Caching a SOP animation

[hopbin9]

Hi,

I've some animated objects that are taking a long time to update in the viewport. Is there a way to cache out the animation of a lot of Copy SOPs so that it loads up quickly. I think it's the cooking time that is slowing everything down.

Thanks,

[lukeiamyourfather]

Hi,

I've some animated objects that are taking a long time to update in the viewport. Is there a way to cache out the animation of a lot of Copy SOPs so that it loads up quickly. I think it's the cooking time that is slowing everything down.

Thanks,

Drop down a ROP Geometry driver node in the SOP and connect what you want to cache. This saves out geometry to whatever format you want and however often you want. Then use a File node to read the cache back.

[hopbin9]

thank you, that works great.

[anim]

if your geometry doesn't change topology over time you can cache to MDD (with MDD Point Cache ROP)

it will store only position values so the resulting cache will be much smaller and load time (MDD SOP) will be also lower

[Stalkerx777]

There is also Cache SOP

[Netvudu]

actually you could use the File SOP both to first write and then to read the cache (i.e. in Automatic mode). I just tend not to do it that way because sometimes I accidentally overwrite caches, which is why I also favour the ROP Output drive/ File combination.

Edited February 11, 2011 by Netvudu

[JeffBenjamin]

actually you could use the File SOP both to first write and then to read the cache (i.e. in Automatic mode). I just tend not to do it that way because sometimes I accidentally overwrite caches, which is why I also favour the ROP Output drive/ File combination.

I always favor writing out cache to disk, as it keeps ram free for the application. if you do decide to use a cache sop, make sure to periodically check your available ram and when it gets low clear the cache out. Another helpful tip, color your cache nodes something bright, or put sticky notes down next to them. One of the msot frustrating things in the world is forgetting to re-cache something when you have made changes upstream, and then losing an hour of work when you finally realize what happened.

[hopbin9]

Thanks, that's a good tip.

Caching didn't give me as much help in performance as I had hoped. So I spent almost two days reworking the entire project to use more instances as the node level. What an amazing difference that made.

Memory used by Mantra before: 4BG after: 1.5BG

Before I was not able to playback the scene in the viewport, and now I get about 18FPS. That might not sound like much, but I can not move the timeline to any frame without feeling like I'm lagged.