude, angle, and framing in flight.