Unable to read back in external baked physics cache #69308

New Issue

Paul Chambers · 2019-08-29T17:50:11+02:00

Paul Chambers commented

2019-08-29 17:50:11 +02:00

System Information
Operating system: Windows-10-10.0.18362 64 Bits
Graphics card: GeForce GTX 1080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 416.34

Blender Version
Broken: version: 2.80 (sub 75), branch: master, commit date: 2019-07-29 14:47, hash: f6cb5f5449
Worked: (optional)

Short description of error
When baking a simple physics particle cache to file, it appears to bake correctly, but then when attempting to read it back in it either reports as no valid file found, or only identifies 1 or 2 points instead of hundreds. Sample file with bake attached,

Exact steps for others to reproduce the error
[Please describe the exact steps needed to reproduce the issue]
[Based on the default startup or an attached .blend file (as simple as possible)]

**System Information** Operating system: Windows-10-10.0.18362 64 Bits Graphics card: GeForce GTX 1080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 416.34 **Blender Version** Broken: version: 2.80 (sub 75), branch: master, commit date: 2019-07-29 14:47, hash: `f6cb5f5449` Worked: (optional) **Short description of error** When baking a simple physics particle cache to file, it appears to bake correctly, but then when attempting to read it back in it either reports as no valid file found, or only identifies 1 or 2 points instead of hundreds. Sample file with bake attached, **Exact steps for others to reproduce the error** [Please describe the exact steps needed to reproduce the issue] [Based on the default startup or an attached .blend file (as simple as possible)]

Paul Chambers commented

2019-08-29 17:50:11 +02:00

Added subscriber: @PaulChambers

blender-admin commented

2019-08-29 17:50:11 +02:00

#68436 was marked as duplicate of this issue

Paul Chambers commented

2019-08-29 18:02:39 +02:00

physics-test.zip

[physics-test.zip](https://archive.blender.org/developer/F7709758/physics-test.zip)

Philipp Oeser commented

2019-09-02 11:30:55 +02:00

Added subscribers: @brecht, @lichtwerk

Philipp Oeser commented

2019-09-02 11:30:55 +02:00

Can confirm.
Even though particles are under complete rewrite atm, this looks pretty bad to me.
I dare setting this to High prio, @brecht correct me if I am wrong.
Or does EOL mean we are shipping unusable versions until the new stuff is in place? (we could, just asking...)

Can confirm. Even though particles are under complete rewrite atm, this looks pretty bad to me. I dare setting this to High prio, @brecht correct me if I am wrong. Or does EOL mean we are shipping unusable versions until the new stuff is in place? (we could, just asking...)

Sybren A. Stüvel was assigned by Philipp Oeser

2019-09-06 12:39:23 +02:00

Philipp Oeser commented

2019-09-06 12:39:23 +02:00

Added subscriber: @dr.sybren

Philipp Oeser commented

2019-09-06 12:39:23 +02:00

@dr.sybren: maybe you could check on this?

Sybren A. Stüvel commented

2019-09-18 14:17:01 +02:00

I did some digging, and it turns out that ptcache_read() reads the correct number of points. It's ptcache_particle_read() that's rejecting the points because psys->totpart is set to a low number (i.e. to the number of points you're seeing). Somehow it's not getting the total number of points from the pointcache. This could be due to the fact that the info file (i.e. the bphys file for frame 0) isn't written to disk, but I'm too unfamiliar with how the code is supposed to work.

A test with 2.79 shows that this issue was already there in 2.79 (and possibly earlier). AFAIK this means that it won't be fixed any more.

PS: Un-checking the 'External Cache' checkbox erases all cache files without any notification, if they happen to be in the same directory as Blender would store them when baking, so it's a dangerous thing to touch.

I did some digging, and it turns out that `ptcache_read()` reads the correct number of points. It's `ptcache_particle_read()` that's rejecting the points because `psys->totpart` is set to a low number (i.e. to the number of points you're seeing). Somehow it's not getting the total number of points from the pointcache. This could be due to the fact that the info file (i.e. the bphys file for frame 0) isn't written to disk, but I'm too unfamiliar with how the code is supposed to work. A test with 2.79 shows that this issue was already there in 2.79 (and possibly earlier). AFAIK this means that it won't be fixed any more. PS: Un-checking the 'External Cache' checkbox erases all cache files without any notification, if they happen to be in the same directory as Blender would store them when baking, so it's a dangerous thing to touch.

Sybren A. Stüvel was unassigned by Brecht Van Lommel

2019-09-18 14:41:33 +02:00

Brecht Van Lommel commented

2019-09-18 14:41:33 +02:00

This is not high priority especially if it happens in 2.79 and with particles being end of life.

That means it's low priority, someone might fix it but we won't prioritize it.

This is not high priority especially if it happens in 2.79 and with particles being end of life. That means it's low priority, someone might fix it but we won't prioritize it.

Philipp Oeser commented

2019-09-18 15:09:21 +02:00

Sorry, my bad, I thought I checked and this was working in 2.79, but apparently it isnt...

Philipp Oeser commented

2020-01-27 11:52:36 +01:00

From my testing in #68436 (External Particle Cache Playback Is Broken (2.8)), the bphys file for frame zero was actually written... So my tests were working in 2.79...
But these two reports are the same, will merge #68436 here.

From my testing in #68436 (External Particle Cache Playback Is Broken (2.8)), the bphys file for frame zero was actually written... So my tests were working in 2.79... But these two reports are the same, will merge #68436 here.

Philipp Oeser commented

2020-01-27 11:53:05 +01:00

Added subscribers: @AtomP, @marks-4, @grosgood, @Pavel_Blend