Alright, let's assume that a file has 2000 instances of \x0A, and 10 of those instances are directly preceded by \x0D. If any of the \x0D bits are deleted the file won't parse. If you deleted all 10 instances of \x0D from the file which are followed by \x0A would it be possible to put them back?
Some things that may help:
If the first instance of \x0D is found, the image (.gif or .png) will be partially visible, if your further find the second position, more of the image is revealed and so on, but it must be found in order (top to bottom).
In other words, you could technically repair the file by placing \x0d before each instance of \x0A in the file until a block is solved and more of the image becomes visible, then move on to the next block, and so on until all \x0D's are placed.
However, it would take time to test each position, meaning for this particular image if you kept a good pace and did it by hand, you could solve it in about 6 hours. Is there any way to write a script to do this?
This would enable recovery of .png and animated .gif files which were corrupted by being transferred in ascii mode.