Speech bubbles in Canvas/FabricJS? - html

Looking for a way to make a regular speech bubble in my website's FabricJS canvas. Now before you flag this post, I did see this question, it just has no proper answers and is designed for WordPress so it's not particularly of any use to me.
What I'm wanting is pretty clear: A speech bubble with text in it and a tail/handle that you can drag to point it to something.
I've found this library but I can't seem to get it to show up in my FabricJS canvas? If you could either explain to me how to add this library into my canvas or provide another way of making speech bubbles, that would be sublime.

I dug a bit into Fabric.js and managed to create a procedual speech bubble, but I'm not able to quickly convert it into a Fabric.js class (which would make sense if you want to have multiple speech bubbles on your canvas). Maybe it's still helpful for you or someone else https://codepen.io/timohausmann/pen/poywXzg
It basically creates a Textbox and based on the bounding box of the text updates the position of the Rect around it.
var bound = textbox.getBoundingRect();
rect.left = bound.left - boxPadding;
rect.top = bound.top - boxPadding;
rect.width = bound.width + (boxPadding*2);
rect.height = bound.height + (boxPadding*2);
For the tail I simply created a transparent Rect that you can drag around and use its coordinates to draw a polygon with three points between the "handle" and the textbox center [A]. To make sure the tail maintains a certain width no matter the position, I calculate the degree between handle and the speech bubble center [B]. To keep the position of textbox and handle in sync, I calculate how much textbox moved and simply add the difference to the handles position [C].
//calculate degree between textbox and handle [B]
var angleRadians = Math.atan2(handle.top - textbox.top,
handle.left - textbox.left);
var offsetX = Math.cos(angleRadians + (Math.PI/2));
var offsetY = Math.sin(angleRadians + (Math.PI/2));
//update the polygon [A]
poly.points[0].x = handle.left;
poly.points[0].y = handle.top;
poly.points[1].x = textbox.left - (offsetX * arrowWidth);
poly.points[1].y = textbox.top - (offsetY * arrowWidth);
poly.points[2].x = textbox.left + (offsetX * arrowWidth);
poly.points[2].y = textbox.top + (offsetY * arrowWidth);
//update the handle when the textbox moved [C]
if(textbox.left !== textbox.lastLeft ||
textbox.top !== textbox.lastTop) {
handle.left += (textbox.left - textbox.lastLeft);
handle.top += (textbox.top - textbox.lastTop);
handle.setCoords();
}
Disclaimer: I'm not a Fabric.js expert, maybe there are a few shortcuts possible with the library.

The answer by #Til Hausmann works nicely (thanks!).
I run into some problems when I tried to store and load the canvas data (via canvas.toJSON and canvas.loadFromJSON, resp.), though.
After some fiddling around, this could be resolved by
storing lastLeft and lastTop for both polygons in the updateBubble() method:
poly.lastLeft = Math.min(handle.left, textBox.left);
poly.lastTop = Math.min(handle.top, textBox.top);
setting the left / top properties for the polygons after the data were loaded:
canvas.loadFromJSON(jsonData, () => {
const poly = // ...
const poly2 = // ...
poly.left = poly.lastLeft;
poly.top = poly.lastTop;
poly2.left = poly2.lastLeft;
poly2.top = poly2.lastTop;
// ...
// Important:
canvas.renderAll();
});
passing the full set of shape properites to canvas.toJSON()
canvas.toJSON(
['lastLeft', 'lastTop'].concat(
Object.keys(handleProperties),
Object.keys(polyProperties),
Object.keys(poly2Properties),
Object.keys(textRectProperties)
))
I was surprised that step (3) is actually necessary, but it didn't work without it...

Related

Actionscript 3 - alternatives to .hitTestObject or position constraints

I need to detect when MC2 is over MC1 that it is inside MC1's borders.
to do this I would usually use 4 separate if x y constraints,
and unfortunately .hitTestObject in my creations also seem to need 4 separate if x y + - constraints.
Does anyone know a more simplistic way to achieve this.
or is x y + - constraints still the only way to do this?
Thank you in advance.
The final solution for your problem to detect hit of two shapes, is to use bitmapData.hitTest(). you can detect hit between any shapes and not only Rectangles. for that, you have to draw both of your shapes on bitmapData like line belo:
var shape1Bitmap:BitmapData = new BitmapData(shape1MC.with,shape1MC.height,true,0x000000);
shape1Bitmap.draw(shape1MC);
var shape2Bitmap:BitmapData = new BitmapData(shape1MC.with,shape1MC.height,true,0x000000);
shape1Bitmap.draw(shape1MC);
shape1Bitmap.hitTest(new Point(),shape2Bitmap):Boolean;******
to continue usint BitmapData.hitTest(), folow the orders here : https://help.adobe.com/en_US/FlashPlatform/reference/actionscript/3/flash/display/BitmapData.html#hitTest()
http://dougmccune.com/blog/2007/02/03/using-hittestpoint-or-hittest-on-transparent-png-images/
It is a little complicated to add the bitmapData.hitTest() samples here. if any further questions left, please let me know to explain.
Good luck
I don't know of a built in way to do this, but it's easy enough using hitTestPoint with each corner of the square:
function isSquareInsideObject(square:DisplayObject, obj:DisplayObject):Boolean {
if(!obj.hitTestPoint(square.x, square.y, true)) return false;
if(!obj.hitTestPoint(square.x + square.width, square.y, true)) return false;
if(!obj.hitTestPoint(square.x + square.width, square.y + square.height, true)) return false;
if(!obj.hitTestPoint(square.x, square.y + square.height, true)) return false;
return true;
}
For more complex shapes than a square, you'd have to add more points to be accurate and it becomes a less elegant and less performant solution then.
You need that shape argument (third parameter for hitTestPoint) set to true if you want to test against the actual circle shape instead of the rectangular bounding box of the circle. If your circle is a bitmap (and not a shape), then I'd suggest putting a circular mask on the object to achieve the same result.
If your square isn't anchored at 0,0, or you don't mind the extra (small) performance hit, you could also use var bounds:Rectangle = square.getBounds(this) and then use the convenience properties of the rectangle object (bounds.bottomLeft, bottomRight, topLeft, topRight)

Scaling points to match background. Actionscript 3

Using some code I found online has helped me create a zoom function for a program I am attempting to make. It is to make a map that allows a user to mark points. Currently the code scales in on the map image alone but I cant get the point icons to realign to where they originally where. I cant workout the maths of it.
Code to zoom in and out
if (mev.shiftKey) {
image.scaleX = Math.max(scaleFactor*image.scaleX, minScale);
image.scaleY = Math.max(scaleFactor*image.scaleY, minScale);
}
if (mev.ctrlKey) {
image.scaleX = Math.min(1/scaleFactor*image.scaleX, maxScale);
image.scaleY = Math.min(1/scaleFactor*image.scaleY, maxScale);
mat = image.transform.matrix.clone();
MatrixTransformer.matchInternalPointWithExternal(mat,internalCenter,externalCenter);
image.transform.matrix=mat;
This allows the image to scale up with the following factors
public var scaleFactor:Number = 0.8;
public var minScale:Number = 0.25;
public var maxScale:Number = 2.0;
The problem occurs when I try to move the pointer icons that are overlaid on this image. They are not to grow or shrink at the moment but they I cant get the maths to get them to move the correct number of pixels away from the mouse location so that they are still in line. Currently I am using the following formulas
//decrease zoom
stage.getChildAt(i).x = stage.getChildAt(i).x * scaleFactor;
//increase zoom
stage.getChildAt(i2).x = stage.getChildAt(i2).x / scaleFactor;
Any thoughts ? Code I am using came from
http://www.flashandmath.com/howtos/zoom/
Quite a few elements missing from the question like the moving map underneath. Anyway now that it's sorted out ...
If you are not a math genius and can't tackle 2 math formulas at the same time then don't and tackle them one by one then combine them. Once again don't use the x,y property of point for calculation but create specific property (like in a custom class for example). I will name them here origin for convenience.
Given a point with origin property of x:100, y:200, its position on the map is (assuming map is top left coordinate, if not adapt accordingly):
point.x = map.x + point.origin.x;
point.y = map.y + point.origin.y;
the positioning is solved now you need to solve for scale which is easy:
point.x = point.origin.x * scaleFactor;
point.y = point.origin.y * scaleFactor;
Both systems are solved now you can combine the two:
point.x = map.x + (point.origin.x * scaleFactor);
point.y = map.y + (point.origin.y * scaleFactor);

How to rotate all objects of canvas at once using Fabric.js?

I am working on custom product designer which uses Fabric.js. I want to rotate all objects of canvas at once by pressing one button (rotate left, rotate right).
I have achieved this using this code :
stage.forEachObject(function(obj){
obj.setAngle(rotation).setCoords();
});
stage.renderAll();
But it has one bug that every element rotates with its own center point. I want that every element rotates with respect to whole canvas element.
Grouping and rotating the group did not work so well for me. Here is another solution based on this js fiddle.
rotateAllObjects (degrees) {
let canvasCenter = new fabric.Point(canvas.getWidth() / 2, canvas.getHeight() / 2) // center of canvas
let radians = fabric.util.degreesToRadians(degrees)
canvas.getObjects().forEach((obj) => {
let objectOrigin = new fabric.Point(obj.left, obj.top)
let new_loc = fabric.util.rotatePoint(objectOrigin, canvasCenter, radians)
obj.top = new_loc.y
obj.left = new_loc.x
obj.angle += degrees //rotate each object buy the same angle
obj.setCoords()
});
canvas.renderAll()
},
You could add all the objects to a group an then rotate the group. This way you can also set the center for rotation.
This is how it could be solved
function rotate(a) {
var group = new fabric.Group(canvas.getObjects());
//angle is var with scope out of this function,
//so you can use this function as rotate(90) and keep rotating
angle = (angle + a) % 360;
group.rotate(angle);
canvas.centerObject(group);
group.setCoords();
canvas.renderAll();
}
FabricJS rotate everything and maintain the relative position also.
You can download the files here - https://drive.google.com/file/d/1UV1nBdfBk6bg9SztyVoWyLJ4eEZJgZRf/view?usp=sharing

Zoom to and from point

I'm trying to zoom a DisplayObject into a certain point. I figured it would be easy, but I've spent a day now trying to figure it out.
Basically, I think this should work. Emphasis on should.
//newPoint is the point being centered. There is no initial scaling, so I do not need to compensate for that (yet)
//scale is the zoom level
//container is the parent of the obj
//obj is the object being scaled/panned
var p:Point = new Point(
( this.container.width - this.obj.width * scale + newPoint.x * scale ) / 2,
( this.container.height - this.obj.height * scale + newPoint.y * scale ) / 2
);
this.obj.scaleX = this.obj.scaleY = scale;
this.obj.x = p.x;
this.obj.y = p.y;
It centers the point if scale is 1, but it gets further and further away from center as you increase the scale. I've tried dozens of different methods. This method, which I have seen on several sites, produced the same exact results. Anyone have any idea how to get this to work?
EDIT 10-1-12:
As a followup, I took the code snippet that LondonDrugs_MediaServices provided as a basis for my original issue. I needed to be able to zoom to a specific point at a specific scale relative to the unscaled image (think how Google Maps zooms to a specific location). To do this, I had to center my image on the point before running the translation code. I've posted the additional code below. For other uses (pinch to zoom, scrolling, and double click), I used the code provided by Vesper, which worked quite well.
//obj is the object being translated
//container is its parent
//x and y are the coordinates to be zoomed to, in untranslated scaling
//obj.scaleX and obj.scaleY are always identical in my class, so there is no need to account for that
//calculates current center point, with scaling
var center:Point = new Point( ( this.container.width - this.obj.width * this.obj.scaleX ) / 2, ( this.container.height - this.obj.height * this.obj.scaleX ) / 2 );
//calulcates the distance from center the point is, with scaling
var distanceFromCenter:Point = new Point( this.obj.width * this.obj.scaleX / 2 - x * this.obj.scaleX, this.obj.height * this.obj.scaleX / 2 - y * this.obj.scaleX );
//center the object on that specific point
this.obj.x = center.x + distanceFromCenter.x;
this.obj.y = center.y + distanceFromCenter.y;
var mat:Matrix=new Matrix();
mat.translate(-p.x,-p.y);
mat.scale(desiredScale,desiredScale);
mat.translate(p.x,p.y);
yourObject.transform.matrix=mat;
The core point is that scaling is done around (0,0), but you can do it with matrix that describes affine transformations. You first make an empty matrix (that is, a matrix that doesn't transform), then apply a set of transformations to it. First, place a desired point at (0,0) by translating by -1*coordinates, then scale, then translate back.
hie guys....
thank's your comments...
i found the answer...
code :
gambar.addEventListener(TransformGestureEvent.GESTURE_ZOOM , onZoom);
function onZoom(event:TransformGestureEvent):void {
var locX:Number=event.localX;
var locY:Number=event.localY;
var stX:Number=event.stageX;
var stY:Number=event.stageY;
var prevScaleX:Number=gambar.scaleX;
var prevScaleY:Number=gambar.scaleY;
var mat:Matrix;
var externalPoint=new Point(stX,stY);
var internalPoint=new Point(locX,locY);
gambar.scaleX *= event.scaleX;
gambar.scaleY *= event.scaleY;
if(event.scaleX>1 && gambar.scaleX>6){
gambar.scaleX=prevScaleX;
gambar.scaleY=prevScaleY;
}
if(event.scaleY>1 && gambar.scaleY>6){
gambar.scaleX=prevScaleX;
gambar.scaleY=prevScaleY;
}
if(event.scaleX<1 && gambar.scaleX<0.8){
gambar.scaleX=prevScaleX;
gambar.scaleY=prevScaleY;
}
if(event.scaleY<1 && gambar.scaleY<0.8){
gambar.scaleX=prevScaleX;
gambar.scaleY=prevScaleY;
}
mat=gambar.transform.matrix.clone();
MatrixTransformer.matchInternalPointWithExternal(mat,internalPoint,externalPoint);
gambar.transform.matrix=mat;
}
The matrix answer is absolutely correct, but if you happen to be a Club GreenSock member you can get some nice functionality with very simple code with the TransformAroundPointPlugin
http://www.greensock.com/as/docs/tween/com/greensock/plugins/TransformAroundPointPlugin.html
You can see an example in the plugin explorer here:
http://www.greensock.com/tweenlite/#plugins
I use this to tween all my zooms and have much better performance than when I tried to do this manually. IMO the whole library is worth it's weight in gold (and no I have no connection other than liking the library). If you need any of the other features I'd look into it. It also has the ThrowProps plugin ( http://www.greensock.com/throwprops/ )which is very important if you are going to have a bounding box on mobile that you want to have a smooth return into the boundaries.
Set obj.x to -p.x and obj.y to -p.y, set the container scaleX and scaleY to the desired value and add p.x to the container x and p.y to the container y. Done!

three.js - How to rotate a object with mouse (but not using camera)

I want to create a lot of objects and rotate each of them separately using the mouse. So far, I can select one of the objects with the mouse, but I can't use the mouse for an object rotation.
As the mouse only has mouse.x = event.clientX - windowHalfX;
mouse.y = event.clientY - windowHalfY;, I only know how to use the mousemove and mousedown event handlers to change the SELECTED.rotation.y and SELECTED.rotation.x (where SELECTED is the selected object) – how can I control the SELECTED.rotation.z too?
If the selected object is upside down, the x-rotation will also backwards which seems not very preferable. Is there any way to modify this?
Lots of the examples I found use camera rotation rather actually rotating the object. I'd like to find an solution that can rotate the object without a camera change.
You should check out the given control examples from three.js. They are not comfortable and mostly you need to copy them and modify like you want. But they take an object which is the controller this could be the camera or object. Here is a small example:
yawLeft = -((event['pageX'] - fullWidth) - halfWidth) / halfWidth;
pitchDown = ((event['pageY'] - fullHeight) - halfHeight) / halfHeight;
rotationVector.x = (-pitchUp + pitchDown);
rotationVector.y = (-yawRight + yawLeft);
rotationVector.z = (-rollRight + rollLeft);
var tmpQuaternion = new THREE.Quaternion();
tmpQuaternion.set(rotationVector.x * rotMult, rotationVector.y * rotMult, rotationVector.z * rotMult, 1).normalize();
object.quaternion.multiplySelf(tmpQuaternion);
object could be camera or model. It does not matter because camera is an THREE.Object3D also.