Considering the image at least as uploaded here is 960x660 and assuming the top panel is about 1/4 the height that means each picture is about 768x330 or 253,440 pixels which is way less than the claimed differences. But if you convert it into bytes at 160kb it would be approximately 1,338,163 per side and if you convert it into raw 1s and 0s that gives a much more believable 4,282,122 bits meaning there’s about 550,000 bits that are the same which I would think would be the real challenge of finding.
Considering the image at least as uploaded here is 960x660 and assuming the top panel is about 1/4 the height that means each picture is about 768x330 or 253,440 pixels which is way less than the claimed differences. But if you convert it into bytes at 160kb it would be approximately 1,338,163 per side and if you convert it into raw 1s and 0s that gives a much more believable 4,282,122 bits meaning there’s about 550,000 bits that are the same which I would think would be the real challenge of finding.
…and now I’m late for work
I think you could do it with a bitwise xor. Any pixel that is not zero should be different.
Each image is 480 x 480, I just put it in paint and trimmed accordingly. 230,400 pixels, so you were pretty close.