Emitted Metaball not showing and rendering correctly #43666

New Issue

Eli Spizzichino · 2015-02-15T03:58:15+01:00

Eli Spizzichino commented

2015-02-15 03:58:15 +01:00

System Information
kubuntu 14.04 with gtx 580 and 590

Blender Version
Broken: >= 2.72
Worked: <= 2.71

Short description of error
Created a simple scene with emiter set to fluid and to render object metaball and a collision object with particle dumping
Metaball are invisible or rendered low res and not stiching to the surface

Exact steps for others to reproduce the error
Based on a (as simple as possible) attached .blend file with minimum amount of steps
DROP.blend

**System Information** kubuntu 14.04 with gtx 580 and 590 **Blender Version** Broken: >= 2.72 Worked: <= 2.71 **Short description of error** Created a simple scene with emiter set to fluid and to render object metaball and a collision object with particle dumping Metaball are invisible or rendered low res and not stiching to the surface **Exact steps for others to reproduce the error** Based on a (as simple as possible) attached .blend file with minimum amount of steps [DROP.blend](https://archive.blender.org/developer/F143382/DROP.blend)

Eli Spizzichino commented

2015-02-15 03:58:16 +01:00

Changed status to: 'Open'

Eli Spizzichino commented

2015-02-15 03:58:16 +01:00

Added subscriber: @EliSpizzichino

ronan ducluzeau commented

2015-02-15 09:54:32 +01:00

Added subscriber: @zeauro

ronan ducluzeau commented

2015-02-15 09:54:32 +01:00

It is not a bug.
A metaball is scaled down relatively to particle size (0.050 by default).
So, they are too small to be represented by polygonisation resolution of metaball that is higher ( at 0.4 by default).

You just have to decrease metaball viewport resolution to 0.025 or increase particle size to 0.3 to see metaballs in viewport.

It is not a bug. A metaball is scaled down relatively to particle size (0.050 by default). So, they are too small to be represented by polygonisation resolution of metaball that is higher ( at 0.4 by default). You just have to decrease metaball viewport resolution to 0.025 or increase particle size to 0.3 to see metaballs in viewport.

Eli Spizzichino commented

2015-02-15 12:30:20 +01:00

Hi,
ok I see it's related to the resolution of the mball, but how you don't call it a bug when from 2.71 to 2.72 those settings are interpreted completely differently and you get a different/unexpected result?

This can cause problems for example if I'm already at the maximum mball resolution (0.05/0.025) to adapt to the new interpretation of those number I'm forced to scale the particle instance up and hence the entire scene!

Hi, ok I see it's related to the resolution of the mball, but how you don't call it a bug when from 2.71 to 2.72 those settings are interpreted completely differently and you get a different/unexpected result? ![mball_2.71.jpg](https://archive.blender.org/developer/F143481/mball_2.71.jpg) ![mball_2.72.jpg](https://archive.blender.org/developer/F143483/mball_2.72.jpg) This can cause problems for example if I'm already at the maximum mball resolution (0.05/0.025) to adapt to the new interpretation of those number I'm forced to scale the particle instance up and hence the entire scene!

Julian Eisel commented

2015-02-16 20:46:56 +01:00

Added subscribers: @LukasTonne, @JulianEisel

Julian Eisel commented

2015-02-16 20:46:56 +01:00

I'm not sure myself what should happen here and what not ;) @LukasTonne, maybe you can tell us a bit more?

Bastien Montagne was assigned by Sergey Sharybin

2015-02-24 10:09:46 +01:00

Sergey Sharybin commented

2015-02-24 10:09:46 +01:00

Added subscribers: @mont29, @Sergey

Sergey Sharybin commented

2015-02-24 10:09:46 +01:00

It was bug in the matrix calculation for metaballs emitted from duplis which was fixed in blender 2.72. That bug lead to unwanted transform in certain circumstances. Fix for it caused the regression in this particular case.

I can't fully understand why @mont29 did what he did, so would ask him to revisit e0d8e62 and see if there's some unwanted side effects are happening there.

It was bug in the matrix calculation for metaballs emitted from duplis which was fixed in blender 2.72. That bug lead to unwanted transform in certain circumstances. Fix for it caused the regression in this particular case. I can't fully understand why @mont29 did what he did, so would ask him to revisit e0d8e62 and see if there's some unwanted side effects are happening there.

Bastien Montagne commented

2015-02-24 12:37:28 +01:00

Hmm… e0d8e62 does not cause that bug (if this is a bug), it reveals it.

Thing is, previously, the org meta object used as dupli source would have its matrix permanently replaced by its one-before-last dupli instance's matrix (as soon as there were two or more dupli instances). Now, that org object's matrix gets correctly reset, after we have cycled through all its dupliinstances. Guess it's that object's matrix that gets used to define metaballs 'definition', current results are expected and correct ones.

More specifically:

BKE_mball_polygonize() calls first mball_count(), which, in old code, would set ob.obmat to the penultimate dupli instance's matrix.
BKE_mball_polygonize() then calls init_meta(), which, in old code, would use that wrong ob.obmat to init metaball data.

In new code, BKE_scene_base_iter_next() properly restores ob.obmat to its real org value.

Not sure what we want to do here, @Sergey? To me, current behavior is correct…

Hmm… e0d8e62 does not cause that bug (if this is a bug), it reveals it. Thing is, previously, the org meta object used as dupli source would have its matrix permanently replaced by its one-before-last dupli instance's matrix (as soon as there were two or more dupli instances). Now, that org object's matrix gets correctly reset, after we have cycled through all its dupliinstances. Guess it's that object's matrix that gets used to define metaballs 'definition', current results are expected and correct ones. More specifically: * `BKE_mball_polygonize()` calls first `mball_count()`, which, in old code, would set `ob.obmat` to the penultimate dupli instance's matrix. * `BKE_mball_polygonize()` then calls `init_meta()`, which, in old code, would use that wrong `ob.obmat` to init metaball data. In new code, `BKE_scene_base_iter_next()` properly restores `ob.obmat` to its real org value. Not sure what we want to do here, @Sergey? To me, current behavior is correct…

Sergey Sharybin commented

2015-02-24 17:56:45 +01:00

@mont29, we for sure don't want the dupli matrix affecting motherball or so and from your explanation it seems indeed the behavior is rather correct. I just didn't fully follow the state machine you wrote, that's a bit tricky to guess..

If all the matricies are restored correct now then let's call it bug was in old blender.

@mont29, we for sure don't want the dupli matrix affecting motherball or so and from your explanation it seems indeed the behavior is rather correct. I just didn't fully follow the state machine you wrote, that's a bit tricky to guess.. If all the matricies are restored correct now then let's call it bug was in old blender.

Bastien Montagne commented

2015-02-24 19:26:41 +01:00

Changed status from 'Open' to: 'Archived'

Bastien Montagne closed this issue

2015-02-24 19:26:41 +01:00

Bastien Montagne commented

2015-02-24 19:26:41 +01:00

Yep, code here is rather convoluted…

@EliSpizzichino Thanks for the report, but current behavior is correct one, and pre-2.72 one was indeed buggy (in that specific case we could call it a 'feature bug' I guess). Also, please note that hard limit of metaball resolution is smaller than 0.025 (think it’s 0.01), you can enter it by typing the value with the keyboard, should it be really needed.

Yep, code here is rather convoluted… @EliSpizzichino Thanks for the report, but current behavior is correct one, and pre-2.72 one was indeed buggy (in that specific case we could call it a 'feature bug' I guess). Also, please note that hard limit of metaball resolution is smaller than 0.025 (think it’s 0.01), you can enter it by typing the value with the keyboard, should it be really needed.

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

Emitted Metaball not showing and rendering correctly #43666