Instance Cache and Clip to Camera

[EdBardet]

If you are using the instance cache, does clip to camera gain you any rendering speed?
Thanks
Ed

[Dune]

I don't know, but you can simply test that in a simple setup, I guess.

[jo]

Hi Ed,

The instance cache is primarily for reducing the amount of time needed to populate, once you generate the cache file. Clip to camera is also aimed at reducing the time needed to populate, because it only populates areas visible to the camera, but of course it also needs repopulating which isn't the case with an instance cache.

I'm not actually sure of the render time benefits of using clip to camera, although I do know that it can be a bit problematic if you need reflections of or shadows from the instances outside the camera view, or if you're animating.

However I can say that if you used clip to camera to generate an instance cache then there would be no render speed difference between using that cache file and using clip to camera without a cache. This is because the same instances would be in the render with either method.

The only clear difference to render speed I can think of is if you're using clip to camera and you haven't repopulated prior to rendering. This is because the populators will need to repopulate as the render process starts and that could take some time which adds to the render time. If you were using an instance cache this wouldn't happen.

Regards,

Jo

[EdBardet]

Jo- Thanks

I just completed a test, like Dune suggested and it makes no difference on rendering if you have populated and set up the caches already. I think I thought it through just like you said.

Jotting down the population times also showed me another item. It takes just about as as long to populate with clip to camera as without. The computations all take place and just the occurrences are less.

Thanks again,
Ed

[Matt]

The main benefit to "clip to camera" is that it uses less memory. I don't believe there would be much difference in population time unless you had been running out of memory.

It should theoretically reduce render times when there are fewer instances in the population. The difference might be measurable if you have an enormous population. If the difference isn't measurable, that means the renderer's acceleration structures are working well to minimise the effect of scene complexity on render times, i.e. that makes me happy

Matt