Page MenuHome
Create Task
Maniphest T61156

Calling Object.to_mesh on a Metaball object yields an empty mesh
Closed, ResolvedPublic
Actions

Description

This is broken in 2.80 but worked in 2.79. Repro script:

bpy.ops.object.metaball_add()
len(bpy.context.active_object.to_mesh(bpy.context.depsgraph, True).polygons) # will be 0

The issue is in BKE_mesh_new_from_object (blenkernel\intern\mesh_convert.c), where on line 958 we check for ob->runtime.curve_cache. It's null, which means we pass null to BKE_mesh_from_metaball, which then bails out without doing any work.

The convert operator works correctly.

Related Objects

Event Timeline

Tom Edwards (artfunkel) created this task.Feb 3 2019, 11:52 AM
Jacques Lucke (JacquesLucke) lowered the priority of this task from 90 to 50.Feb 4 2019, 12:09 PM
Philipp Oeser (lichtwerk) added a subscriber: Philipp Oeser (lichtwerk).Feb 4 2019, 12:24 PM

bandaid:

P902 T61156 snippet
1
2
3diff --git a/source/blender/blenkernel/intern/mesh_convert.c b/source/blender/blenkernel/intern/mesh_convert.c
4index a61d9da2c70..cd2e91fb29c 100644
5--- a/source/blender/blenkernel/intern/mesh_convert.c
6+++ b/source/blender/blenkernel/intern/mesh_convert.c
7@@ -955,9 +955,10 @@ Mesh *BKE_mesh_new_from_object(
8 }
9 else {
10 ListBase disp = {NULL, NULL};
11- if (ob->runtime.curve_cache) {
12- disp = ob->runtime.curve_cache->disp;
13+ if (!ob->runtime.curve_cache) {
14+ BKE_displist_make_mball(depsgraph, sce, ob);
15 }
16+ disp = ob->runtime.curve_cache->disp;
17 BKE_mesh_from_metaball(&disp, tmpmesh);
18 }
19

Philipp Oeser (lichtwerk) assigned this task to Sergey Sharybin (sergey).Feb 4 2019, 1:16 PM
Philipp Oeser (lichtwerk) added a subscriber: Sergey Sharybin (sergey).

hold on, or should we pass the evaluated object here in the first place?

ob = bpy.context.active_object
eval_ob = bpy.context.depsgraph.objects.get(ob.name, None)
len(eval_ob.to_mesh(bpy.context.depsgraph, True).polygons)

this works, it is unclear though if we should do this on the python side (or get the evaluated object on the C side -- since we are passing a depsgraph anyways...)

If we should get the evaluated object from the passed depsgraph, we could do here:

P903 T61156 snippet 2
1
2
3diff --git a/source/blender/makesrna/intern/rna_main_api.c b/source/blender/makesrna/intern/rna_main_api.c
4index a4c4acb8006..1da2a68fd0d 100644
5--- a/source/blender/makesrna/intern/rna_main_api.c
6+++ b/source/blender/makesrna/intern/rna_main_api.c
7@@ -309,6 +309,7 @@ Mesh *rna_Main_meshes_new_from_object(
8 Object *ob, bool apply_modifiers, bool calc_undeformed)
9 {
10 Scene *sce = DEG_get_evaluated_scene(depsgraph);
11+ Object *ob_eval = DEG_get_evaluated_object(depsgraph, ob);
12
13 switch (ob->type) {
14 case OB_FONT:
15@@ -322,7 +323,7 @@ Mesh *rna_Main_meshes_new_from_object(
16 return NULL;
17 }
18
19- return BKE_mesh_new_from_object(depsgraph, bmain, sce, ob, apply_modifiers, calc_undeformed);
20+ return BKE_mesh_new_from_object(depsgraph, bmain, sce, ob_eval, apply_modifiers, calc_undeformed);
21 }
22
23 static Lamp *rna_Main_lights_new(Main *bmain, const char *name, int type)

or even earlier here:

P904 T61156 snippet 3
1
2
3diff --git a/source/blender/makesrna/intern/rna_object_api.c b/source/blender/makesrna/intern/rna_object_api.c
4index 43f41f2704a..0e6c81aafcb 100644
5--- a/source/blender/makesrna/intern/rna_object_api.c
6+++ b/source/blender/makesrna/intern/rna_object_api.c
7@@ -217,8 +217,8 @@ static Mesh *rna_Object_to_mesh(
8 bool apply_modifiers, bool calc_undeformed)
9 {
10 Main *bmain = CTX_data_main(C);
11-
12- return rna_Main_meshes_new_from_object(bmain, reports, depsgraph, ob, apply_modifiers, calc_undeformed);
13+ Object *ob_eval = DEG_get_evaluated_object(depsgraph, ob);
14+ return rna_Main_meshes_new_from_object(bmain, reports, depsgraph, ob_eval, apply_modifiers, calc_undeformed);
15 }
16
17 static PointerRNA rna_Object_shape_key_add(Object *ob, bContext *C, ReportList *reports,

@Sergey Sharybin (sergey): what do you think? (reminds me of T60083 btw. where we argued that evaluated object should be done on the python side of things...)

Tom Edwards (artfunkel) added a comment.Mar 4 2019, 8:34 PM

This bug has been open a month without being addressed, so I'm going to use Phillip's suggested Python workaround. A fix before 2.80 ships would be appreciated!

Sergey Sharybin (sergey) changed the task status from Unknown Status to Resolved.May 17 2019, 2:05 PM

This is now addressed by the re-designed python API of the related calls. See the depsgraph python examples in the documentation.

If there are other issues which we did not spot yet please make a new report.

For until then considering the issue solved, closing.