Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Repo info
Activity
  • 06:55
    jz0224 starred mosra/magnum
  • 01:21
    dec05eba starred mosra/magnum
  • Nov 30 18:55
    durexyl starred mosra/corrade
  • Nov 30 18:52
    durexyl starred mosra/magnum
  • Nov 30 15:33
    devpikachu commented #67
  • Nov 30 15:33
    devpikachu commented #67
  • Nov 30 05:53
    yunyuncc starred mosra/magnum
  • Nov 30 02:41
    aslights starred mosra/magnum
  • Nov 29 23:58
    wagnermc506 starred mosra/magnum
  • Nov 28 19:28
    mosra closed #116
  • Nov 28 19:28
    mosra commented #116
  • Nov 28 19:27

    mosra on master

    AssimpImporter: add missing sou… AssimpImporter: add a .gitattri… AssimpImporter: add light names and 28 more (compare)

  • Nov 28 18:22

    mosra on next

    package/ci: go back to Assimp 5… (compare)

  • Nov 28 18:09
    mosra edited #91
  • Nov 28 18:07
    mosra demilestoned #91
  • Nov 28 18:07
    mosra edited #91
  • Nov 28 18:07
    mosra edited #91
  • Nov 28 18:06
    mosra labeled #116
  • Nov 28 18:06
    mosra milestoned #116
  • Nov 28 17:58

    mosra on master

    Link to latest Assimp build. (compare)

Vladimír Vondruš
@mosra
ahahahahah
sounds like you're pretty deep in this, what was the original problem? :D
pezcode
@pezcode
I just wanted to add support for the scene name added in assimp/assimp#3510
but with master half the tests fail now and I'm about to give up
Vladimír Vondruš
@mosra
:see_no_evil:
that doesn't seem like a feature worth dying for :D
and i thought you did not want to touch assimp ever again?
pezcode
@pezcode
it seemed simple enough to spend 30 minutes on
no use putting all that work in until there's a new tagged release, anyway
Vladimír Vondruš
@mosra
if you want something else knee-deep in assimp, i have this half-done custom assimp material support WIP PR opened on the plugins repo, it "just" needs to find some files that the new codepaths could be tested with
pezcode
@pezcode
that's a lot of open checkboxes

that's a lot of open checkboxes

yeah exactly, and i gave up (they're apparently reworking this to be less shit), so from the PR i would only take the custom materials, which is the very last checkbox, nothing else

assimp has a clear set of types (nice) and predefined attribute name strings with varying level of shittiness (ew), which, if imported as-is (without trying to make sense of) could work as at least some way of providing access to, say, FBX PBR attributes
pezcode
@pezcode
does indeed sound useful
Vladimír Vondruš
@mosra
i want to do the same for gltf (extensions as layers, extras as custom attribs) because the new PBR extensions are appearing way faster than i'm able to integrate them :D
pezcode
@pezcode
most of them seem to need additions to MaterialAttribute, like IOR or transmission
sheen was the only one I saw that can be done with a layer (color + roughness)
Vladimír Vondruš
@mosra
yep, MaterialAttribute / MaterialLayer additions mostly
the thinfilm extension looked like a good candidate for a layer as well
pezcode
@pezcode
I've only looked at the released ones so far... is thin film for iridescence?
Vladimír Vondruš
@mosra
i don't remember where i saw the example renders (autodesk standard surface maybe? filament?) but thinfilm was for the green/blue tint that's on well-treated leather chairs
iridescence is an extension on its own i think
Vladimír Vondruš
@mosra
ah... KhronosGroup/glTF#2027 , which replaces the thinfilm extension
that explains why i thought it's two separate things
i'm a forever pbr noob
pezcode
@pezcode
they're moving really fast with these extensions, but no visible results
Vladimír Vondruš
@mosra
and yeah those pictures on that link are maybe exactly what i remember

no visible results

i already saw a dataset that used like 10 of those fancy new extensions

pezcode
@pezcode
what I meant is that they have like 10 in the pipeline that aren't finished, but keep changing names and feature sets, only confusing everyone
Vladimír Vondruš
@mosra
ah :) yeah, true
though honestly i like this better than a total silence for 5 years and then "here's glTF 2.1 with PBR Next, enjoy!" with no community involvement
i imagine this is how the internal vulkan extension development looks like as well, only we don't see it :D
pezcode
@pezcode
yeah, this is probably the better alternative
Vladimír Vondruš
@mosra
I just made a commit ccccc9d9281f779b59ff587568fe4f869dab9eb3, does that already count as a mined cryptocurrency or do I need more cs?
pezcode
@pezcode
don't give Microsoft ideas :eyes:
Vladimír Vondruš
@mosra
you mean further gamification of github? :D
Mathew
@lectroMathew
mining on git commits would be the cleanest way to ensure that github's servers go down as quickly as possible and never go up again
Fo Nz
@FoNz99089892_twitter

so myOwnDefaultFramebuffer has a non-zero ID, right? just double-checking i understand this properly

Exactly. Both "default" framebuffer and "default" renderbuffer have a ID of 1, in my case.

irfna
@irfna

Hey, I'm having some issues rendering some basic textured cubes that use the same texture for each side, I end up with weird artifacts that change with every tiny change in the camera angle:

https://i.imgur.com/VH6HOuG.png

Currently using phong shader to test but the FlatGL shader has the same problem. My current vertices are just a copy and paste/rename of cubeSolid() and its dependencies, except I shoved in texture coordinates:

constexpr struct TileBlockVertex2 {
    Vector3 position;
    Vector3 normal;
    Vector2 textureCoordinates;
} TileBlockVertices2[]{
    {{-1.0f, -1.0f,  1.0f}, { 0.0f,  0.0f,  1.0f}, { 0.0f,  0.0f}},
    {{ 1.0f, -1.0f,  1.0f}, { 0.0f,  0.0f,  1.0f}, { 1.0f,  0.0f}},
    {{ 1.0f,  1.0f,  1.0f}, { 0.0f,  0.0f,  1.0f}, { 1.0f,  1.0f}}, /* +Z */
    {{-1.0f,  1.0f,  1.0f}, { 0.0f,  0.0f,  1.0f}, { 0.0f,  1.0f}},

...

And so on, then I'm just using cubeSolid() as is shown in examples. When I test FlatGL, I have a version of the array without the normal vector in it but is otherwise the same. There has to be something small/dumb I'm not seeing that is resulting in these errors, but everything I'm attempting is failing

Vladimír Vondruš
@mosra
hi! could it be just .. the depth test not being enabled?
leading to the (black) backfaces sometimes rendered after the faces that should be in front
irfna
@irfna
    GL::defaultFramebuffer.clear(GL::FramebufferClear::Color|GL::FramebufferClear::Depth);

    ViewRegion();

    GL::Renderer::enable(GL::Renderer::Feature::DepthTest);

    sprites.draw(*_camera);
    terrain.draw(*_camera);
I have actually experimented without depth test, but then some sides and other things are randomly missing as well. There's something really wrong that I can't nail down
I've used various textured cube vertex data from random people online to try different versions, but there's always artifacts and bugs in the cube. I'm able to render sprites on planes just fine, so I'm not sure why cubes are being so silly
 Shaders::PhongGL shader{Shaders::PhongGL::Flag::DiffuseTexture};

 GL::Mesh mesh = MeshTools::compile(cubeSolid());
  auto size = Vector3(TilePixelDimensions) / 4.0f;

  category.shader
  .setTransformationMatrix(transformation*Matrix4::scaling(size))
  .setProjectionMatrix(camera.projectionMatrix())
  .bindDiffuseTexture(*texture)
  .draw(category.mesh);
There's nothing really complicated going on in the code either, but maybe I'm missing a necessary feature or something?
Vladimír Vondruš
@mosra
huh
what's your setup? is it just a pure magnum application or combined with some 3rd party lib that could potentially touch OpenGL as well?
irfna
@irfna
This is a pure magnum setup based on the basic bootstrap project, setup just for testing out rendering sprites and cubes
Vladimír Vondruš
@mosra
ah that could be rather easy to go through, can you post the whole code somewhere? gist, pastebin... i'll look at it