If it’s good to change rotation probably each body, then a script that runs each body is just about the best way to go.

In the instance screenshot there are perhaps 30 sprites on display. It can be very tough to trigger a noticeable efficiency affect from doing something a mere 30 occasions per body. So as TomTsagk says in a remark, it is best to undoubtedly check out this easiest, most direct possibility first, and profile it to confirm whether or not there’s any actual downside that may pressure you to do one thing extra complicated.

If your digicam solely rotates in discrete hops, you would arrange this script to run solely when the digicam strikes (so, in the identical script the place you pay attention for the Q/E key press to start out the digicam transfer, you additionally hearth off the billboarding replace). Then at the very least it isn’t working redundantly when the digicam angle stays fastened.

You need to orient the sprites or quads to face alongside (the negation of) the digicam’s ahead vector. That means you needn’t change the sprite rotation when the digicam interprets, solely when it rotates.

Technically, you would additionally accomplish this by abuse of a particle system, which have built-in billboarding choices. Those compute new billboarded quads for each particle each body with out important efficiency affect, even for 1000’s of simultaneous particles on display. But the indirection of routing all of your sprite positioning and animation updates by way of a particle buffer is more likely to get onerous and sap extra productiveness than it’s value.

Lastly, you would implement the billboarding impact in a shader, repositioning the vertices to snap them to face the digicam simply earlier than rasterizing the sprite. This is somewhat sophisticated to do with out breaking sprite batching although, because the vertices would one way or the other want to determine the place the sprite’s pivot is, to rotate the billboard round that. Normally that data has already been baked down by that stage of the pipeline, so it could be further work to reconstruct.

So, I’d say: attempt the straightforward rotation script first. Profile it. If you discover it has an unacceptable price, replace your query along with your findings and we will discover potential fixes. But these fixes are themselves sophisticated sufficient to be a web drain in your productiveness if they don’t seem to be really wanted.