"Stitched" image, big galaxy missed?

Along the S edge, no g band data (maybe no r band either?).

Running horizontally ~four thin strips, almost lines; mostly red:

I guess these are where tiles/bricks overlap, or slightly underlap.

Somehow the model keeps these linear features:

But it deletes a well-defined galaxy, and adds a large, diffuse streak, not seen in the data:

How did this come about?