Page MenuHome

Hair particle systems' caches vanishing
Closed, InvalidPublicKNOWN ISSUE


System Information
Operating system: Windows-10-10.0.18362 64 Bits
Graphics card: GeForce RTX 2080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 431.60

Blender Version
Broken: version: 2.81 (sub 14), branch: master, commit date: 2019-10-07 20:16, hash: rBa856c5eccffb
Worked: (optional)

Short description of error
As adressed in T70642 (but the report was split upon request as it turned out it contained two different issues).

When working with multiple hair particle systems on a same object, the baking process for one particle system will also bake the second particle system simultaneously. However, a click on the other particle system will permanently erase its cache.
It is even more evident when turning on Disk Cache and naming the two caches differently, the vanishing of the second cache can be witnessed in real time in windows explorer.

Exact steps for others to reproduce the error
Simply bake one of the two particle systems present on the object, see how the viewport is reactive once baked, for both systems. Then switch to the other particle system, and notice how when scrolling in the viewport, this particle system has to be recalculated.

Again, as stated in T70642, Bake All Dynamics cannot solve this issue as it has no effect on hair systems.

Event Timeline

Philipp Oeser (lichtwerk) lowered the priority of this task from 90 to 50.Nov 28 2019, 12:48 PM
Sybren A. Stüvel (sybren) claimed this task.
Sybren A. Stüvel (sybren) changed the subtype of this task from "Report" to "Known Issue".

At first I couldn't reproduce this with the current 2.82 branch (rBfb0136f908652a0fef). I used this simplified blend file, with explicit names for the top & side hair systems to ease identification:

This made me think, maybe the issue only occurs when the particle systems are not named for ease of identification. And indeed, when you don't give the point cache a name, it computes one from the object name. In your case, both particle systems are on the same object and thus use the same filename. This is a limitation of the current architecture, and as such is not a bug. Of course it's annoying that bakes disappear, and the UI could warn about this, but that's a different thing. This is also not easy to resolve, as changing this behaviour would invalidate existing point caches.

As a workaround, give each point cache a descriptive name, and this shouldn't happen any more.