Keyed Particle Physics Incorrectly Rendered Under Eevee and Cycles, but Correctly Rendered in Viewport Render #80680

New Issue

Jonathan Manton · 2020-09-11T05:28:04+02:00

Jonathan Manton commented

2020-09-11 05:28:04 +02:00

System Information
Operating system: Windows-10-10.0.19041-SP0 64 Bits
Graphics card: Radeon RX550/550 Series ATI Technologies Inc. 4.5.14736 Core Profile Context 20.8.3 27.20.12027.1001

Blender Version
Broken: version: 2.91.0 Alpha, branch: master (modified), commit date: 2020-09-13 23:10, hash: 0f61b27e93
Worked: Never (2.80+)

Short description of error
Create two objects. Attach a normal particle system to the first one. Attach a keyed particle system to the second one, with the two relations "first object" then "second object". Therefore, particles should move from the first object to the second object. If the starting velocities are non-trivial and the keyed particles are generated over three or more frames (e.g., start time 10, end time at least 12) then the particles of the keyed system are correctly rendered in the viewport (rendered mode) but appear in different positions when actually rendering (pressing F12).

I first discovered this with a more complicated example, and the same behaviour occurred under both 2.90 and 2.83 (normal releases). Googling did not turn up anything relevant. (Motion blur is off.)

I have been able to recreate it with a simple .blend file attached. Included are two screenshots at frame 20, one of the viewport, one of the rendered image. While the difference is just one particle here (a red particle is over the left square in the viewport but off the square in the rendered image), vastly different behaviour can be achieved by increasing the difference between start and end times.

In the more complicated example, not included here, a particle would appear at the top of the screen and move towards its final destination. When rendered though, it would appear in the correct place on that frame but then immediately jump to a very different location on the next frame.

Note: This bug might depend on the first object's particles having a short lifetime - the original motivation is having one object explode then morph into another object. The rendered animation looked very different (and incorrect) from the viewport animation.

system-info.txt

Exact steps for others to reproduce the error
Debug 2.blend

Open it.
Set viewport to Rendered mode.
Move playhead to frame 20.
Observe the locations of the 4 red particles near Plane.001. (They form a loose "T" shape and none are touching Plane.001.)
Press F12.
Observe the locations of the 4 red particles near Plane.001 are very different. (Two are touching Plane.001 and the other two are in line ready to arrive at Plane.001.)

**System Information** Operating system: Windows-10-10.0.19041-SP0 64 Bits Graphics card: Radeon RX550/550 Series ATI Technologies Inc. 4.5.14736 Core Profile Context 20.8.3 27.20.12027.1001 **Blender Version** Broken: version: 2.91.0 Alpha, branch: master (modified), commit date: 2020-09-13 23:10, hash: `0f61b27e93` Worked: Never (2.80+) **Short description of error** Create two objects. Attach a normal particle system to the first one. Attach a keyed particle system to the second one, with the two relations "first object" then "second object". Therefore, particles should move from the first object to the second object. If the starting velocities are non-trivial and the keyed particles are generated over three or more frames (e.g., start time 10, end time at least 12) then the particles of the keyed system are correctly rendered in the viewport (rendered mode) but appear in different positions when actually rendering (pressing F12). I first discovered this with a more complicated example, and the same behaviour occurred under both 2.90 and 2.83 (normal releases). Googling did not turn up anything relevant. (Motion blur is off.) I have been able to recreate it with a simple .blend file attached. Included are two screenshots at frame 20, one of the viewport, one of the rendered image. While the difference is just one particle here (a red particle is over the left square in the viewport but off the square in the rendered image), vastly different behaviour can be achieved by increasing the difference between start and end times. In the more complicated example, not included here, a particle would appear at the top of the screen and move towards its final destination. When rendered though, it would appear in the correct place on that frame but then immediately jump to a very different location on the next frame. Note: This bug might depend on the first object's particles having a short lifetime - the original motivation is having one object explode then morph into another object. The rendered animation looked very different (and incorrect) from the viewport animation. [system-info.txt](https://archive.blender.org/developer/F8864029/system-info.txt) ![Debug2_viewport.png](https://archive.blender.org/developer/F8864036/Debug2_viewport.png) ![Debug2_eevee_render.png](https://archive.blender.org/developer/F8864037/Debug2_eevee_render.png) **Exact steps for others to reproduce the error** [Debug 2.blend](https://archive.blender.org/developer/F8883931/Debug_2.blend) 1) Open it. 2) Set viewport to Rendered mode. 3) Move playhead to frame 20. 4) Observe the locations of the 4 red particles near Plane.001. (They form a loose "T" shape and none are touching Plane.001.) 5) Press F12. 6) Observe the locations of the 4 red particles near Plane.001 are very different. (Two are touching Plane.001 and the other two are in line ready to arrive at Plane.001.)

Jonathan Manton commented

2020-09-11 05:28:04 +02:00

Added subscriber: @jmanton

Germano Cavalcante commented

2020-09-14 16:06:16 +02:00

Added subscribers: @fclem, @mano-wii

Germano Cavalcante commented

2020-09-14 16:06:16 +02:00

Changed status from 'Needs Triage' to: 'Archived'

Germano Cavalcante closed this issue

2020-09-14 16:06:16 +02:00

Germano Cavalcante self-assigned this 2020-09-14 16:06:16 +02:00

Germano Cavalcante commented

2020-09-14 16:06:16 +02:00

Thanks for the report,

but the default particle drawing is an overlay.
Overlays do not appear in the F12 render.

So the issue reported here is a request for modified/improved behavior and not a bug in current behavior. Closing as this bug tracker is only for bugs and errors.

For user requests and feedback, please use other channels: https://wiki.blender.org/wiki/Communication/Contact#User_Feedback_and_Requests

For more information on why this isn't considered a bug, visit: https://wiki.blender.org/wiki/Reference/Not_a_bug

I just find it a little strange that these particles are not hidden when you disable overlays.
@fclem, is this intentional?

Thanks for the report, but the default particle drawing is an overlay. Overlays do not appear in the F12 render. So the issue reported here is a request for modified/improved behavior and not a bug in current behavior. Closing as this bug tracker is only for bugs and errors. For user requests and feedback, please use other channels: https://wiki.blender.org/wiki/Communication/Contact#User_Feedback_and_Requests For more information on why this isn't considered a bug, visit: https://wiki.blender.org/wiki/Reference/Not_a_bug ---- I just find it a little strange that these particles are not hidden when you disable overlays. @fclem, is this intentional?

Germano Cavalcante commented

2020-09-14 16:11:21 +02:00

Changed status from 'Archived' to: 'Needs Triage'

Germano Cavalcante reopened this issue

2020-09-14 16:11:21 +02:00

Germano Cavalcante removed their assignment 2020-09-14 16:11:21 +02:00

Germano Cavalcante commented

2020-09-14 16:11:21 +02:00

After reading the report again, it seems that the problem involves the red particles. Maybe I got it wrong.

Jonathan Manton commented

2020-09-15 02:36:23 +02:00

Yes, the problem involves the red particles (which are displayed as actual objects).

Further experimentation suggests that the problem occurs when a keyed particle travels from a particle A in system 1 to a particle B in system 2, so to speak, and particle A dies before the keyed particle reaches its destination. This does not cause a problem in the viewport display but it does cause a problem in the rendered display.

Yes, the problem involves the red particles (which are displayed as actual objects). Further experimentation suggests that the problem occurs when a keyed particle travels from a particle A in system 1 to a particle B in system 2, so to speak, and particle A dies before the keyed particle reaches its destination. This does not cause a problem in the viewport display but it does cause a problem in the rendered display.

Richard Antalik commented

2020-09-15 14:39:05 +02:00

Added subscriber: @iss

Richard Antalik commented

2020-09-15 14:39:05 +02:00

Changed status from 'Needs Triage' to: 'Needs User Info'

Richard Antalik commented

2020-09-15 14:39:05 +02:00

I can't reproduce any difference with file provided.

Does this happen when you bake cache for particles?
Please check this with build from https://builder.blender.org/download/

I can't reproduce any difference with file provided. Does this happen when you bake cache for particles? Please check this with build from https://builder.blender.org/download/

Jonathan Manton commented

2020-09-15 15:40:44 +02:00

There is no need to bake.
In this simple example, it is only one particle that is incorrect; this is shown in the attached screenshots which were taken at frame 20.
Are you able to reproduce the exact screenshots using Blender 2.90? (I am using the standard download of Blender.)
One screenshot is of the viewport and shows the correct location of all the particles.
The other is of the F12 render and one particle is in the incorrect position.
(Others have experienced this problem as I saw a YouTube video where a workaround was given... I tried to reduce it to as simple an example as possible.)

There is no need to bake. In this simple example, it is only one particle that is incorrect; this is shown in the attached screenshots which were taken at frame 20. Are you able to reproduce the exact screenshots using Blender 2.90? (I am using the standard download of Blender.) One screenshot is of the viewport and shows the correct location of all the particles. The other is of the F12 render and one particle is in the incorrect position. (Others have experienced this problem as I saw a YouTube video where a workaround was given... I tried to reduce it to as simple an example as possible.)

Jonathan Manton commented

2020-09-15 15:56:44 +02:00

I just downloaded the file and you are right, there is no difference. Let me work this out then get back to you soon. Sorry!

Jonathan Manton commented

2020-09-15 16:19:11 +02:00

Ok. I can recreate the problem but not the exact screenshots from before. In the file I sent, go to Plane and set the particle system's lifetime to 15 (down from 50). Set the playhead to frame 0 or 1. Then advance to frame 20. Then compare the red particles in the viewport (turn on Render view) with the F12 render particles. I can upload a new file if you prefer.

(Another interesting thing is that if the file is saved with the playhead at frame 20 then re-opened, the viewport will display the particles in the F12 render position rather than in the viewport position, unless you first rewind the playhead so the cache gets updated.)

Ok. I can recreate the problem but not the exact screenshots from before. In the file I sent, go to Plane and set the particle system's lifetime to 15 (down from 50). Set the playhead to frame 0 or 1. Then advance to frame 20. Then compare the red particles in the viewport (turn on Render view) with the F12 render particles. I can upload a new file if you prefer. (Another interesting thing is that if the file is saved with the playhead at frame 20 then re-opened, the viewport will display the particles in the F12 render position rather than in the viewport position, unless you first rewind the playhead so the cache gets updated.)

Jonathan Manton commented

2020-09-15 16:27:05 +02:00

With the new Debug 2.blend file attached here:

Open it.
Set viewport to Rendered mode.
Move playhead to frame 20.
Observe the locations of the 4 red particles near Plane.001. (They form a loose "T" shape and none are touching Plane.001.)
Press F12.
Observe the locations of the 4 red particles near Plane.001 are very different. (Two are touching Plane.001 and the other two are in line ready to arrive at Plane.001.)

Debug 2.blend

With the new Debug 2.blend file attached here: 1) Open it. 2) Set viewport to Rendered mode. 3) Move playhead to frame 20. 4) Observe the locations of the 4 red particles near Plane.001. (They form a loose "T" shape and none are touching Plane.001.) 5) Press F12. 6) Observe the locations of the 4 red particles near Plane.001 are very different. (Two are touching Plane.001 and the other two are in line ready to arrive at Plane.001.) [Debug 2.blend](https://archive.blender.org/developer/F8883931/Debug_2.blend)

Richard Antalik commented

2020-09-15 18:03:58 +02:00

Changed status from 'Needs User Info' to: 'Confirmed'

Richard Antalik commented

2020-09-15 18:05:09 +02:00

I think this may be limitation, but I will confirm anyway.

@jmanton do you know if this ever worked correctly?

I think this may be limitation, but I will confirm anyway. @jmanton do you know if this ever worked correctly?

Jonathan Manton commented

2020-09-16 01:50:09 +02:00

@iss Thanks for looking into this. I don't know if it ever worked correctly. Since the viewport render computes the correct trajectories, hopefully it is not an inherent limitation but rather a bug that the F12 render system can put particles in very different positions to the viewport render (so what-you-see-is-not-what-you-get).

Jacques Lucke commented

2020-09-16 12:54:30 +02:00

Added subscriber: @JacquesLucke

Jacques Lucke commented

2020-09-16 12:54:30 +02:00

When I try to render frame 20 in a debug build I get this:

libc.so.6!raise (Unknown Source:0)
libc.so.6!abort (Unknown Source:0)
_BLI_assert_abort() (/home/jacques/blender-git/blender/source/blender/blenlib/intern/BLI_assert.c:50)
interp_qt_qtqt(float * q, const float * a, const float * b, const float t) (/home/jacques/blender-git/blender/source/blender/blenlib/intern/math_rotation.c:874)
do_particle_interpolation(ParticleSystem * psys, int p, ParticleData * pa, float t, ParticleInterpolationData * pind, ParticleKey * result) (/home/jacques/blender-git/blender/source/blender/blenkernel/intern/particle.c:1310)
psys_get_particle_on_path(ParticleSimulationData * sim, int p, ParticleKey * state, const _Bool vel) (/home/jacques/blender-git/blender/source/blender/blenkernel/intern/particle.c:4422)
psys_get_particle_state(ParticleSimulationData * sim, int p, ParticleKey * state, int always) (/home/jacques/blender-git/blender/source/blender/blenkernel/intern/particle.c:4693)
make_duplis_particle_system(const DupliContext * ctx, ParticleSystem * psys) (/home/jacques/blender-git/blender/source/blender/blenkernel/intern/object_dupli.c:1340)
make_duplis_particles(const DupliContext * ctx) (/home/jacques/blender-git/blender/source/blender/blenkernel/intern/object_dupli.c:1453)
object_duplilist(Depsgraph * depsgraph, Scene * sce, Object * ob) (/home/jacques/blender-git/blender/source/blender/blenkernel/intern/object_dupli.c:1525)
(anonymous namespace)::deg_iterator_objects_step(BLI_Iterator * iter, blender::deg::IDNode * id_node) (/home/jacques/blender-git/blender/source/blender/depsgraph/intern/depsgraph_query_iter.cc:230)
DEG_iterator_objects_next(BLI_Iterator * iter) (/home/jacques/blender-git/blender/source/blender/depsgraph/intern/depsgraph_query_iter.cc:302)
DRW_render_object_iter(void * vedata, RenderEngine * engine, struct Depsgraph * depsgraph, void (*)(void *, Object *, RenderEngine *, struct Depsgraph *) callback) (/home/jacques/blender-git/blender/source/blender/draw/intern/draw_manager.c:1903)
eevee_render_to_image(void * vedata, RenderEngine * engine, struct RenderLayer * render_layer, const rcti * rect) (/home/jacques/blender-git/blender/source/blender/draw/engines/eevee/eevee_engine.c:524)
DRW_render_to_image(RenderEngine * engine, struct Depsgraph * depsgraph) (/home/jacques/blender-git/blender/source/blender/draw/intern/draw_manager.c:1868)
RE_engine_render(Render * re, int do_all) (/home/jacques/blender-git/blender/source/blender/render/intern/source/external_engine.c:878)
do_render_3d(Render * re) (/home/jacques/blender-git/blender/source/blender/render/intern/source/pipeline.c:1136)
do_render(Render * re) (/home/jacques/blender-git/blender/source/blender/render/intern/source/pipeline.c:1213)
do_render_composite(Render * re) (/home/jacques/blender-git/blender/source/blender/render/intern/source/pipeline.c:1353)
do_render_all_options(Render * re) (/home/jacques/blender-git/blender/source/blender/render/intern/source/pipeline.c:1613)

Can someone reproduce that?

When I try to render frame 20 in a debug build I get this: ``` libc.so.6!raise (Unknown Source:0) libc.so.6!abort (Unknown Source:0) _BLI_assert_abort() (/home/jacques/blender-git/blender/source/blender/blenlib/intern/BLI_assert.c:50) interp_qt_qtqt(float * q, const float * a, const float * b, const float t) (/home/jacques/blender-git/blender/source/blender/blenlib/intern/math_rotation.c:874) do_particle_interpolation(ParticleSystem * psys, int p, ParticleData * pa, float t, ParticleInterpolationData * pind, ParticleKey * result) (/home/jacques/blender-git/blender/source/blender/blenkernel/intern/particle.c:1310) psys_get_particle_on_path(ParticleSimulationData * sim, int p, ParticleKey * state, const _Bool vel) (/home/jacques/blender-git/blender/source/blender/blenkernel/intern/particle.c:4422) psys_get_particle_state(ParticleSimulationData * sim, int p, ParticleKey * state, int always) (/home/jacques/blender-git/blender/source/blender/blenkernel/intern/particle.c:4693) make_duplis_particle_system(const DupliContext * ctx, ParticleSystem * psys) (/home/jacques/blender-git/blender/source/blender/blenkernel/intern/object_dupli.c:1340) make_duplis_particles(const DupliContext * ctx) (/home/jacques/blender-git/blender/source/blender/blenkernel/intern/object_dupli.c:1453) object_duplilist(Depsgraph * depsgraph, Scene * sce, Object * ob) (/home/jacques/blender-git/blender/source/blender/blenkernel/intern/object_dupli.c:1525) (anonymous namespace)::deg_iterator_objects_step(BLI_Iterator * iter, blender::deg::IDNode * id_node) (/home/jacques/blender-git/blender/source/blender/depsgraph/intern/depsgraph_query_iter.cc:230) DEG_iterator_objects_next(BLI_Iterator * iter) (/home/jacques/blender-git/blender/source/blender/depsgraph/intern/depsgraph_query_iter.cc:302) DRW_render_object_iter(void * vedata, RenderEngine * engine, struct Depsgraph * depsgraph, void (*)(void *, Object *, RenderEngine *, struct Depsgraph *) callback) (/home/jacques/blender-git/blender/source/blender/draw/intern/draw_manager.c:1903) eevee_render_to_image(void * vedata, RenderEngine * engine, struct RenderLayer * render_layer, const rcti * rect) (/home/jacques/blender-git/blender/source/blender/draw/engines/eevee/eevee_engine.c:524) DRW_render_to_image(RenderEngine * engine, struct Depsgraph * depsgraph) (/home/jacques/blender-git/blender/source/blender/draw/intern/draw_manager.c:1868) RE_engine_render(Render * re, int do_all) (/home/jacques/blender-git/blender/source/blender/render/intern/source/external_engine.c:878) do_render_3d(Render * re) (/home/jacques/blender-git/blender/source/blender/render/intern/source/pipeline.c:1136) do_render(Render * re) (/home/jacques/blender-git/blender/source/blender/render/intern/source/pipeline.c:1213) do_render_composite(Render * re) (/home/jacques/blender-git/blender/source/blender/render/intern/source/pipeline.c:1353) do_render_all_options(Render * re) (/home/jacques/blender-git/blender/source/blender/render/intern/source/pipeline.c:1613) ``` Can someone reproduce that?

Richard Antalik commented

2020-09-17 13:37:59 +02:00

In #80680#1016822, @JacquesLucke wrote:
When I try to render frame 20 in a debug build I get this:
Can someone reproduce that?

No problems here

> In #80680#1016822, @JacquesLucke wrote: > When I try to render frame 20 in a debug build I get this: > Can someone reproduce that? No problems here

Dalai Felinto commented

2022-05-09 15:18:35 +02:00

Added subscriber: @dfelinto

Dalai Felinto commented

2022-05-09 15:18:35 +02:00

Thanks for the report. Marking this as known issue since it depends on the old particle settings which is end of life .

Thanks for the report. Marking this as known issue since it depends on the [old particle settings which is end of life ](https://developer.blender.org/tag/nodes_physics/).

Philipp Oeser removed the

Download

What's New

Blender Studio

Manual

Developers Blog

Documentation

Benchmark

Blender Conference

Development Fund

One-time Donations

Keyed Particle Physics Incorrectly Rendered Under Eevee and Cycles, but Correctly Rendered in Viewport Render #80680