Animations stretching geometry - three.js

I read that the latest threejs exporter added support for multiple animations so I created a quick experiment and exported a model with the latest plugin..after a bit of playing around with the code I managed to get each of the animations within the model playing, however the geometry is getting skewed horribly.
I'm pretty sure the issue is related to the following error: "gl.getProgramInfoLog() WARNING: Could not find vertex shader attribute 'position' to match BindAttributeLocation request." but not positive. Maybe I'm just not aware of something, such as a particular blender export flag configuration or something else that needs to be set within the code. Here's the important parts:
for (var i = 0; i < materials.length; i++) {
var mat = materials[i];
mat.skinning = true;
}
scene.add(mesh);
// add animation data to the animation handler
var animations = [];
for(var a in mesh.geometry.animations) {
THREE.AnimationHandler.add(mesh.geometry.animations[a]);
var animation = new THREE.Animation(mesh,mesh.geometry.animations[a].name);
animations.push(animation);
}
animations[0].play();
Any ideas guys? Here's a screenshot: https://drive.google.com/file/d/0B6zdGNflpdJLSVVHemszWWFTTHc/edit?usp=sharing

Related

How do I load texture files designed for an FBX model so that the model appears correctly in my ThreeJS scene?

I have been buying animated 3D models from TurboSquid and they fall into two starkly different categories:
Ready to use - I just load the FBX file and it looks fine in my ThreeJS scene
Missing textures - The loaded FBX model looks geometrically fine and the animations work, but all the surfaces are white
I want to know how to load the provided textures using the FBXLoader module. I have noticed that all of the FBX models, which I believe are PBR material based, have this set of files:
*002_BaseColor.png
*002_Emissive.png
*002_Height.png
*002_Metallic.png
*002_Roughness.png
*002_Normal.png
I have these questions:
How would I use the FBXLoader module to load the textures "in the right places"? In other words, how do I make the calls so that the BaseColor texture goes to the right place, the Emissive texture so it goes where it should, etc?
Do I need to pass it a custom material loader to the FBXLoader and if so, which one and how?
UPDATE: I trace through the ThreeJS FBX loader code I discovered that the FBX model that is not showing any textures does not have a Texture node at all. The FBX model that does show the textures does, as you can see in this screenshot of the debugger:
I'll keep tracing. If worse comes to worse I'll write custom code to add a Textures node with the right fields since I don't know how to use Blender to add the textures there. My concern is of course that even if I create a Textures node at run-time, they still won't end up in the proper place on the model.
This is the code I am currently using to load FBX models:
this.initializeModel = function (
parentAnimManagerObj,
initModelArgs= null,
funcCallWhenInitialized = null,
) {
const methodName = self.constructor.name + '::' + `initializeModel`;
const errPrefix = '(' + methodName + ') ';
self.CAC.parentAnimManagerObj = parentAnimManagerObj;
self.CAC.modelLoader = new FBXLoader();
self.CAC.modelLoader.funcTranslateUrl =
((url) => {
// Fix up the texture file names.
const useRobotColor = misc_shared_lib.uppercaseFirstLetter(robotColor);
let useUrl =
url.replace('low_Purple_TarologistRobot', `Tarologist${useRobotColor}`);
return useUrl;
});
// PARAMETERS: url, onLoad, onProgress, onError
self.CAC.modelLoader.load(
MODEL_FILE_NAME_TAROLOGIST_ROBOT_1,
(fbxObj) => {
CommonAnimationCode.allChildrenCastAndReceiveShadows(fbxObj);
fbxObj.scale.x = initModelArgs.theScale;
fbxObj.scale.y = initModelArgs.theScale;
fbxObj.scale.z = initModelArgs.theScale;
self.CAC.modelInstance = fbxObj;
// Set the initial position.
self.CAC.modelInstance.position.x = initModelArgs.initialPos_X;
self.CAC.modelInstance.position.y = initModelArgs.initialPos_Y;
self.CAC.modelInstance.position.z = initModelArgs.initialPos_Z;
// Create an animation mixer for this model.
self.CAC.modelAnimationMixer = new THREE.AnimationMixer(self.CAC.modelInstance);
// Speed
self.CAC.modelAnimationMixer.timeScale = initModelArgs.theTimeScale;
// Add the model to the scene.
g_ThreeJsScene.add(self.CAC.modelInstance);
// Don't let this part of the code crash the entire
// load call. We may have just added a new model and
// don't know certain values yet.
try {
// Play the active action.
self.CAC.activeAction = self.CAC.modelActions[DEFAULT_IDLE_ANIMATION_NAME_FOR_TAROLOGIST_ROBOT_1];
self.CAC.activeAction.play();
// Execute the desired callback function, if any.
if (funcCallWhenInitialized)
funcCallWhenInitialized(self);
} catch (err) {
const errMsg =
errPrefix + misc_shared_lib.conformErrorObjectMsg(err);
console.error(`${errMsg} - try`);
}
},
// onProgress
undefined,
// onError
(err) => {
// Log the error message from the loader.
console.error(errPrefix + misc_shared_lib.conformErrorObjectMsg(err));
}
);
}

Aframe Load lightmap after loading GLTF - lightmap not showing

I'm trying to add a lightmap to some mesh after loading them from a GLTF file.
All my objects have 2UV channel.
I'm waiting 'object3dset' and here is my code :
const mesh = this.el.getObject3D('mesh');
var textureLoader = new THREE.TextureLoader();
textureLoader.load("lightmap.png", function(lmap){
mesh.traverse((node) => {
if (!node.isMesh) return;
node.material.lightMap = lmap;
lmap.flipY = node.material.map.flipY; //needed to flip the texture
node.material.needsUpdate = true;
});
});
If I replace the material with a new one and set the lightmap, it's working.
But I want to find a way without recreating all materials.
The lightmap was loaded, but not easy to see.
By default metalness from Khronos Blender Exporter converted in threejs after loading GLTF result to a level 1.0. With this configuration, the lightmap is hard to see and is not corresponding to what we see in Blender.
I hope my mistake can help someone else losing too much time.

THREE.js repeating UV texture using JSONLoader

I need help for getting UV Textures to be rendered correctly in three.js. I've created a model with repeating tile texture in Blender. The tile texture is applied using UV mapping, and it looks like this if it is being rendered correctly:
Render image using Blender
.However, when it is loaded using JSONLoader by three.js, the tiles are just stretched to fill each polygon, giving weird result like this:Screenshot of render using three.js
. I've tried setting THREE.RepeatWrapping in my code but nothing changed:
bodyLoader = new THREE.JSONLoader();
bodyLoader.load('./starofthesea_threejs.json', function(geometry, materials) {
mainBodyMaterials = new THREE.MeshFaceMaterial(materials);
console.log(materials);
mainBodyMaterials.wrapS = THREE.RepeatWrapping;
mainBodyMaterials.wrapT = THREE.RepeatWrapping;
mainBodyMaterials.needsUpdate = true;
mainBody = new THREE.Mesh(geometry, mainBodyMaterials);
mainBody.traverse ( function (child) {
if (child instanceof THREE.Mesh) {
child.castShadow = true;
child.receiveShadow = true;
}
});
mainBody.scale.x = mainBody.scale.y = mainBody.scale.z = 1;
geometry.computeBoundingBox();
geometry.computeFaceNormals();
geometry.computeFlatVertexNormals();
scene.add(mainBody);
});
Is there anything wrong in my code, or workaround to get it rendered correctly? All help is deeply appreciated.
Finally I've figured out the problem by myself, where both the Blender model and JS are misconfigured. RepeatWrapping should be applied to texture but not material. I need to study the structure of THREE.MeshFaceMaterial to find the handle for the underlying textures. I need to traverse through the materials to find out all materials with image textures:
mainBodyMaterials = new THREE.MeshFaceMaterial(materials);
for(prop in mainBodyMaterials.materials) {
if(mainBodyMaterials.materials[prop].map != null) {
mainBodyMaterials.materials[prop].map.wrapS = THREE.RepeatWrapping;
mainBodyMaterials.materials[prop].map.wrapT = THREE.RepeatWrapping;
tex.push(mainBodyMaterials.materials[prop].map.clone());
tex[tex_sequence].needsUpdate = true;
tex_sequence++;
}
}
After applying wrapS and wrapT to textures correctly, one of the tile materials get rendered correctly, but 'Texture marked for update but image is undefined' error keep throwing out. I need to clone the texture to get rid of the error, according to another question: Three.js r72 - Texture marked for update but image is undefined?
As there are several materials with repeating tiles, I need to create a global array at the beginning of the JS routine, and push the modified textures one by one when looping through the materials:
var tex = new Array();
var tex_sequence = 0;
After fixing the JS calls, one of the textures are still not working. I forgot that only ONE UV slot is allowed for three.js. I need to unwrap the UVs again under the same UV slot in Blender. Everything works like a charm, and I hope my painful experience can help those who are getting mad by similar problems.

ILNumerics: Export part of a scene as image without affecting original scene

I have created a Scene with ILNumerics that consists of 3 PlotCubes and a Colorbar.
Screenshot of the ILPanel
I wanted to add a method that exports the scene as an image in two ways, the first being the screenshot you see above.
The second export is supposed to only show the centric cube.
I attempted to follow the guidelines of ILNumerics for scene management.
I wrote the following code:
public void ExportAsImage(int resolutionWidth, int resolutionHeight, string path, bool includeSubCubes)
{
using (ILScope.Enter())
{
ILGDIDriver backgroundDriver = new ILGDIDriver(resolutionWidth, resolutionHeight, ilPanel1.Scene);
if (includeSubCubes)
{
// code for standard export here
}
else
{
// setting left and top cube and color bar invisible and
// adjusting main cube size is affecting the ilPanel.Scene
backgroundDriver.Scene.First<ILColorbar>().Visible = false;
GetElementByTag<ILPlotCube>(backgroundDriver.Scene, _leftCubeTag).Visible = false;
GetElementByTag<ILPlotCube>(backgroundDriver.Scene, _topCubeTag).Visible = false;
GetElementByTag<ILPlotCube>(backgroundDriver.Scene, _mainCubeTag).ScreenRect = new RectangleF(0, 0, 1, 1);
GetElementByTag<ILPlotCube>(backgroundDriver.Scene, _mainCubeTag).DataScreenRect = new RectangleF.Empty;
backgroundDriver.Scene.Configure();
backgroundDriver.Render();
// save image
backgroundDriver.BackBuffer.Bitmap.Save(path,System.Drawing.Imaging.ImageFormat.Png);
// revert changes done to cubes and color bar
backgroundDriver.Scene.First<ILColorbar>().Visible = true;
GetElementByTag<ILPlotCube>(backgroundDriver.Scene, _leftCubeTag).Visible = true;
GetElementByTag<ILPlotCube>(backgroundDriver.Scene, _topCubeTag).Visible = true;
AdjustCubeSizes();
}
}
}
Note: "GetElementByTag" is an own implementation to retrieve objects in the ILNumerics Scene.
I first expected that the new driver basically creates a copy of the Scene I can work on, but like the code shows I have to revert all changed after the export or the displayed ilPanel only shows the scene the way I exported it.
Is it possible at all to do export to image without affecting the real Scene? Am I just missing some details?
Regards,
Florian S.
Florian, it does make a copy. But you need to add the interesting part to a new scene. The magic is happening in the Add() method:
var scene4render = new ILScene();
scene4render.Add(oldScene.First<ILPlotCube>(mytag));
// ... configure scene4render here, it will be detached from the original scene
// with the exception of shared buffers.
// ... proceed with rendering
In order to also include + render interactive state changes to the original plot cube (let's say rotations by the users mouse) you'd use something like that:
scene4render.Add(panel.SceneSyncRoot.First<ILPlotCube>(mytag));
Also, I wonder what GetElementByTag does better than ILGroup.First<T>(tag, predicate) or ILGroup.Find<T>(...)?
See also: http://ilnumerics.net/scene-management.html

Three.js Hemisphere Light

I`m examining a different types of light with dat.GUI and three.js (r72) and get stucked on dinamically turning on/off the HemisphereLight.
I have a one instance of each light being added to the scene:
var pointLight = new THREE.PointLight(this._whiteColor);
pointLight.visible = false;
var hemisphereLight = new THREE.HemisphereLight(this._whiteColor, this._whiteColor);
hemisphereLight.visible = false;
ect...
and a buttons for turning on/off with simple handler. Like this:
me._sceneObjects.hemisphereLight.visible = value;
So, before rendering all lights are present in the scene, but not visible.
When executing a handler for Hemisphere light - it stay invisible. After excluding hemisphereLight.visible = false; works fine.
Currently I`m disabling this light after rendering a 1st frame:
function render() {
some code ...
if (!sentenceExecuted && firstCall > 0) {
hemisphereLight.light.visible = false;
sentenceExecuted = true;
} else {
firstCall++;
}
};
I will be grateful for any suggestions to fix this without workaround.
Sorry for possible mistakes in English.
WestLangley, thank you for the link to Wiki article.
After reading and trying the examples I have found my mistake.
Originally I wanted to add "invisible" (using Object3D.visible) for render lights into a scene and then turn them on/off dynamically.
Some lights (in my case HemisphereLight and SpotLight) do not work this way because of internal three.js rendering algorythm (buffers for geometries and materials were built without considering my lights).
To achieve dynamic lightning an author of the article suggest to:
Add light with intensity = 0 and then increase it (such light is taken into account by render)
Force material to be updated (material.needsUpdate = true) for affected scene objects after setting light visibility to true.

Resources