more work

02fddda5 · Jerome Etienne · eccf0d1a · 02fddda5 · 02fddda5 · eccf0d1a
9 changed file
--- a/README.md
+++ b/README.md
@@ -106,8 +106,7 @@ by

 ```html
 <script src="https://aframe.io/releases/0.5.0/aframe.min.js"></script>
-<script src="https://rawgit.com/jeromeetienne/ar.js/master/aframe/build/aframe-ar.js"></script>
-<script>THREEx.ArToolkitContext.baseURL = 'https://rawgit.com/jeromeetienne/ar.js/master/three.js/'</script>
+<script src="https://jeromeetienne.github.io/AR.js/aframe/build/aframe-ar.js"></script>
 <body style='margin : 0px; overflow: hidden;'>
 	<a-scene embedded artoolkit='sourceType: webcam;'>
 		<a-box position='0 0 0.5' material='opacity: 0.5;'></a-box>

--- a/TODO.md
+++ b/TODO.md
 - do a webvr demo with a hole in the ground. - good for tango
  - possible demo: hole in the wall, refraction, liquid marker but on all the wall

-
 - release soon and start doing dev/master
  - create a dev branch
  - release AR.js as 1.2

--- a/three.js/examples/multi-markers/ALGO.md
+++ b/three.js/examples/multi-markers/ALGO.md
-# High Level Description
-
-The workflow is splitted into 3 steps :
-
-1. collect statistics about the relative position of each marker couple
-1. generation of multi-marker description
-1. using the multi-marker 
-
-So we first have a learning phase to collect data and generating result, then
-we have a usage phase, where the multi-marker is used in a AR application.
-
-So in short, you put all your markers where you like them. Then you scan them
-with your phone to collect statistics about their relative position.
-When you are done, you start using the AR application with this multi-marker.
-
-# Definitions
- a multi-marker is several markers acting as one
- a markerA is said to be a sub-marker of a multi-marker
-
-# Learning Algo description
- every time i see markerA and markerB at the same time, i store statistics on their relative position
- such couple is called a seen-couple
- the statistics collected are simply an average of position/quaternion/scale
- this statistics can be trivially expressed as a transformation matrix
-
-# Generation of Multi-Marker description
- when the learning is considered completed, we generate a multi marker description
-  - it is a json file in which each sub-markers has a transformation matrix relative to the origin sub-marker
- the first sub-marker is the anchor/root sub-marker.
- The origin position of the multi-marker is the origin sub-marker
- the origin sub-marker MUST be seen at least once during the learning phase
-
-## Description Format
- the description format is simple: 
- it is a list with one item for each sub-markers 
-  - it contains a description of the sub-markers (e.g. marker barcode 3)
-  - and a poseMatrix which is the transformation matrix relative to the origin sub-markers
-  - the first item is the origin sub-marker
-
-
---
-
-# Computation of the average matrix for each sub-marker
-
-Once statistics are collected, we need to produce a description of this new 
-area. This is a json structure describing each sub-marker and their respective
-position.
-
-Here are the algo which can be used
-
-## Minimal Algo - assume origin always visible
- Assume that the origin sub-marker is ALWAYS visible
- Look on all the seen-couple of the origin sub-marker
- for each sub-markers, you got the relative position with the origin sub-marker
- just generate the transformation matrix for it
-
-This is the one implemented
-
-## Minimal Algo - origin may sometime be invisible
- do first pass where you compute all couple which contain the origin submarkers
- so you computed the transformation matrix for each sub-marker which has been 
-  visible with the origin marker
- you computed the level 1 of transformation matrices
- the level 0 would the transformation matrix of the origin sub-marker. But it is 
-  always the identity matrix.
- you do an iterative process, where each iterations compute one more level.
-
-Algo for one iteration
-1. for each sub-markers, check all the seen couple it appears in
-2. if at least one already has a transformation matrix, compute the transformation
-3. Loop until all sub-markers got computed
--- a/three.js/examples/multi-markers/README.md
+++ b/three.js/examples/multi-markers/README.md
@@ -35,3 +35,78 @@ Thus even if only one sub-marker is visible, multi-marker position will still be
 I tried to workaround those issues, but it rapidely became a pain. 
 So i prefered to devote my energy at building a good version instead of
 working around those bugs.
+
+
+---
+
+# High Level Description
+
+The workflow is splitted into 3 steps :
+
+1. collect statistics about the relative position of each marker couple
+1. generation of multi-marker description
+1. using the multi-marker 
+
+So we first have a learning phase to collect data and generating result, then
+we have a usage phase, where the multi-marker is used in a AR application.
+
+So in short, you put all your markers where you like them. Then you scan them
+with your phone to collect statistics about their relative position.
+When you are done, you start using the AR application with this multi-marker.
+
+# Definitions
+- a multi-marker is several markers acting as one
+- a markerA is said to be a sub-marker of a multi-marker
+
+# Learning Algo description
+- every time i see markerA and markerB at the same time, i store statistics on their relative position
+- such couple is called a seen-couple
+- the statistics collected are simply an average of position/quaternion/scale
+- this statistics can be trivially expressed as a transformation matrix
+
+# Generation of Multi-Marker description
+- when the learning is considered completed, we generate a multi marker description
+  - it is a json file in which each sub-markers has a transformation matrix relative to the origin sub-marker
+- the first sub-marker is the anchor/root sub-marker.
+- The origin position of the multi-marker is the origin sub-marker
+- the origin sub-marker MUST be seen at least once during the learning phase
+
+## Description Format
+- the description format is simple: 
+- it is a list with one item for each sub-markers 
+  - it contains a description of the sub-markers (e.g. marker barcode 3)
+  - and a poseMatrix which is the transformation matrix relative to the origin sub-markers
+  - the first item is the origin sub-marker
+
+
+---
+
+# Computation of the average matrix for each sub-marker
+
+Once statistics are collected, we need to produce a description of this new 
+area. This is a json structure describing each sub-marker and their respective
+position.
+
+Here are the algo which can be used
+
+## Minimal Algo - assume origin always visible
+- Assume that the origin sub-marker is ALWAYS visible
+- Look on all the seen-couple of the origin sub-marker
+- for each sub-markers, you got the relative position with the origin sub-marker
+- just generate the transformation matrix for it
+
+This is the one implemented
+
+## Minimal Algo - origin may sometime be invisible
+- do first pass where you compute all couple which contain the origin submarkers
+- so you computed the transformation matrix for each sub-marker which has been 
+  visible with the origin marker
+- you computed the level 1 of transformation matrices
+- the level 0 would the transformation matrix of the origin sub-marker. But it is 
+  always the identity matrix.
+- you do an iterative process, where each iterations compute one more level.
+
+Algo for one iteration
+1. for each sub-markers, check all the seen couple it appears in
+2. if at least one already has a transformation matrix, compute the transformation
+3. Loop until all sub-markers got computed
--- a/three.js/examples/multi-markers/TODO.md
+++ b/three.js/examples/multi-markers/TODO.md
+- what to do with the official multimarker support
+  - make an example for it, and keep supporting it
+
+- make all markers children of a parent THREEx.ArBaseControls()
+  - GOAL: make it explicit what is expected from a AR controls
+  - emit event
+  - have id
+  - anything which is common
+
 - TODO put a smoother on sub-marker while learning ? as a way to remove noise ?
+  - do it and hide it behind a flags
+  - in relation with THREEx.ArBaseControls

 - do an apps which does something with it
-  - a minecraft going from markers to markers
+  - a minecraft going from submarkers to submarkers

 - add a-frame support
-  - support for learning new area
+  - support for learning new area - can i just use the area-learner.html
  - support of multi-marker description

- do a post about multimarker
-  - how to use them, 
-  - workflow: learning then using
-  - pro: larger area can be coverded by 
-
- what to do with the official multimarker support
-  - make an example for it, and keep supporting it
-
-
-
-# How to compute the result of the learning
- we got statistics on the position/orientation/scale between each marker seen simultaneously
- we want a transposition matrix relative to the origin markersControls
+- do a post about multimarker - https://medium.com/p/4bcafc785dfd/edit

- step1 - build the transposition matrix relative the origin marker controls
-  - store it in the userData
-  - use similar format as local matrix
- step2 - generate the file
- 


- LATER: how to get more robust statistics?
+- DONE merge it all in the README.md
--- a/three.js/examples/multi-markers/examples/learner.html
+++ b/three.js/examples/multi-markers/examples/learner.html
@@ -164,6 +164,8 @@
 			markerControls.object3d.add( markerHelper.object3d )			
 		}
 		
+		// TODO here put a THREEx.ArSmoothedControls behind a flag - could be useful for tunning
+		
 		// store it in the parameters
 		subMarkersControls.push(markerControls)
 	})

--- a/three.js/examples/multi-markers/examples/player.html
+++ b/three.js/examples/multi-markers/examples/player.html
@@ -302,8 +302,6 @@

 		multiMarkerPlay(JSON.stringify(multiMarkerFile))
 	}
-	
-	

 	//////////////////////////////////////////////////////////////////////////////////
 	//		render the whole thing on the page

--- a/three.js/threex-arbasecontrols.js
+++ b/three.js/threex-arbasecontrols.js
+var THREEx = THREEx || {}
+
+THREEx.ArBaseControls = function(object3d){
+	this.id = THREEx.ArBaseControls.id++
+	this.object3d = object3d
+
+	// Events to honor
+	// this.dispatchEvent({ type: 'becameVisible' })
+	// this.dispatchEvent({ type: 'becameUnVisible' })
+}
+
+THREEx.ArBaseControls.id = 0
+
+Object.assign( THREEx.ArBaseControls.prototype, THREE.EventDispatcher.prototype );
+
+//////////////////////////////////////////////////////////////////////////////
+//		Functions
+//////////////////////////////////////////////////////////////////////////////
+/**
+ * error catching function for update()
+ */
+THREEx.ArBaseControls.prototype.update = function(){
+	console.assert(false, 'you need to implement your own update')
+}
--- a/three.js/threex-arsmoothedcontrols.js
+++ b/three.js/threex-arsmoothedcontrols.js
@@ -103,7 +103,7 @@ THREEx.ArSmoothedControls.prototype.update = function(targetObject3d){
 		object3d.scale.lerp(targetObject3d.scale, parameters.lerpScale)
 	}

-	// disabled the lerp by directly copying targetObject3d position/quaternion/scale
+	// disable the lerp by directly copying targetObject3d position/quaternion/scale
 	// if( false ){		
 	// 	this.object3d.position.copy( targetObject3d.position )
 	// 	this.object3d.quaternion.copy( targetObject3d.quaternion )