Hard to explain (in part because I'm not a scientist) but it isn't when you look at it, it's when the photon interacts with something. I'm gonna do my best, and if I'm hard to follow, that's because I suck at writing
Before interacting with something, a photon acts like a wave, kind of like a wave in water, or a sound wave. The wave goes through both slits at the same time, which causes it to split into multiple waves. In places where two waves meet, their magnitude is added together. That is, where the peaks of those waves meet, the peak gets higher. Where the valleys meet, they get lower. Where a peak meets a valley, they cancel each other out. The empty parts on the detector are where peaks met valleys, and there was no measurable wave in those parts.
When a photon interacts with something, it collapses from a wave to a particle, and interacts with the detector only in one spot. I've seen it compared to a speck of dust in a raindrop. Before that raindrop hits the ground, you know that the speck of dust is somewhere in the drop, but not where it is in the drop. When it hits the ground, the speck can only end up in one spot. When the wave collapses, the photon is forced to interact with the detector in only one place. The location is random, but is more likely in spots with a larger magnitude of wave. Those are the places with the spots on the detector in the top picture.
If the photon interacts with something at the slits, like a polarizing filter, it collapses before an interference pattern is able to form. No interference pattern means it ends up interacting with the detector in one of the two areas like in the bottom picture.