 In neuroscience, people want to figure out how the brain works. You want to understand how the information flow inside the circuits. And with modern calcium indicators, you can really capture the neural dynamics and function time. And therefore, you can really see how the neurons think in real animals. The many things happen deep in the brain. But the thing is, when you image deep, your image quality gets worse, because tissue induces distortion to your optical beam. AO is adaptive optics. With AO, you can have high quality environment, but it's restricted to a tiny image field of view. So what we aim to solve with this system is that we want to make the field of view really large, and we want to make it scalable, so you can simultaneously measure a very large region without any constraint. The idea of multi-pupil adaptive optics is really that we wanted a prism array. So this array of indeed tiny prisms, but each of the elements is tilting the beam to a different direction. Now, with this prism array, we have one individual pupil for each of those quadrant. That way, we can handle them independently and simultaneously. And in this way, they can all be optimally corrected at the same time and over multiple regions. And in this work, we use a 3x3 prism array to generate 9 pupils. Of course, this thing is scalable. You can potentially do 4x4 or 5x5 to get 25 regions simultaneously corrected.