Hey both, I just want to check in regarding the above. I have followed the approach outlined in the link provided by Marcel above. My exact use case is as follows (Marcel, we had a conversation about this privately a few months back so if you check your conversation history with me, its the only thing we've ever spoken about privately):
I have a path fbx object that controls the paths each horse (6 in total) takes. The horses themselves, and the animations for the horses are applied separately to each horse, completely independant of the path animation. I have 120 total possible path animations, and creating an animator controller with 120 animation states is not a solution because of the fact it would bloat the download size. Initially, I had ben creating a separate prefab for each possible outcome, so for example, animationData_123456.glb, animationData_654321 etc , and would download an animation from a different server only when needed. Following the advice here yesterday, specifically the approach outline in the link shared above by Marcel (not the SceneSwitcher approach as you said that that uses the same AssetReference logic under the hood), I now have a list of AssetReferences, which are the animationData prefabs I created, and I load them when needed like shown in the screenshot (I am only loading the first one for testing purposes in the screenshot). Is this a valid approach?
Apologies for going in circles with this point, but I still just cant get my head around how this will keep the project build small. If I do a build, is my build not going to be massively bloated by these animations, and if I make this publically downloadable how is going to remain small? Without these changes today, my project size is about 15MB without gzip compression, which is already large enough.