replace polyvox with simpler implementation by RTUnreal · Pull Request #2026 · overte-org/overte

RTUnreal · 2026-01-14T18:45:43Z

This replaces the functionality of the polyvox library, so we don't have to ship it anymore. :)

This will to introduce a bit of a breaking change, where if borders are filled in, then the faces at the border is not rendered. But considering, that polyvox already did it at some of the sides, I tought it would be better to always not render faces there, so that tiling can be done in the future. This can be changed back if needed

ksuprynowicz

Thank you! Everything looks good :)
I only have minor feedback about some of the formatting changes, let me know what you think about it.

ksuprynowicz

Thank you! Everything looks good :)

ksuprynowicz · 2026-02-20T15:20:49Z

I'm getting an assert in Overte_hub:

[02/20 16:13:01] [FATAL] [default] ASSERT: "isInside(index)" in file /home/ksuprynowicz/overte/overte_opt/overte/libraries/entities-renderer/src/RenderablePolyVoxEntityItem.cpp, line 192

ksuprynowicz · 2026-02-20T15:24:50Z

Backtrace:

__pthread_kill_implementation 0x00007fffec29e95c
__pthread_kill_internal 0x00007fffec29e9ff
__GI_raise 0x00007fffec249cc2
__GI_abort 0x00007fffec2324ac
QMessageLogger::fatal(char const*, ...) const 0x00007ffff6a92514
qt_assert(const char *, const char *, int) 0x00007ffff6a915ae
VoxelVolume::getVoxelAt RenderablePolyVoxEntityItem.cpp:192
operator() RenderablePolyVoxEntityItem.cpp:614
loop3<…>(const glm::vec<…> &, const glm::vec<…> &, struct {...}) RenderablePolyVoxEntityItem.cpp:153
CubicSurfaceExtractorWithNormals::createMesh RenderablePolyVoxEntityItem.cpp:603
operator() RenderablePolyVoxEntityItem.cpp:1933
ReadWriteLockable::withReadLock<…>(struct {...} &&) const ReadWriteLockable.h:109
operator() RenderablePolyVoxEntityItem.cpp:1918
QtConcurrent::StoredFunctorCall0::runFunctor(void) qtconcurrentstoredfunctioncall.h:70
QtConcurrent::RunFunctionTask::run qtconcurrentrunbase.h:142
<unknown> 0x00007ffff6ae5e21
<unknown> 0x00007ffff6ae28fd
start_thread 0x00007fffec29cb7b
__GI___clone3 0x00007fffec31a7b8

JulianGro

The CMake and Conan changes are trivial and fine.
Thanks for adding the additional copyright notices, or whatever we want to call them.

Would be cool if someone else could review cb49bca

ksuprynowicz · 2026-03-28T16:34:54Z

The newest version hits an assert for me here:

__pthread_kill_implementation 0x00007fffec4a13bc
__pthread_kill_internal 0x00007fffec4a147f
__GI_raise 0x00007fffec44a942
__GI_abort 0x00007fffec4324ac
QMessageLogger::fatal(char const*, ...) const 0x00007ffff6a92514
qt_assert(const char *, const char *, int) 0x00007ffff6a915ae
VoxelVolume::getVoxelAt RenderablePolyVoxEntityItem.cpp:240
operator() RenderablePolyVoxEntityItem.cpp:894
loop3<…>(const glm::vec<…> &, const glm::vec<…> &, struct {...}) RenderablePolyVoxEntityItem.cpp:153
MarchingCubesSurfaceExtractor::createMesh RenderablePolyVoxEntityItem.cpp:883
operator() RenderablePolyVoxEntityItem.cpp:2064
ReadWriteLockable::withReadLock<…>(struct {...} &&) const ReadWriteLockable.h:109
operator() RenderablePolyVoxEntityItem.cpp:2049
QtConcurrent::StoredFunctorCall0::runFunctor(void) qtconcurrentstoredfunctioncall.h:70
QtConcurrent::RunFunctionTask::run qtconcurrentrunbase.h:142
<unknown> 0x00007ffff6ae5f01
<unknown> 0x00007ffff6ae29bd
start_thread 0x00007fffec49f469
__GI___clone3 0x00007fffec51dcf8

Index is {21, 21, 21}, valid size is {21, 21, 21} and allocated_size is {21, 21, 21}.

ksuprynowicz · 2026-03-28T18:25:05Z

__pthread_kill_implementation 0x00007fffec4a13bc
__pthread_kill_internal 0x00007fffec4a147f
__GI_raise 0x00007fffec44a942
__GI_abort 0x00007fffec4324ac
QMessageLogger::fatal(char const*, ...) const 0x00007ffff6a92514
qt_assert(const char *, const char *, int) 0x00007ffff6a915ae
VoxelVolume::getVoxelAt RenderablePolyVoxEntityItem.cpp:240
operator() RenderablePolyVoxEntityItem.cpp:894
loop3<…>(const glm::vec<…> &, const glm::vec<…> &, struct {...}) RenderablePolyVoxEntityItem.cpp:153
MarchingCubesSurfaceExtractor::createMesh RenderablePolyVoxEntityItem.cpp:883
operator() RenderablePolyVoxEntityItem.cpp:2064
ReadWriteLockable::withReadLock<…>(struct {...} &&) const ReadWriteLockable.h:109
operator() RenderablePolyVoxEntityItem.cpp:2049
QtConcurrent::StoredFunctorCall0::runFunctor(void) qtconcurrentstoredfunctioncall.h:70
QtConcurrent::RunFunctionTask::run qtconcurrentrunbase.h:142
<unknown> 0x00007ffff6ae5f01
<unknown> 0x00007ffff6ae29bd
start_thread 0x00007fffec49f469
__GI___clone3 0x00007fffec51dcf8

ksuprynowicz · 2026-03-29T20:57:32Z

__pthread_kill_implementation 0x00007fffec4a13bc
__pthread_kill_internal 0x00007fffec4a147f
__GI_raise 0x00007fffec44a942
__GI_abort 0x00007fffec4324ac
QMessageLogger::fatal(char const*, ...) const 0x00007ffff6a92514
qt_assert(const char *, const char *, int) 0x00007ffff6a915ae
VoxelVolume::getVoxelAt RenderablePolyVoxEntityItem.cpp:242
VoxelVolume::raycast RenderablePolyVoxEntityItem.cpp:305
operator() RenderablePolyVoxEntityItem.cpp:1338
ReadWriteLockable::withReadLock<…>(struct {...} &&) const ReadWriteLockable.h:109
RenderablePolyVoxEntityItem::findDetailedRayIntersection RenderablePolyVoxEntityItem.cpp:1338
operator() EntityTreeElement.cpp:255
operator() EntityTreeElement.h:163
ReadWriteLockable::withReadLock<…>(struct {...} &&) const ReadWriteLockable.h:109
EntityTreeElement::forEachEntity<…> EntityTreeElement.h:161
EntityTreeElement::evalDetailedRayIntersection EntityTreeElement.cpp:199
EntityTreeElement::evalRayIntersection EntityTreeElement.cpp:180
evalRayIntersectionOp EntityTree.cpp:811
std::__invoke_impl<…> invoke.h:63
std::__invoke_r<…> invoke.h:116
std::_Function_handler::_M_invoke std_function.h:292
std::function::operator() std_function.h:593
Octree::recurseElementWithOperationSorted Octree.cpp:108
Octree::recurseTreeWithOperationSorted Octree.cpp:96
operator() EntityTree.cpp:860
ReadWriteLockable::withReadLock<…>(struct {...} &&) const ReadWriteLockable.h:109
ReadWriteLockable::withReadLock<…>(struct {...} &&, bool) const ReadWriteLockable.h:124
EntityTree::evalRayIntersection EntityTree.cpp:859
EntityScriptingInterface::evalRayIntersectionWorker EntityScriptingInterface.cpp:1524
EntityScriptingInterface::evalRayIntersectionVector EntityScriptingInterface.cpp:1514
RayPick::getEntityIntersection RayPick.cpp:70
PickCacheOptimizer::update PickCacheOptimizer.h:94
PickManager::update PickManager.cpp:176
Application::update qsharedpointer_impl.h:307
Application::idle Application.cpp:1875
Application::event Application_Events.cpp:103
QApplicationPrivate::notify_helper(QObject *, QEvent *) 0x00007fffed364240
Application::notify Application_Events.cpp:61
QCoreApplication::notifyInternal2(QObject *, QEvent *) 0x00007ffff6ce9f88
QCoreApplicationPrivate::sendPostedEvents(QObject *, int, QThreadData *) 0x00007ffff6ced444
<unknown> 0x00007ffff6d48ba7
<unknown> 0x00007fffed1015ee
<unknown> 0x00007fffed10497f
g_main_context_iteration 0x00007fffed105110
QEventDispatcherGlib::processEvents(QFlags<…>) 0x00007ffff6d481ca
QEventLoop::exec(QFlags<…>) 0x00007ffff6ce88ba
QCoreApplication::exec() 0x00007ffff6cf146f
main main.cpp:850
__libc_start_call_main 0x00007fffec433f75
__libc_start_main_impl 0x00007fffec434027
_start 0x0000555555710c61

ksuprynowicz · 2026-03-29T21:35:37Z

It looks like it's missing smooth normals:

Current master branch:

ksuprynowicz · 2026-03-30T19:47:38Z

__pthread_kill_implementation 0x00007fffec4a13bc
__pthread_kill_internal 0x00007fffec4a147f
__GI_raise 0x00007fffec44a942
__GI_abort 0x00007fffec4324ac
QMessageLogger::fatal(char const*, ...) const 0x00007ffff6a92514
qt_assert(const char *, const char *, int) 0x00007ffff6a915ae
MarchingCubesSurfaceExtractor::createMesh RenderablePolyVoxEntityItem.cpp:984
operator() RenderablePolyVoxEntityItem.cpp:2114
ReadWriteLockable::withReadLock<…>(struct {...} &&) const ReadWriteLockable.h:109
operator() RenderablePolyVoxEntityItem.cpp:2099
QtConcurrent::StoredFunctorCall0::runFunctor(void) qtconcurrentstoredfunctioncall.h:70
QtConcurrent::RunFunctionTask::run qtconcurrentrunbase.h:142
<unknown> 0x00007ffff6ae5f01
<unknown> 0x00007ffff6ae29bd
start_thread 0x00007fffec49f469
__GI___clone3 0x00007fffec51dcf8

ksuprynowicz · 2026-03-30T20:12:45Z

Works perfectly now :)
I just need to test it with neighboured voxels

ksuprynowicz · 2026-04-04T18:45:10Z

Master branch:

2026:

ksuprynowicz · 2026-04-12T09:32:13Z

I tested a bit more and noticed an issue with meshing. Maybe there's different threshold?
Original:

This PR:

RTUnreal · 2026-04-18T13:27:33Z

import itertools

def rotr_edge(a):
    assert a in range(12)
    if a == 0: return 1
    elif a == 1: return 2
    elif a == 2: return 3
    elif a == 3: return 0
    elif a == 4: return 5
    elif a == 5: return 6
    elif a == 6: return 7
    elif a == 7: return 4
    elif a == 8: return 9
    elif a == 9: return 10
    elif a == 10: return 11
    elif a == 11: return 8
    assert False

def rotr_vert(a):
    assert a in range(8), f"{a}"
    if a == 0: return 1
    elif a == 1: return 2
    elif a == 2: return 3
    elif a == 3: return 0
    elif a == 4: return 5
    elif a == 5: return 6
    elif a == 6: return 7
    elif a == 7: return 4
    assert False

def mirroru_edge(a):
    assert a in range(12)
    if a == 0: return 4
    elif a == 1: return 5
    elif a == 2: return 6
    elif a == 3: return 7
    elif a == 4: return 0
    elif a == 5: return 1
    elif a == 6: return 2
    elif a == 7: return 3
    elif a == 8: return 8
    elif a == 9: return 9
    elif a == 10: return 10
    elif a == 11: return 11
    assert False

def mirroru_vert(a):
    assert a in range(8)
    if a == 0: return 4
    elif a == 1: return 5
    elif a == 2: return 6
    elif a == 3: return 7
    elif a == 4: return 0
    elif a == 5: return 1
    elif a == 6: return 2
    elif a == 7: return 3
    assert False

def ltoi(l):
    assert all(map(lambda i: i in range(8), l))
    r = 0
    for i in l:
        r = r|(1<<i)
    return r

def itol(v):
    assert v in range(256)
    r = []
    for i in range(8):
        if v&1 != 0:
            r.append(i)
        v = v>>1
    return r

def list_group(v,rot=4,mirror=True):
    out = []
    e = []
    for _ in range(rot):
        out.append(v)
        v,e = apply_trans(rotr_vert, rotr_edge, v, e)
    if mirror:
        v,e = apply_trans(mirroru_vert, mirroru_edge, v, e)
        for _ in range(rot):
            out.append(v)
            v,e = apply_trans(rotr_vert, rotr_edge, v, e)
    return out

RESULTS = [None for _ in range(256)]

def add_res(v, e, expect_dup = False):
    global RESULTS

    assert expect_dup or RESULTS[ltoi(v)] is None
    assert all(map(lambda a: a in range(12), e))

    RESULTS[ltoi(v)] = e + [-1] * (16 - len(e))

def apply_trans(dv, de, v, e):
    return list(map(dv,v)), list(map(de, e))


def repeat_group(v,e,rot=4,mirror=True, expect_dup_mirror=False):
    assert len(e) % 3 == 0
    for _ in range(rot):
        add_res(v,e)
        v,e = apply_trans(rotr_vert, rotr_edge, v, e)

    if mirror:
        v,e = apply_trans(mirroru_vert, mirroru_edge, v, e)
        e = list(itertools.chain.from_iterable(map(lambda t: [t[0],t[2],t[1]], [e[i:i+3] for i in range(0, len(e), 3)])))
        for _ in range(rot):
            add_res(v,e, expect_dup=expect_dup_mirror)
            v,e = apply_trans(rotr_vert, rotr_edge, v, e)


#print(itol(min(map(lambda v:ltoi(v), list_group([0,1,4,5,3,6])))))
#exit(0)

add_res([],[])
add_res(itol(255), [])
repeat_group([0], [0,8,3])
repeat_group([0, 1], [3,1,8, 8,1,9])
repeat_group([0, 4], [3,0,4, 3,4,7], mirror=False)
repeat_group([0, 5], [3,4,8, 3,5,4, 3,0,5, 0,9,5])
repeat_group([0, 6], [0,8,3, 5,6,10], expect_dup_mirror=True)
repeat_group([0, 2], [0,8,1, 1,8,10, 3,10,8, 3,2,10], rot=2)
repeat_group([0, 1, 2], [8,10,9, 3,10,8, 3,2,10])
repeat_group([0, 1, 2, 3], [8,10,9, 8,11,10], rot=1)
repeat_group([0, 1, 4], [3,1,7, 7,9,4, 9,7,1])
repeat_group([0, 2, 4], [10,4,7, 2,10,7, 3,2,7, 10,1,4, 1,0,4])
repeat_group([1, 2, 4], [2,10,7, 7,8,2, 8,0,2, 7,10,9, 4,7,9])
repeat_group([0, 1, 2, 4], [2,10,9, 2,9,7, 2,7,3, 7,9,4])
repeat_group([0, 3, 4], [11,4,7, 11,2,4, 2,0,4])
repeat_group([1, 3, 4], [0,3,8, 2,7,11, 1,9,4, 2,1,7, 7,1,4])
repeat_group([0, 1, 3, 4], [2,7,11, 1,9,4, 2,1,7, 7,1,4])
#repeat_group([0, 1, 6], [3,6,8, 3,10,6, 1,10,3, 6,5,8, 8,5,9])
repeat_group([0, 1, 6], [3,6,8, 3,1,6, 1,10,6, 6,9,8, 6,5,9])
repeat_group([0, 1, 2, 6], [3,9,8, 3,2,9, 2,6,9, 9,6,5])
repeat_group([1, 2, 3, 4], [3,8,0, 11,4,7, 11,9,4, 11,10,9])
repeat_group([0, 1, 2, 3, 4], [11,4,7, 11,9,4, 11,10,9])
repeat_group([0, 1, 4, 5], [3,5,7, 3,1,5], mirror=False)
repeat_group([0, 2, 4, 5], [0,9,1, 3,5,7, 3,2,5, 2,10,5])
#repeat_group([1, 2, 4, 5], [0,7,8, 2,10,5, 0,5,7, 5,0,2])
repeat_group([1, 2, 4, 5], [2,10,0, 0,10,8, 8,10,7, 7,10,5])
repeat_group([0, 1, 2, 4, 5], [3,5,7, 3,10,5, 3,2,10])
repeat_group([1, 3, 4, 5], [3,8,0, 7,11,2, 2,1,7, 1,5,7])
repeat_group([0, 1, 3, 4, 5], [1,5,7, 11,2,7, 7,2,1])
repeat_group([2, 3, 4, 5], [7,11,5, 5,11,10, 3,8,1, 1,8,9], mirror=False)
repeat_group([0, 1, 3, 5, 6], [8,11,4, 4,11,6, 1,10,2])
repeat_group([1, 2, 3, 4, 5], [3,8,0, 11,5,7, 11,10,5])
repeat_group([0, 1, 2, 3, 4, 5], [11,5,7, 11,10,5])
repeat_group([0, 2, 4, 6], [3,6,7, 3,2,6, 1,0,4, 1,4,5], rot=2, mirror=False)
repeat_group([0, 1, 2, 4, 6], [9,4,5, 3,6,7, 3,2,6])
repeat_group([1, 3, 4, 6], [3,8,0, 2,1,10, 11,6,7, 9,4,5], rot=2, mirror=False)
repeat_group([0, 1, 3, 4, 6], [1,10,2, 9,4,5, 11,6,7])
repeat_group([0, 1, 2, 3, 4, 6], [11,6,7, 9,4,5], rot=2)
repeat_group([0, 1, 2, 4, 5, 6], [3,6,7, 3,2,6], mirror=False)
#repeat_group([0, 1, 3, 4, 5, 6], [11,1,7, 11,2,1, 7,1,10, 7,10,6])
repeat_group([0, 1, 3, 4, 5, 6], [11,6,7, 1,10,2])
repeat_group([0, 2, 3, 4, 5, 6], [11,6,7, 0,9,1], mirror=False)
repeat_group([0, 1, 2, 3, 4, 5, 6], [11,6,7])


done = all(map(lambda x: x is not None, RESULTS))
if done:
    print("{ ", end='')
    for el in RESULTS:
        print("{ ", end='')
        for ii, i in enumerate(el):
              if ii > 0:
                  print(", ", end='')
              print(i, end='')
        print(" },")
    print("}")
else:
    print(RESULTS, sum(map(lambda x: int(x is not None), RESULTS)), done)

    next_none = next((i for i, x in enumerate(RESULTS) if x is None))

    print(next_none, itol(next_none))

If there is an interest, here is the script I used to generate the new table (the commented-out repeat_group calls are my attempt at a table).

ksuprynowicz · 2026-04-18T14:15:36Z

The newest version works perfectly :)
Normals are also very nice improvement over stable branch:
Original:

PR 2026:

ksuprynowicz

Thank you!
There's one typo, but all is good otherwise :)

RTUnreal requested review from JulianGro and ksuprynowicz January 14, 2026 19:45

RTUnreal added needs CR This pull request needs to be code reviewed needs QA This pull request needs to be tested labels Jan 21, 2026

ksuprynowicz requested changes Jan 25, 2026

View reviewed changes

RTUnreal requested a review from ksuprynowicz January 25, 2026 11:11

ksuprynowicz approved these changes Feb 13, 2026

View reviewed changes

ksuprynowicz added CR approved This pull request has been successfully code reviewed and removed needs CR This pull request needs to be code reviewed labels Feb 13, 2026

JulianGro added needs CR This pull request needs to be code reviewed and removed CR approved This pull request has been successfully code reviewed labels Mar 20, 2026

RTUnreal added 6 commits March 20, 2026 13:42

replace polyvox with simpler implementation

f916b4a

remove polyvox as project dependencies

4fefc97

make cubic surfacemesher not render border

b3d1570

remove unused debug code

16b92df

some formatting fixup

20349be

make code hopefully more reliable

cb49bca

RTUnreal force-pushed the remove_polyvox branch from f81956c to cb49bca Compare March 20, 2026 12:43

RTUnreal added 2 commits March 20, 2026 14:06

add notice of source for code

bae9079

improve copyright notice

5d09007

JulianGro approved these changes Mar 20, 2026

View reviewed changes

replicate old behavior for now

228e588

ksuprynowicz reviewed Mar 27, 2026

View reviewed changes

Comment thread libraries/entities-renderer/src/RenderablePolyVoxEntityItem.cpp

RTUnreal added 2 commits March 27, 2026 23:47

libraries/entities-renderer: add ensureAllocation docs for explanation

ee25928

libraries/entities-renderer: docs try 2

216eb9d

ksuprynowicz reviewed Mar 27, 2026

View reviewed changes

Comment thread libraries/entities-renderer/src/RenderablePolyVoxEntityItem.cpp Outdated

test fix

106c880

add more output of check in marching cubes

1d91bb8

post debugging cleanup

471a3d2

deduplicate and take the mean of normals in marching cubes

fd6da9d

try fixxing assert in dedup code

59a37fb

make edge connect with neighbor work

14100c5

RTUnreal added 3 commits April 4, 2026 20:53

use pointers instead of invalidId check

7ae14e0

make edged voxel objects work

58bc265

fixing try 2

ecfeb3b

RTUnreal added 3 commits April 18, 2026 15:11

use custom triTable to match polyvox behavior

c7d393e

normalize the vectors correctly

aa0ace8

add some explanatory comments

4f6beaa

RTUnreal force-pushed the remove_polyvox branch from 92bf88d to 4f6beaa Compare April 18, 2026 13:13

ksuprynowicz added QA approved This pull request has been successfully tested and removed needs QA This pull request needs to be tested labels Apr 18, 2026

fix graph

c6511e7

ksuprynowicz approved these changes Apr 18, 2026

View reviewed changes

Comment thread libraries/entities-renderer/src/RenderablePolyVoxEntityItem.cpp Outdated

fix typo

b95722f

ksuprynowicz merged commit 0523faf into overte-org:master Apr 18, 2026
5 checks passed

RTUnreal deleted the remove_polyvox branch April 18, 2026 17:43

Uh oh!

Conversation

RTUnreal commented Jan 14, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

ksuprynowicz left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

ksuprynowicz left a comment

Choose a reason for hiding this comment

Uh oh!

ksuprynowicz commented Feb 20, 2026

Uh oh!

ksuprynowicz commented Feb 20, 2026

Uh oh!

JulianGro left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

ksuprynowicz commented Mar 28, 2026

Uh oh!

ksuprynowicz commented Mar 28, 2026

Uh oh!

ksuprynowicz commented Mar 29, 2026

Uh oh!

ksuprynowicz commented Mar 29, 2026

Uh oh!

ksuprynowicz commented Mar 30, 2026

Uh oh!

ksuprynowicz commented Mar 30, 2026

Uh oh!

ksuprynowicz commented Apr 4, 2026

Uh oh!

ksuprynowicz commented Apr 12, 2026

Uh oh!

RTUnreal commented Apr 18, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

ksuprynowicz commented Apr 18, 2026

Uh oh!

ksuprynowicz left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

RTUnreal commented Jan 14, 2026 •

edited

Loading

RTUnreal commented Apr 18, 2026 •

edited

Loading