Some time has passed and nobody has disputed my and jarlmai's calculations that the object must be at a MINIMUM 1 meter in width.
Here are the conclusions (REDACTED to compress see original posts for full details)
So Max approximate size is several meters across and most probably no more than 5 meters across.
Counting pixels the object appears approx 30x32 pixels. So basically 1x1 meter or 5x5 meters is the maximum range approximately.
Typical batman balloon sizes (I initially found 45x70cm but can't find that format anymore):
71cm w x 69cm h https://www.instaballoons.com/products/batman-cape-shape-28-mylar-foil-balloon
69cm x 99cm https://rakkoonsballoons.com/product/supershape-balloon-batman-action/ (different design)
From 1 meter to 70cm we would need a big (but not impossible) error given our approximations.
I wouldn't exactly dispute your conclusions, because I don't understand either cameras or geometry well enough to follow your calculations. As the metadata about the photo seemed to exclude the use of zoom, I reapplied my own crude method, using 65.5 degrees as the FoV, and got results very similar to yours. I didn't think it was worth commenting again, and threw my scribbles away. But my calculations relied heavily on the assumption that your estimates of viewing distance were correct, at least for the upper and lower limits. Before anyone starts quoting your conclusions as definitive I think the underlying method and assumptions need to be reviewed. I'm bound to say I start from the strong presumption that the object is a Batman balloon, since it looks exactly like one, and it doesn't look much like anything else. There are literally many points of resemblance! So my question is: are your findings robust enough to overturn this presumption? For example, how sensitive are they to slight variations in the estimate of the viewing angle? And how are they affected by the actual change in direction of view between the two photos, as shown by the changing position of the pilot's helmet, etc?