The reason is that the planner requires the implementation of an interface, which normally has about 20 interactive objects.
One solution is to make them into spine animations, load them directly into the interface, and play the actions when interaction is needed.
Another solution is to make them into spine animations and switch to a static image in standby mode. In the normal standby mode, the static image is displayed, and when interaction is required, the spine is loaded and the actions are played, and then the static image is switched back to after the playback is completed.
The first one is very convenient and easy to maintain, but the disadvantage is that too many spines may affect performance.
The second one is easy to batch, but the disadvantage is that it is very troublesome and the extra cutting will affect the package size.
Considering the difficulty of iteration, I definitely prefer the first option, but the planner asked me a question: How much performance overhead can the second option save? How much impact on frame rate can it reduce? How much heat generation can it reduce?
I didn’t know how to answer this question. Can anyone please tell me how to quantify the impact of these two situations on performance?