Particle Collision failing in blender 2.80 #68128

New Issue

Sangwook LEE · 2019-08-02T13:04:51+02:00

Sangwook LEE commented

2019-08-02 13:04:51 +02:00

Blender Version
Broken: (example: 2.80rc1, 2.83
Worked: (optional) : 2.79b

Short description of error
My case is that when a collision object moves towards a bunch of particles which has zero velocity, in 2.79b the particles were ricocheted but in 2.80 they go straight.

Exact steps for others to reproduce the error

Open Attached file in 2.79b and 2.80
Run animation and compare
test_279.blend

**Blender Version** Broken: (example: 2.80rc1, 2.83 Worked: (optional) : 2.79b **Short description of error** My case is that when a collision object moves towards a bunch of particles which has zero velocity, in 2.79b the particles were ricocheted but in 2.80 they go straight. **Exact steps for others to reproduce the error** - Open Attached file in 2.79b and 2.80 - Run animation and compare [test_279.blend](https://archive.blender.org/developer/F7645544/test_279.blend)

Sangwook LEE commented

2019-08-02 13:04:51 +02:00

Added subscriber: @sangwook71

Jacques Lucke commented

2019-08-07 16:22:18 +02:00

Added subscriber: @JacquesLucke

Jacques Lucke commented

2019-08-07 16:22:18 +02:00

I can confirm the issue, but it is unlikely that this is going to be fixed.. The current particle system is considered to be end-of-life.

Sangwook LEE commented

2019-08-08 11:32:03 +02:00

Hmm... I think the problem doesn't belong to particle system, but belongs to collision system.
In 2.79 the collision works greatly!, but in advanced 2.8 version, it doesn't work correctly.
I think the new advanced version should include all of the prior version's functionalities except when there are some inevitable reasons that some functionalities should be excluded.

I discovered this error when I was following the Blender Guru tutorial (https:*www.blenderguru.com/), and the title is 'glass smashing (https:*www.youtube.com/watch?v=9L8qOq1Shiw&t=2096s)' which explains the glass shards tutorial using
collision system with particles.

Thank you !

Hmm... I think the problem doesn't belong to particle system, but belongs to collision system. In 2.79 the collision works greatly!, but in advanced 2.8 version, it doesn't work correctly. I think the new advanced version should include all of the prior version's functionalities except when there are some inevitable reasons that some functionalities should be excluded. I discovered this error when I was following the Blender Guru tutorial (https:*www.blenderguru.com/), and the title is 'glass smashing (https:*www.youtube.com/watch?v=9L8qOq1Shiw&t=2096s)' which explains the glass shards tutorial using collision system with particles. Thank you !

Germano Cavalcante changed title from ~~Particle Collision Bug~~ to Particle Collision failing in blender 2.80

2020-02-04 14:56:00 +01:00

Germano Cavalcante commented

2020-02-04 14:56:00 +01:00

Added subscriber: @LucaRood-3

Germano Cavalcante commented

2020-02-04 14:59:52 +01:00

Added subscriber: @mano-wii

Germano Cavalcante commented

2020-02-04 14:59:52 +01:00

@LucaRood-3, shouldn't this problem have been solved by blender/blender-staging@0ef881cc57?

Germano Cavalcante commented

2020-02-12 16:38:20 +01:00

After investigating a little, I realized that the reason for the failure is the Rigdy Body.
There is something wrong with updating the object's matrices in 2.80.
I suspect it is a lack of sync between the evaluated object and the original.

After investigating a little, I realized that the reason for the failure is the `Rigdy Body`. There is something wrong with updating the object's matrices in 2.80. I suspect it is a lack of sync between the evaluated object and the original.

Germano Cavalcante commented

2020-02-13 00:02:07 +01:00

Added subscriber: @Sergey

Germano Cavalcante commented

2020-02-13 00:02:07 +01:00

After investigating, I realized that depsgraph calls 3 eval functions in a problematic order.

BKE_object_eval_local_transform
BKE_object_eval_uber_data
BKE_rigidbody_eval_simulation

BKE_rigidbody_eval_simulation updates ob->obmat according to Rigid Body simulation. This obmat is that seen in the 3dview.
BKE_object_eval_local_transform takes values like ob->loc, ob->rot and ob->scale to update ob->obmat, resetting it to the value before BKE_rigidbody_eval_simulation.
BKE_object_eval_uber_data updates the collision modifier with the same obmat before BKE_rigidbody_eval_simulation. (It is as if the object does not move).

One solution would be to call BKE_object_apply_mat4 and change ob->loc, ob->rot and ob->scale during the BKE_rigidbody_eval_simulation. But I don't know if this is the most correct. BKE_object_eval_local_transform should only be called at the end. No?

@Sergey, would you have any light on this subject?

After investigating, I realized that depsgraph calls 3 eval functions in a problematic order. 1. `BKE_object_eval_local_transform` 2. `BKE_object_eval_uber_data` 3. `BKE_rigidbody_eval_simulation` `BKE_rigidbody_eval_simulation` updates `ob->obmat` according to `Rigid Body` simulation. This `obmat` is that seen in the 3dview. `BKE_object_eval_local_transform` takes values like `ob->loc`, `ob->rot` and `ob->scale` to update `ob->obmat`, resetting it to the value before `BKE_rigidbody_eval_simulation`. `BKE_object_eval_uber_data` updates the collision modifier with the same `obmat` before `BKE_rigidbody_eval_simulation`. (It is as if the object does not move). One solution would be to call `BKE_object_apply_mat4` and change `ob->loc`, `ob->rot` and `ob->scale` during the `BKE_rigidbody_eval_simulation`. But I don't know if this is the most correct. `BKE_object_eval_local_transform` should only be called at the end. No? @Sergey, would you have any light on this subject?

Sergey Sharybin commented

2020-05-04 11:40:39 +02:00

This is the same root issue as #58044. Will merge this report there.

Long story short: generally, rigid body needs to know object's geometry to have proper collision mesh, but geometry might need transform. In this case you're running into a feedback loop. There are ways to solve this, but all of them are rather big design changes.

See #58044 and the discussion in D6834 for more technical insights.

This is the same root issue as #58044. Will merge this report there. Long story short: generally, rigid body needs to know object's geometry to have proper collision mesh, but geometry might need transform. In this case you're running into a feedback loop. There are ways to solve this, but all of them are rather big design changes. See #58044 and the discussion in [D6834](https://archive.blender.org/developer/D6834) for more technical insights.

Sergey Sharybin commented

2020-05-04 11:40:58 +02:00

Closed as duplicate of #58044

Sergey Sharybin closed this issue

2020-05-04 11:40:58 +02:00

Thomas Dinges added this to the 2.83 LTS milestone 2023-02-08 16:39:39 +01:00

Sign in to join this conversation.

No Label

Download

What's New

Blender Studio

Manual

Developers Blog

Documentation

Benchmark

Blender Conference

Development Fund

One-time Donations

Particle Collision failing in blender 2.80 #68128