particle cache always getting corrupted with replay !?

I have an odd result here. The kernels bounce and rotate fine from frame 1 to 100.
But when the animation loops or replays it seems as there is a new invisible platform
above the collision ground plane and the particles seem to swim in air and also the particle rotation is off.

Anybody has idea what could corrupt the cache or well the particle physics?
It always works fine when I close the file and re-open it. Or it works when I create a new cache data block and delete the old.


Setting the Cache Step to 1 should do it.


how did you know that? The online Blender 2.6 wiki does not explain the step function.

And yes of course your tip fixed it. Shame and ashes on me …
I spend 3 days experimenting …

I think that like other esoteric Blender functions the devs assume that “of course, it’s so obvious what this is all about” when it is not at all obvious unless you are the developer. I’ve run into this many times. It’s led me to develop a “tweak everything” reflex.

Step function: As I understand it, this creates Cache data for only a subset of the entire frame range, based on the step value. So a step of 10 creates 1/10 the total amount of data that a Step 1 creates. I suppose that in playback the frames that have no data are interpolated from the Cache data, but not necessarily in exactly the same manner each time through the frame range, hence the glitching.

but if you would try this out with step 10 the animation is totally goofy so I cannot see how that can even be useful.
it is not like jpeg compression and you can still see the image.

here check this recording you can see how different the animations are.

a bit more doc or tool tip help would be great indeed.

I completely agree, imho opinion default should be 1. Who uses it with 10? It’s one (of the many) things i do need to setup every single f… time. One of those things in Blender - once you found it :wink: - you just need to know.

The default should be changed to 1!

no, its to save memory you only calculate the cache every 10 frames. the particles are then supposed to be interpolated between the cached frames. however given that its an interpolation… thats where the inconsistencies slip in. and if your randomizing any of the data…

I can see the Memory saving Point. But i would prefer to increase Steps if i see i’m running low on Memory, instead of saving Memory by default while most of the time i don’t run low on memory. I just see that i change this setting every single time. I never used step-size of 10, so far. It would be as well (probably) easier for newbies. just my opinion

Oh, I agree. I was just pointing out what was probobly going through the dev’s heads; remember that blender has been around for several years and state of the rat in memory capacity has improved greatly since the first days of particle sims

mh the cache interpolation can work well for linear motions but at one point it will break.

well good to know this is not a bug.