I want to make a universal item animation blueprint that houses all of my item animations but does not require all to be loaded at a given time. I think anim layer interfaces would be useful for this, however, I am finding that these are not applicable to child animation blueprints which means I cannot just plug in a single state machine in the parent and then modify the animations for each state in the child blueprints and override the interface function. As a result, I am considering having a singular state machine and including five interface functions in my anim layer interface. This way each will control one of my five movement states (idle, idle + aiming, walking, walking + Aiming, and running) and the states can be altered by linking different ABPs (each of which can be loaded from soft references before activation). This will essentially mean I am making animation blueprints exclusively to store the 5 desired animations and nothing else and will have 5 anim layer interface functions in my main state machine. Is this an inappropriate use of anim layer interfaces? Will this be really inefficient? Is there a better way to incorporate layer interfaces?
All animations are loaded to memory at runtime. Regardless of current use. If it is referenced anywhere in code it will be loaded to memory when the owning Actor/Pawn is loaded. This prevents loading stalls and hitches when the anim is called to run. Already in memory and ready.
Correct me if this explanation is wrong but what I was thinking is that I would have a basic animation blueprint for my character that has a handful of basic animations for things like running, walking, jumping, etc. These animations will always be loaded into memory. The idea I was trying to ask about was the possibility of plugging each of these animations into a anim link interface, and then asynchronously loading and linking another animation blueprint (which is not associated with anything directly but holds another set of animations for a given item pickup). From my understanding, this would let me override my basic animations with the animations for the new item. If I were to then swap this item for a different one, I could unlink the current one (so those animations are no longer held in memory) and link the new one so only my default animations and these new ones are loaded. My main point of uncertainty (assuming that explanation is correct) is whether or not this is a less efficient way that just loading a different animation blueprint when needed. The main benefit I was hoping for is that it seems like animation layer interface linking is smoother visually and would make my main animation blueprint generally less confusing to work with.
Any graph, interface, layer that includes animations would have those anims loaded to memory when the pawn/SK that uses or would “Potentially” use them is loaded. Regardless if the graphs, interfaces, layers etc are being used at the time.
Linked Graphs, Layers and Interfaces can be used to modularize animation complexity.