I think there is quite a common misconception about the "splitting of universes". From Everett own work we referes to an object not being described by the wave funciton as a superposition of states, but rather that the object IS the superposition of all its states. That is to say each state is like a different aspect of the photon. Now the concept of wave function collapse becomes a subjective and not a objective experience. Let our screen also be a superposition of all its possible states (whatever these might be).
The premisse is that each state of the poton "correlates" with each state of the screen. For example consider the photon in a superposition of states ##|x>##, with respective measure ##a(x)## spanning some interval ##x \in [x_0, x_1]##, where x is here translation along the screen from the centre. Then the states of the photon correlates with the screen such that the state of the whole system (which in Everett's theory is the actual object itself) is now each state of the photon "multiplied" by each state of the screen. Consider then the case where there is only one state of the screen, then it will appear as if the photon impacts one place on the screen, an impact at point d is just the state ##|d>## "corrolated" with the 1 state of the screen. However, all the other states ##|x>## for ##x\neq d## also corrolate with the screen and are seen as the photon impacting only at the point ##x \neq d## for each respective x. The probability that you will observe the photon in state ##|x>## is then ##|a(x)|^2##.
The term "splitting" was introduced to say that although the whole object is still all the states of the photon "multiplied" by all the states of the screen (i.e. the photon hitting every point in the screen). We subjectively only experience one colission, because that happens to be the combined state of the now corrolated photon and screen where we find ourselves. Yet that is not to say that there are not other corrolated states of the photon and screen that exist. In other words, every time you observe aquantum even there is a different you observing every single possible outcome. While this might sound farfetched it is one of the few interpretations that are self-consistent and don't have wavefuncitons "collapsing". Also as any object can corrolate with any other object this means that even if a photon interacts with an electron, the electron is an observer for the photon and the photon is an observer for the election. This then removes the paradox of what is an observer and where the quantum realm begins and ends.
Another example which might be easier to consider is the interaction of two electron both described by ##|\psi_i>=\frac{1}{\sqrt{2}}|up_i>+\frac{1}{\sqrt{2}}|down_i>## where up and down denote spin up and spin down states and ##i \in {1,2}##. When these two states become corrolated the resulting system is ##|\psi_1> \otimes |\psi_2>=\left[\frac{1}{\sqrt{2}}|up_1>+\frac{1}{\sqrt{2}}|down_1>\right] \otimes \left[\frac{1}{\sqrt{2}}|up_2>+\frac{1}{\sqrt{2}}|down_2>\right]##. This then results in 4 different possible events (all equally as likely) and all physically real according to MWT. If we "are" electron 2, then we may either find ourselves in spin up, observing electron 1 with spin up or spin down; or find ourselves in spin down, observing electron 1 with spin up or spin down. Yet there is no mention of splitting anywhere, this only comes in in an attempt to visualise the interaction.
The MWT by Everett arrises naturally if you only consider probability theory and then apply it to quantum mechanics, as was his method. The whole "splitting" term is misleading and was only introduced to describe states of a system that will forever be inaccessible to us. Visually it might help people understand what is happening, but it is just that, a visual aid and doesn't actually imply that the universe has "multiplied" and that there now are as many times the energy of the universe as there are possible ways to corrolate the states of two systems.
I hope this helps you understand the concept a bit better, Everett's work is facinating and self consistent, but is unfortunately often talked away or dismissed as noncense by people who have not actually taken the time to go through his paper and follow the mathematics which lead only to the conclusion that all possible outcomes are real, yet we can only observe one!