/gdx-gltf

GLTF 2.0 3D format support and PBR shader implementation for LibGDX

Primary LanguageJavaApache License 2.0Apache-2.0

status GitHub release (latest SemVer including pre-releases) Release GitHub release (latest SemVer including pre-releases)

LibGDX GLTF 2.0 support and PBR shader implementation. Alternative to libGDX G3D format.

Introduction

What's glTF befenits over G3D/FBX in libGDX?

  • Simpler workflow : no fbx-conv required, you can load gltf files directly.
  • Load cameras, lights, custom properties from Blender and other 3D softwares.
  • Shape keys / Animated shape keys (aka MorphTarget) feature.
  • Multiple animations playback
  • Non linear animations keyframes interpolation ("step" and "cubic" supported)
  • Out of the box shaders for normal maps, metallic/roughness, Image based lighting (IBL) and more.
  • Texture coordinates transform.
  • 64k vertices supported (instead of 32k)
  • Faster loading time, see benchmark

What's more than a 3D format parser in gdx-gltf library?

  • Scene management facility : Sky box, shadows, and more.
  • Physic Based Rendering (PBR) shaders : for realistic (or not) high quality rendering.
  • Spot light support.
  • Export various objects to glTF file (whole scene, model, mesh, etc).

Can i only load glTF files and use them with regular libgdx 3D API?

  • Yes, it's the same API, only materials differs : by default gdx-gltf uses its own shader (PBR) to enable all glTF features.
  • Note that libgdx default shader doesn't implements spot lights.
  • If you don't want/need high quality rendering (PBR), you still can configure loaders to use libgdx materials (and libgdx DefaultShader).

Demo and gallery

Library demo (aka model viewer) is available for several platforms:

Games made with this library:

GL Transmission Format (glTF) 2.0 Support

Implementation based on official glTF 2.0 Specification

Shaders inspired by glTF-WebGL-PBR demo :

GLTF extensions implemented:

Getting started

Compatibility

  • gdx-gltf 1.x requires libGDX 1.9.10+
  • gdx-gltf 2.x requires libGDX 1.9.11+

Before upgrading your gdx-gltf version, please read changes history.

Install

gdx-gltf is available via Jitpack.

ensure you have jitpack repository declared in your Gradle configuration and add a gltfVersion variable.

Version can be any release (latest release is recommended) or master-SNAPSHOT

allprojects {
	repositories {
		...
		maven { url 'https://jitpack.io' }
	}
	ext {
        ...
        gltfVersion = 'master-SNAPSHOT'
    }
}

Add dependency in your core project (replace master-SNAPSHOT by latest release to use a stable version) :

project(":core") {
    dependencies {
    	...
        api "com.github.mgsx-dev.gdx-gltf:gltf:$gltfVersion"
    }
}

For GWT (html) projects you need to add source dependency and inherit GWT module in your core .gwt.xml file.

project(":html") {
    dependencies {
    	...
        api "com.github.mgsx-dev.gdx-gltf:gltf:$gltfVersion:sources"
    }
}
<module>
	<inherits name='GLTF' />
	...
</module>

Loading asset files

Directly

SceneAsset sceneAsset = new GLTFLoader().load(Gdx.files.internal("myModel.gltf"));
SceneAsset sceneAsset = new GLBLoader().load(Gdx.files.internal("myModel.glb"));

Using AssetManager loaders

assetManager.setLoader(SceneAsset.class, ".gltf", new GLTFAssetLoader());
assetManager.setLoader(SceneAsset.class, ".glb", new GLBAssetLoader());
...
assetManager.load("myModel.gltf");
...
SceneAsset sceneAsset = assetManager.get("myModel.gltf", SceneAsset.class);

Render models

This library provides a convenient scene manager to handle glTF models and PBR shader.

see few provided examples for more information:

For advanced usage, please read full documentation

Export objects from libgdx

This library provides convenient methods to export various object type to glTF file. For instance, You can create some mesh programmatically in libgdx and export them to glTF files and optionally import them in Blender:

new GLTFExporter().export(model, Gdx.files.local("myModel.gltf")

You can also export a scene with its lights and camera. All gltf features are supported for export: animations, bones, etc. Note that only "gltf separate files" mode is currently supported for export.

Export models from Blender

As Blender 2.80, glTF exporter addon is included and enabled by default.

Image Based Lighting (IBL)

Demo is shipped with a pre-generated lighting environment. If you want to use others or generate them yourself, please read IBL guide Alternatively this library provide some quick IBL generators, it's not as accurate as HDRI based IBL but can be useful to quickly setup a lighting environement. see IBLBuilder class.

More about the library

Project structure

This repository is made of a library and a demo :

  • gltf library module (LibGDX extension). see gdx-gltf readme for futher information.
  • demo folder contains a LibGDX demo project with usual modules (core, desktop, android, html, ...) see gdx-gltf-demo readme for futher information.

Limitations

Mesh limitations

Both LibGDX and gdx-gltf supports unsigned short indices : a mesh is then limited to 65535 vertices. Both signed int and unsigned int indices are not supported (and not supported by all GPUs anyway).

WebGL limitations

LibGDX Pixmap loading from binary data is not supported by its GWT emulation. So, GLTF embeded and binary formats are not supported for html/WebGL target.

CubeMap seamless limitations

CubeMap seamless is a feature that perform correct cubeMap filtering. Without this feature, you may notice some artifacts for materials with high roughness since these artifacts are more visible at low resolution. The library automatically enable this feature when possible, based on these conditions :

  • For GLES 2 and WebGL 1, this feature is not supported at all.
  • For GLES 3 and WebGL 2, this feature is always supported and always activated.
  • For Desktop, it can be enabled if platform supports OpenGL 3.2+ or GL_ARB_seamless_cube_map

Troubleshooting

Max uniforms: Constant register limit exceeded, do not fit in N vectors

You may encounter this shader compilation error in case too many uniform needed on current hardware.

Constant register limit exceeded at ... more than 1024 registers needed to compiled program

or

Error: uniform variables in vertex shader do not fit in 256 vectors.

It typically means you may have too many bones. A single bone takes 4 uniforms (mat4), desktop GPU typically supports 1024 uniforms and lowend mobile 256 uniforms. That mean you should keep bones count under 50 per skeleton.

Max vertices: high index detected

You may encounter high index detected warnings or errors.

It means you may have too many vertices in a mesh. Try to reduce or split them before exporting to GLTF :

  • 64k vertices are fully supported.
  • more vertices are not supported at all.

Note that this limitation is per mesh, not for a whole scene.

Note that Blender vertex count can be misleading because exported geometry may contains more vertices because of normal split, texture coordinates split or vertex color split. Note that Blender glTF exporter split meshes for you to keep your vertices count below 64k.

Tangent vertex attributes

Without tangents, some old GPUs / OpenGL version, require an extension which may not be available.

In this case you'll get the following error : GL_OES_standard_derivatives extension or tangent vertex attribute required

It's then highly recommended to always provide tangent vertex attributes, it improve performances as well.

Max vertex attributes : too many vertex attributes

You may encounter this error if you have too many vertex attributes in one of your mesh.

Most GPU support up to 16 vertex attributes

This limit can be quickly reached depending on mesh information :

  • a_position: 1
  • a_normal: 1
  • a_tangent: 1 (optional)
  • a_color: 1 (optional)
  • a_texCoordX: up to 2 UVs layers
  • a_boneWeightX: up to 8 bones influences
  • a_positionX: up to 8 positions
  • a_normalX: up to 8 normals
  • a_tangentX: up to 8 tangents