I don’t really use social media anymore (or my phone, for that matter), so I’ll just use this two-impressions-a-month corner of the internet to quickly say: I have been and continue to be in awe of the talent and collective drive of all the people involved in Visual Intelligence (and Siri as a whole) here at Apple. This…

In the previous write-up , I mentioned briefly that changes in POV were handled by a specific entry in the transformation matrix, m34. I noted how the derivation was a whole separate thing in and of itself, so here’s the overview of how we are able to understand why this entry affects perspective!

When it comes to transforming a view, one can think of it as applying a calculation to each individual point of the view’s layer, such that for every (x, y, z), we obtain a new (x', y', z'). That calculation is actually a multiplication of the coordinates (x, y, z) by a matrix (good ol' linear algebra). How we construct our matrix is through the use of various types of CATransform3Ds, which we’ll now dive into.