sea3d update #8892

sunag · 2016-05-15T18:19:54Z

Based in new Three.JS Animation System
Compatible with Physical Material
Orthographic Camera
Add LightAnimator and CameraAnimator
Removed hacks
Port NodeMaterial - https://github.com/sunag/sea3d/blob/gh-pages/Labs/sea3d-1.8/SEA3DNodeMaterial.js
Port Ammo - https://github.com/sunag/sea3d/blob/gh-pages/Labs/sea3d-1.8/SEA3DAmmoLoader.js
Assets updated to 1.8 - http://community.poonya.com/201
New loader.clone() to clone complex assets
Fix sea3d animation duration

WestLangley · 2016-05-15T18:38:53Z

examples/webgl_materials_nodes.html

@@ -33,7 +33,7 @@
 			<a href="http://threejs.org" target="_blank">three.js</a> - Node-Based Material
 		</div>

-		<script src="../build/three.js"></script>
+		<script src="../build/three.min.js"></script>


We are using the un-minified build in examples now.

mrdoob · 2016-05-15T22:15:24Z

Based in new Three.JS Animation System

Sweet! Should the examples stop including js/loaders/collada/Animation.js and co.?

sunag · 2016-05-15T23:41:16Z

@mrdoob Yes! Vertex, keyframe, skeleton and others animators use only the new animation system now. Works great!

mrdoob · 2016-05-16T04:22:04Z

@bhouston and @tschw will love to hear that 😀

mrdoob · 2016-05-16T04:22:36Z

Are you planning on removing the redundant includes?

tschw · 2016-05-16T13:29:28Z

examples/js/loaders/sea3d/SEA3D.js

@@ -4749,7 +4159,7 @@ SEA3D.EventDispatcher.prototype = {

 };

-SEA3D.EventDispatcher.apply = function ( object ) {
+SEA3D.EventDispatcher.apply = function( object ) {

 	object.addEventListener = SEA3D.EventDispatcher.prototype.addEvenListener;


Not that this line has been changed, but it jumped right at me skimming the code: Should it read addEventListener - note the extra "t"?

Also worth mentioning: We have recently deprecated THREE.EventListener.prototype.apply and instead do

Object.assign( AClass.prototype, THREE.EventListener.prototype /* , { ... methods } */ );

now (there's a polyfill in Three.js, so yes, this works with all browsers).

Thanks @tschw . I updated EventDispatcher at https://github.com/mrdoob/eventdispatcher.js/ and some other things with Object.assign. Nice!

sunag · 2016-05-16T18:19:52Z

Are you planning on removing the redundant includes?

I even have a page with builds here: https://github.com/sunag/sea3d/tree/gh-pages/Build/Three.JS

Currently the hierarchy is:

SEA3D.js - SDK only no depedencies
SEA3DDeflate.js - Only If the file use Deflate compression
SEA3DLZMA.js - Only If the file use LZMA compression

< Three.JS >
SEA3DLoader.js - Three.JS SEA3D Loader
SEA3DLegacy.js - For non-TJS files
SEA3DNodeMaterial.js Compatible with 'NodeMaterial' for fresnel effect and others
SEA3DAmmoLoader.js Compatible with Ammo Physics Engine

SEA3DNodeMaterial and SEA3DAmmoLoader I think in a PR with examples.in the next updates...

sunag · 2016-05-16T18:24:38Z

How come you are not directly using THREE.EventDispatcher?

I will try to transfer the Script, DomainManager and others for SEA3DLoader for use THREE.EventDispatcher

sunag · 2016-05-16T19:39:09Z

All references to THREE.EventDispatcher updated!

sunag · 2016-05-16T21:25:49Z

@mrdoob Added support for filters in THREE.Audio and Legacy alert e.g:

// new
sound.setFilters( [ soundAnalyser, biquadFilter ] );

// old
sound.setFilter( biquadFilter );

and updated the example webgl_loader_sea3d.sound.html light intensity is based in sound amplitude.

mrdoob · 2016-05-16T21:34:36Z

All references to THREE.EventDispatcher updated!

Sweet!

mrdoob · 2016-05-16T21:34:40Z

@mrdoob Added support for filters in THREE.Audio and Legacy alert e.g:

I wouldn't deprecate setFilter() yet. Lets keep both setFilter() and setFilters() for now.

and updated the example webgl_loader_sea3d.sound.html light intensity is based in sound amplitude.

Nice! Maybe you can use AudioAnalyser though?

https://github.com/mrdoob/three.js/blob/master/examples/misc_sound.html#L143
https://github.com/mrdoob/three.js/blob/master/examples/misc_sound.html#L256

sunag · 2016-05-16T23:06:32Z

Maybe you can use AudioAnalyser though?

Nice! The Biquad Filter was obstructed a bit the sound amplitude by AudioAnalyser capture the output but it was good I think.

mrdoob · 2016-05-16T23:37:34Z

Lovely!

mrdoob · 2016-05-16T23:48:56Z

Thanks!

mrdoob · 2016-05-17T03:10:53Z

src/audio/AudioAnalyser.js

@@ -20,6 +20,22 @@ Object.assign( THREE.AudioAnalyser.prototype, {
 		this.analyser.getByteFrequencyData( this.data );
 		return this.data;

+	},
+
+	getAverage: function() {


Actually, wouldn't be more efficient to do this?

var analyser = new THREE.AudioAnalyser( sound, 2 ); var average = analyser.getData()[ 0 ];

Hmm, I think so. Sound great. I'll update the examples to an other PR. We remove getAverage function?

Maybe I did not update the cache. I reload and show:

Uncaught IndexSizeError: Failed to set the 'fftSize' property on 'AnalyserNode': The FFT size provided (2) is outside the range [32, 32768].

In console message. The the minimum is 32.

Ah, interesting. Never mind then!

mrdoob · 2016-05-17T03:13:12Z

Also, webgl_loader_sea3d_hierarchy doesn't seem to work now.

sunag · 2016-05-17T03:50:29Z

Also, webgl_loader_sea3d_hierarchy doesn't seem to work now.

It works here in my desktop. Maybe have some compatibility problem with the hardware because now has shadow and physical material!? It has some new colors too

mrdoob · 2016-05-17T03:59:47Z

I think it has something to do with the post processing. I can see the robot for one frame when resizing then it's only vignette.

sunag · 2016-05-17T04:21:00Z

Here appeared some warn too:

[.CommandBufferContext]RENDER WARNING: there is no texture bound to the unit 1

sunag · 2016-05-17T04:35:49Z

I add

loader.setShadowMap = function() {}

to ignore the shadow not showed more warnings.

mrdoob · 2016-05-17T04:58:25Z

Have you pulled from dev?

sunag · 2016-05-17T05:19:30Z

Have you pulled from dev?

Yes! The 77dev last revision than two hours ago.

mrdoob · 2016-05-17T13:51:34Z

Strange... If I comment out the vignette it works.

var vignettePass = new THREE.ShaderPass( THREE.VignetteShader );
vignettePass.uniforms[ "darkness" ].value = 1.0;
// composer.addPass( vignettePass );

mrdoob · 2016-05-17T13:52:07Z

Firefox doesn't have the problem though!

mrdoob · 2016-05-17T13:52:26Z

Also, in Safari...

tschw · 2016-05-17T19:21:08Z

Don't use TypedArray.prototype.slice - it's not portable!

Either use .subarray (it's a view and will change the original array when changed) or use new TypedArray( sourceArray.subarray( from, to ) ) if you need a slice that can be changed independently.

tschw · 2016-05-17T19:21:35Z

examples/webgl_loader_sea3d_hierarchy.html

@@ -120,6 +116,9 @@
 				renderer.setPixelRatio( window.devicePixelRatio );
 				renderer.setSize( window.innerWidth, window.innerHeight );
 				renderer.setClearColor( 0x333333, 1 );
+				renderer.shadowMap.enabled = true;
+				renderer.shadowMap.type = THREE.PCFSoftShadowMap;
+				renderer.shadowMap.cullFace = THREE.CullFaceBack;


Deprecated. See the warning.

@tschw About the Mixer you know if it is possible to change the timeScale from action?

I think I found it. I'll update it after I make a new PR.

Yes, certainly is.

You can also call setDuration passing how much global mixer time your animation is supposed to take as a more high-level way of setting .timeScale.

The value of .timeScale is effective when the action is not paused and not being time warped. The effective time scale during warping can be queried with getEffectiveTimeScale. The result is zero when the action is paused. The corresponding setter stops all warping, but does not release the .paused state.

@tschw Thanks! I'll test as soon as possible.

@tschw In unification of mixers I had a problem.
AnimationMixer works with unique names of AnimationClip but the variable name not represents an uuid semantically. If rename name to uuid or ns open space to declare the real animation name in name slot like idle, walk, run, etc.

https://github.com/mrdoob/three.js/blob/dev/src/animation/AnimationClip.js#L11

For example in skinning example I have:

https://github.com/mrdoob/three.js/blob/dev/examples/webgl_loader_sea3d_skinning.html#L171

if (player.currentAnimation.name == "idle")

currentAnimation is an AnimationClip. I could assign currentAnimation.ns or currentAnimationName but I think it would be clearer this way currentAnimation.name and currentAnimation.uuid for clip name currently.

In keyframe example I had the same problem by the names conflict.

https://github.com/mrdoob/three.js/blob/dev/examples/webgl_loader_sea3d_keyframe.html

@sunag
Yes, I remember having the same problem porting one of the examples and worked around it prefixing with the target uuid. I'm not much an expert when it comes to loaders and all the formats we support or halfway support, so I didn't really know how to best fix the underlying problem.

You OTOH seem to be having a plan, so I'll be grateful if you fixed it. I guess your changes will be sound and so there shouldn't be any problem getting them merged.

@tschw It will be a pleasure. I will present a possible solution in the next PR.

* node material r5 + mirror nodes example * threejs.min in node material examples * sea3d update * reverted to un-minified * update mrdoob EventDispatcher.js * fix eventdispatcher update * proto defines replace by Object.assign * native script support transferred to loader * sound filters support * light intensity from sound amplitude * keep .setFilter() method * get freq amplitude average * cleanup * use THREE.AudioAnalyser to get amplitude

sunag · 2016-05-19T03:29:41Z

@mrdoob The robot example has bug yet?

mrdoob · 2016-05-19T03:57:13Z

@sunag It's still broken here yes... It's the only sea3d example that is broken though. The turtle one also has a vignette so I'm not sure where the issue is.

mrdoob · 2016-05-20T17:34:13Z

Ok, seems to be an issue with Chrome only and, maybe, my setup. There seems to be an issue when using a renderTarget as source and target at the same time.

I'll check with Google next week.

sunag · 2016-05-20T18:11:30Z

Ok, seems to be an issue with Chrome only and, maybe, my setup. There seems to be an issue when using a renderTarget as source and target at the same time.

I have same problem in update of postprocessing_nodes example using renderTarget and NodeMaterial issue only in Chrome too:

I tested in r74 at r76 same problem. I modified this in r77 to fix but the issue remains open. This happened recently in my Chrome.

If replace the current fadeScreen:

https://github.com/mrdoob/three.js/blob/master/examples/webgl_postprocessing_nodes.html#L435

to:

var fadeScreen = new THREE.Math3Node(
    uv,
    coord,
    new THREE.Math1Node( new THREE.TextureNode( lensflare2 ), THREE.Math1Node.INVERT ),
    THREE.Math3Node.MIX
);

only the lensflare2 is rendered in Chrome, Firefox and others is done.

<- Left is Firefox
-> Right is Chrome

mrdoob · 2016-05-20T18:14:58Z

Yeah... I guess it's a Chrome regression.

mrdoob · 2016-05-20T18:15:18Z

Btw, there is still one slice left.

sunag · 2016-05-20T18:18:36Z

Btw, there is still one slice left.

Ops! I forgot to send in the last PR. I will send right now 👍

mrdoob · 2016-05-21T16:39:46Z

@sunag I checked with the Chrome guys. The behaviour is as intended. Chrome set that limitation because some mobiles can't render to a texture while they're using it. So I guess we'll have to do some pingponging... 😁

sunag · 2016-05-21T18:29:21Z

Thanks @mrdoob

Still about this! I forgot to point out in postprocessing_nodes issue no warning appeared.

bhouston · 2016-05-22T01:12:18Z

I think I've seen this when setting a uniform["name"].value = renderTarget instead of uniform["name"].value = renderTarget.texture. Or it may have been this erroneous form that caused that Chrome warning: uniform["name"] = renderTarget.texture. Basically once I fixed setting my uniforms correctly, this warning went away. So look for something like that.

sunag · 2016-05-22T05:43:18Z

Basically once I fixed setting my uniforms correctly, this warning went away. So look for something like that.

I did some tests about this but without success. I thought that the problems were into Three.JS core until test in others browsers.

I did a minimal reproduction of the problem in replace of mosaic example of postprocessing_nodes:

JavaScript

lensflare2.name = "lensflare2";

var uv = new THREE.UVNode();

var fadeScreen = new THREE.Math3Node(
    uv,
    uv,
    new THREE.Math1Node( new THREE.TextureNode( lensflare2 ), THREE.Math1Node.INVERT ),
    THREE.Math3Node.MIX
);

nodepass.value = new THREE.ScreenNode( fadeScreen );

NodeMaterial GLSL fragment:

#ifdef GL_EXT_shader_texture_lod
#define texCube(a, b) textureCube(a, b)
#define texCubeBias(a, b, c) textureCubeLodEXT(a, b, c)
#define tex2D(a, b) texture2D(a, b)
#define tex2DBias(a, b, c) texture2DLodEXT(a, b, c)
#else
#define texCube(a, b) textureCube(a, b)
#define texCubeBias(a, b, c) textureCube(a, b, c)
#define tex2D(a, b) texture2D(a, b)
#define tex2DBias(a, b, c) texture2D(a, b, c)
#endif
varying vec2 vUv;

uniform sampler2D renderTexture;
uniform sampler2D nVu1;



void main(){



gl_FragColor = tex2D(renderTexture,mix(vec4(vUv,0.0,1.0),vec4(vUv,0.0,1.0),(1.0-tex2D(nVu1,vUv))).xy);
}

Chrome uniforms

Textures names

composer.render(); // to first rtt

console.log( nodepass.node.uniforms.nVu1.value.name ); // output: "lensflare2"
console.log( nodepass.node.uniforms.renderTexture.value.name ); // output: ""

Result

The strange is that no warning was showing in my tests as shown the printscreen.

Another two important detail is that this bug only happens when use lensflare2 as UV and this worked in previous Chrome versions.

Firefox version: 46.0.1
Chrome version: 50.0.2661.102 - 32bit

<- Left Firefox
-> Right Chrome

…r to be the same. See #8892.

mrdoob · 2016-05-25T23:13:57Z

Ok, I think I figured it out. It was tricky...

What was "funny" was that if I duplicated the vignette pass the problem went away. So it only happened when the number of passes was odd.

…r to be the same. See mrdoob#8892.

sunag added 3 commits May 14, 2016 19:42

node material r5 + mirror nodes example

f13522f

threejs.min in node material examples

082488d

sea3d update

24a325a

WestLangley reviewed May 15, 2016
View reviewed changes

reverted to un-minified

8d68b6f

tschw reviewed May 16, 2016
View reviewed changes

sunag added 3 commits May 16, 2016 14:26

update mrdoob EventDispatcher.js

ae3729e

fix eventdispatcher update

ef749a0

proto defines replace by Object.assign

36eeef2

native script support transferred to loader

6cba6d2

sunag added 2 commits May 16, 2016 18:18

sound filters support

0b32a7a

light intensity from sound amplitude

74a0857

sunag added 4 commits May 16, 2016 19:36

keep .setFilter() method

6db30a5

get freq amplitude average

610cb42

cleanup

44b81e6

use THREE.AudioAnalyser to get amplitude

92373ba

mrdoob merged commit 0eaaa84 into mrdoob:dev May 16, 2016

mrdoob reviewed May 17, 2016
View reviewed changes

tschw reviewed May 17, 2016
View reviewed changes

mrdoob added a commit that referenced this pull request May 25, 2016

EffectComposer: Reseting pingpong can cause readBuffer and writeBuffe…

0f9b587

…r to be the same. See #8892.

carlosanunes pushed a commit to carlosanunes/three.js that referenced this pull request May 26, 2016

EffectComposer: Reseting pingpong can cause readBuffer and writeBuffe…

1181574

…r to be the same. See mrdoob#8892.

Uh oh!

sea3d update #8892

sea3d update #8892

Uh oh!

Conversation

sunag commented May 15, 2016

Uh oh!

Choose a reason for hiding this comment

Uh oh!

mrdoob commented May 15, 2016 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

sunag commented May 15, 2016

Uh oh!

mrdoob commented May 16, 2016

Uh oh!

mrdoob commented May 16, 2016

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

sunag commented May 16, 2016

Uh oh!

sunag commented May 16, 2016

Uh oh!

sunag commented May 16, 2016

Uh oh!

sunag commented May 16, 2016

Uh oh!

mrdoob commented May 16, 2016

Uh oh!

mrdoob commented May 16, 2016

Uh oh!

sunag commented May 16, 2016

Uh oh!

mrdoob commented May 16, 2016

Uh oh!

mrdoob commented May 16, 2016

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

mrdoob commented May 17, 2016

Uh oh!

sunag commented May 17, 2016

Uh oh!

mrdoob commented May 17, 2016

Uh oh!

sunag commented May 17, 2016

Uh oh!

sunag commented May 17, 2016

Uh oh!

mrdoob commented May 17, 2016

Uh oh!

sunag commented May 17, 2016

Uh oh!

mrdoob commented May 17, 2016

Uh oh!

mrdoob commented May 17, 2016

Uh oh!

mrdoob commented May 17, 2016

Uh oh!

tschw commented May 17, 2016

Uh oh!

Choose a reason for hiding this comment

Uh oh!

sunag May 17, 2016 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

mrdoob commented May 15, 2016 •

edited

Loading

sunag May 17, 2016 •

edited

Loading

sunag May 17, 2016 •

edited

Loading

sunag commented May 21, 2016 •

edited

Loading