Is dynamic backface culling worth it? about three.js HOT 6 CLOSED

Are we talking CanvasRenderer or WebGLRenderer here?

Both have back-face culling already. Depends of your scene and what the user is supposed to see. But yes, it's always good to remove faces that will never be seen (less computations), it's like having code that does nothing.

from three.js.

I currently use WebGLRenderer.

That they both already have backface-culling is great!

I then however wonder why you (here 52ef987) did remove the bottom faces of the cubes in the Minecraft demo. These faces are always backfaces, so they are already culled, right?

As for the culling of invisible faces, it save computations for the GPU, but it takes extra computations for the CPU every time the camera moves (because then, it should be recalculated what is visible and what is not...). There is a trade-off, I guess?

from three.js.

That was alteredq though... Anyway, it's not that we're -removing- the bottom sides of the cube. It's that we're -not adding- them. Kinda different ;)

Backface culling is useful for faces that may be seen at some point. Like the back of the Walt Disney head which you would be seeing. But, in theory, the bottom of the minecraft cubes shouldn't ever be seen, so you can -not add- them to save memory and CPU/GPU computations.

from three.js.

Ah, I see... Should have thought about that myself :-)

Thanks for the explanation!

How can I contribute back to the project and to you? Is there anything you would like?

from three.js.

You can contribute back by doing amazing stuff with it ;)

from three.js.

Will do ;-)

from three.js.