Develop Reference

HTML Canvas: ctx.stroke restroke behaviour with transparent colors

I am working on a sketch tool with html canvas.
I am using a common algorithm for this, that uses the mousedown, mousemove, mouseup events.
mousedown
I beginPath(), and moveTo(// the mouse coordinates).
mousemove
I draw lineTo(// the mouse coordinates), and then stoke(// the line to render it)
mouseup
I do nothing, // closePath()
And I noticed that, calling the stroke method without first calling closePath or beginPath, will redraw or restroke all previous paths or lines, which makes them appear thicker than the define color.
without a transparent color its is barely noticeable, but the colors do appear thicker than they should be.
but with color with transparency|alpha e.g. rgba(). The most recent path or line respects the color's transparency, but all previous line due to the redraw, the transparent colored line overlap and that causes previous lines to get thicker in sequence or succession.
is there a way to avoid|prevent this behavior. thank in advance.
sample is below, try drawing really fast!
var cvs = document.querySelector("canvas");
cvs.width = cvs.parentElement.clientWidth;
var colorInput = document.querySelector("input");
var ctx = cvs.getContext("2d");
ctx.strokeStyle = "rgba(0, 0, 0, 0.4)"
ctx.lineWidth = 20;
onDraw(cvs, {
penDown: function(e) {
var x = e.pageX - this.offsetLeft;
var y = e.pageY - this.offsetTop;
ctx.strokeStyle = colorInput.value;
ctx.beginPath();
ctx.moveTo(x, y);
},
penMove: function(e) {
var x = e.pageX - this.offsetLeft;
var y = e.pageY - this.offsetTop;
ctx.lineTo(x, y);
ctx.stroke();
},
penUp: function() {
// ctx.closePath;
}
});
function onDraw(node, drawHandler, beginHandler, endHandler, outOfBoundHandler, sticky) {
var mouseDown = false, mouseOut = false;
if( typeof drawHandler === "object" ) {
var drawEvents = drawHandler;
drawHandler = get(drawEvents.penMove);
beginHandler = get(drawEvents.penDown);
endHandler = get(drawEvents.penUp);
outOfBoundHandler = get(drawEvents.penOff);
sticky = drawEvents.sticky;
}
function get(name) {
return typeof name === "string" ? drawEvents[ name ] : name;
}
node.addEventListener('mousedown', function(e) {
mouseDown = true;
beginHandler&&beginHandler.call(this, e);
});
node.addEventListener('mousemove', function(e) {
mouseDown&&drawHandler&&drawHandler.call(this, e);
});
node.addEventListener('mouseup', function(e) {
mouseDown = false;
endHandler&&endHandler.call(this, e);
});
node.addEventListener('mouseout', function(e) {
mouseDown&&outOfBoundHandler&&outOfBoundHandler.call(this, e);
if( !sticky ) {
mouseDown = false;
}
});
}
.wrapper { border: 1px solid #aaa }
<div class="wrapper">
<canvas border="1" width="600" hieght="400">Canvas is not supported</canvas>
<input type="text" value="rgba(0, 0, 0, 0.3)" placeholder="rgba(#, #, #, #)">
</div>

If no Path argument is passed to stroke and fill methods they will use the path currently being declared with the context's drawing methods.
const ctx = c.getContext('2d');
// starts Path declaration
ctx.moveTo(20, 20);
ctx.lineTo(30, 80);
ctx.stroke(); // first rendering
setTimeout(() => {
ctx.clearRect(0, 0, 300, 150); // even if we clear the canvas
ctx.lineTo(70, 20); // this will continue path declaration
setTimeout(() => {
ctx.stroke(); // and this will draw everything
}, 1000);
}, 1000);
<canvas id="c"></canvas>
The only ways to start a new path declaration (except for the first one) are to either reset the whole context (not good), or to use beginPath method.
const ctx = c.getContext('2d');
// starts Path declaration
ctx.moveTo(20, 20);
ctx.lineTo(30, 80);
ctx.stroke(); // first rendering
setTimeout(() => {
ctx.clearRect(0, 0, 300, 150);
ctx.beginPath(); // start a new Path declaration
ctx.moveTo(30, 80); // we need to move to the previous coords
ctx.lineTo(70, 20); // this will be alone
ctx.stroke(); // and this will draw only the new path
}, 1000);
<canvas id="c"></canvas>
About closePath, it's just a lineTo(last_point_in_current_path_declaration), and doesn't ends a path declaration in no way.
So for your problem, there are two strategies you can adopt :
keep only the last coordinates, and at every mousemove,
ctx.beginPath();
ctx.moveTo(lastX, lastY);
ctx.lineTo(nextX, nextY);
keep all your coordinates in an array and redraw everything every time
ctx.clearRect(0, 0, canvas.width, canvas.height);
ctx.beginPath();
coords.forEach(pt => ctx.lineTo(pt.x, pt.y));
ctx.stroke();
Personally, I prefer the second one, which allows some undo - redo, and to e.g change your pen's style.

Image in canvas leaves a tiled trail when panned

I am trying to create a pannable image viewer which also allows magnification. If the zoom factor or the image size is such that the image no longer paints over the entire canvas then I wish to have the area of the canvas which does not contain the image painted with a specified background color.
My current implementation allows for zooming and panning but with the unwanted effect that the image leaves a tiled trail after it during a pan operation (much like the cards in windows Solitaire when you win a game). How do I clean up my canvas such that the image does not leave a trail and my background rectangle properly renders in my canvas?
To recreate the unwanted effect set magnification to some level at which you see the dark gray background show and then pan the image with the mouse (mouse down and drag).
Code snippet added below and Plnkr link for those who wish to muck about there.
http://plnkr.co/edit/Cl4T4d13AgPpaDFzhsq1
<!DOCTYPE html>
<html>
<head>
<style>
canvas{
border:solid 5px #333;
}
</style>
</head>
<body>
<button onclick="changeScale(0.10)">+</button>
<button onclick="changeScale(-0.10)">-</button>
<div id="container">
<canvas width="700" height="500" id ="canvas1"></canvas>
</div>
<script>
var canvas = document.getElementById('canvas1');
var context = canvas.getContext("2d");
var imageDimensions ={width:0,height:0};
var photo = new Image();
var isDown = false;
var startCoords = [];
var last = [0, 0];
var windowWidth = canvas.width;
var windowHeight = canvas.height;
var scale=1;
photo.addEventListener('load', eventPhotoLoaded , false);
photo.src = "http://www.html5rocks.com/static/images/cors_server_flowchart.png";
function eventPhotoLoaded(e) {
imageDimensions.width = photo.width;
imageDimensions.height = photo.height;
drawScreen();
}
function changeScale(delta){
scale += delta;
drawScreen();
}
function drawScreen(){
context.fillRect(0,0, windowWidth, windowHeight);
context.fillStyle="#333333";
context.drawImage(photo,0,0,imageDimensions.width*scale,imageDimensions.height*scale);
}
canvas.onmousedown = function(e) {
isDown = true;
startCoords = [
e.offsetX - last[0],
e.offsetY - last[1]
];
};
canvas.onmouseup = function(e) {
isDown = false;
last = [
e.offsetX - startCoords[0], // set last coordinates
e.offsetY - startCoords[1]
];
};
canvas.onmousemove = function(e)
{
if(!isDown) return;
var x = e.offsetX;
var y = e.offsetY;
context.setTransform(1, 0, 0, 1,
x - startCoords[0], y - startCoords[1]);
drawScreen();
}
</script>
</body>
</html>

You need to reset the transform.
Add context.setTransform(1,0,0,1,0,0); just before you clear the canvas and that will fix your problem. It sets the current transform to the default value. Then befor the image is draw set the transform for the image.
UPDATE:
When interacting with user input such as mouse or touch events it should be handled independently of rendering. The rendering will fire only once per frame and make visual changes for any mouse changes that happened during the previous refresh interval. No rendering is done if not needed.
Dont use save and restore if you don't need to.
var canvas = document.getElementById('canvas1');
var ctx = canvas.getContext("2d");
var photo = new Image();
var mouse = {}
mouse.lastY = mouse.lastX = mouse.y = mouse.x = 0;
mouse.down = false;
var changed = true;
var scale = 1;
var imageX = 0;
var imageY = 0;
photo.src = "http://www.html5rocks.com/static/images/cors_server_flowchart.png";
function changeScale(delta){
scale += delta;
changed = true;
}
// Turns mouse button of when moving out to prevent mouse button locking if you have other mouse event handlers.
function mouseEvents(event){ // do it all in one function
if(event.type === "mouseup" || event.type === "mouseout"){
mouse.down = false;
changed = true;
}else
if(event.type === "mousedown"){
mouse.down = true;
}
mouse.x = event.offsetX;
mouse.y = event.offsetY;
if(mouse.down) {
changed = true;
}
}
canvas.addEventListener("mousemove",mouseEvents);
canvas.addEventListener("mouseup",mouseEvents);
canvas.addEventListener("mouseout",mouseEvents);
canvas.addEventListener("mousedown",mouseEvents);
function update(){
requestAnimationFrame(update);
if(photo.complete && changed){
ctx.setTransform(1,0,0,1,0,0);
ctx.fillStyle="#333";
ctx.fillRect(0,0, canvas.width, canvas.height);
if(mouse.down){
imageX += mouse.x - mouse.lastX;
imageY += mouse.y - mouse.lastY;
}
ctx.setTransform(scale, 0, 0, scale, imageX,imageY);
ctx.drawImage(photo,0,0);
changed = false;
}
mouse.lastX = mouse.x
mouse.lastY = mouse.y
}
requestAnimationFrame(update);
canvas{
border:solid 5px #333;
}
<button onclick="changeScale(0.10)">+</button><button onclick="changeScale(-0.10)">-</button>
<canvas width="700" height="500" id ="canvas1"></canvas>

Nice Code ;)
You are seeing the 'tiled' effect in your demonstration because you are painting the scaled image to the canvas on top of itself each time the drawScreen() function is called while dragging. You can rectify this in two simple steps.
First, you need to clear the canvas between calls to drawScreen() and second, you need to use the canvas context.save() and context.restore() methods to cleanly reset the canvas transform matrix between calls to drawScreen().
Given your code as is stands:
Create a function to clear the canvas. e.g.
function clearCanvas() {
context.clearRect(0, 0, canvas.width, canvas.height);
}
In the canavs.onmousemove() function, call clearCanvas() and invoke context.save() before redefining the transform matrix...
canvas.onmousemove = function(e) {
if(!isDown) return;
var x = e.offsetX;
var y = e.offsetY;
/* !!! */
clearCanvas();
context.save();
context.setTransform(
1, 0, 0, 1,
x - startCoords[0], y - startCoords[1]
);
drawScreen();
}
... then conditionally invoke context.restore() at the end of drawScreen() ...
function drawScreen() {
context.fillRect(0,0, windowWidth, windowHeight);
context.fillStyle="#333333";
context.drawImage(photo,0,0,imageDimensions.width*scale,imageDimensions.height*scale);
/* !!! */
if (isDown) context.restore();
}
Additionally, you may want to call clearCanvas() before rescaling the image, and the canvas background could be styled with CSS rather than .fillRect() (in drawScreen()) - which could give a performance gain on low spec devices.
Edited in light of comments from Blindman67 below
See Also
Canvas.context.save : https://developer.mozilla.org/en-US/docs/Web/API/CanvasRenderingContext2D/save
Canvas.context.restore : https://developer.mozilla.org/en-US/docs/Web/API/CanvasRenderingContext2D/restore
requestAnimationFrame : https://developer.mozilla.org/en-US/docs/Web/API/Window/requestAnimationFrame
Paul Irish, requestAnimationFrame polyfill : http://www.paulirish.com/2011/requestanimationframe-for-smart-animating/

Call context.save to save the transformation matrix before you call context.fillRect.
Then whenever you need to draw your image, call context.restore to restore the matrix.
For example:
function drawScreen(){
context.save();
context.fillStyle="#333333";
context.fillRect(0,0, windowWidth, windowHeight);
context.restore();
context.drawImage(photo,0,0,imageDimensions.width*scale,imageDimensions.height*scale);
}
Also, to further optimize, you only need to set fillStyle once until you change the size of canvas.

HTML5 canvas zoom my drawings

Hello i create program like a paint on HTML5 canvas. I have problem i need create few tools drawing and zoom. I don't have idea how to create zoom without delay. Drawing example: http://jsfiddle.net/x5rrvcr0/
How i can zooming my drawings?
drawing code:
<style>
canvas {
background-color: #CECECE;
}
html, body {
background-color: #FFFFFF;
}
</style>
<script>
$(document).ready(function () {
var paintCanvas = document.getElementById("paintCanvas");
var paintCtx = paintCanvas.getContext("2d");
var size = 500;
paintCanvas.width = size;
paintCanvas.height = size;
var draw = false;
var prevMouseX = 0;
var prevMouseY = 0;
function getMousePos(canvas, evt) {
evt = evt.originalEvent || window.event || evt;
var rect = canvas.getBoundingClientRect();
if (evt.clientX !== undefined && evt.clientY !== undefined) {
return {
x: evt.clientX - rect.left,
y: evt.clientY - rect.top
};
}
}
$("#paintCanvas").on("mousedown", function(e) {
draw = true;
var coords = getMousePos(paintCanvas);
prevMouseX = coords.x;
prevMouseY = coords.y;
});
$("#paintCanvas").on("mousemove", function(e) {
if(draw) {
var coords = getMousePos(paintCanvas, e);
paintCtx.beginPath();
paintCtx.lineWidth = 10;
paintCtx.strokeStyle = "#000000";
paintCtx.moveTo(prevMouseX, prevMouseY);
paintCtx.lineTo(coords.x, coords.y);
paintCtx.stroke();
prevMouseX = coords.x;
prevMouseY = coords.y;
}
});
$("#paintCanvas").on("mouseup", function(e) {
draw = false;
});
});
</script>
<body>
<canvas id="paintCanvas"></canvas>
</body>

If you want to keep the pixelated effect in the zoom, you need to draw on a temp canvas, then only after copy that temp canvas to the main screen.
You no longer need to zoom in the temp canvas, just draw on 1:1 scale always. When copying to the view canvas, then you apply the zoom (and maybe translate) that you want.
Keep in mind that drawings are anti-aliased, so you when zooming you will see some shades of grey when drawing in black, for instance.
I kept the recording code of #FurqanZafar since it is a good idea to record things in case you want to perform undo : in that case just delete the last record entry and redraw everything.
http://jsfiddle.net/gamealchemist/x5rrvcr0/4/
function updatePaintCanvas() {
paintContext.clearRect(0, 0, paintContext.canvas.width, paintContext.canvas.height);
paintContext.save();
paintContext.translate(cvSize * 0.5, cvSize * 0.5);
paintContext.scale(scale, scale);
paintContext.drawImage(drawCanvas, -cvSize * 0.5, -cvSize * 0.5);
paintContext.restore();
}

Heres the updated fiddle: http://jsfiddle.net/x5rrvcr0/2/ with basic zooming functionality
If you draw multiple paths on mouse move then your sketch will appear broken or disconnected, instead you should only stroke a single path until "mouseup" event.
You can then store these paths in an array and later redraw them at different zoom levels:
function zoom(context, paths, styles, scale) {
context.clearRect(0, 0, context.canvas.width, context.canvas.height);
context.save();
applyStyles(context, styles);
scaleFromCenter(context, scale);
for (var i = 0; i < paths.length; i++) {
context.beginPath();
context.moveTo(paths[i][0].x, paths[i][0].y);
for (var j = 1; j < paths[i].length; j++)
context.lineTo(paths[i][j].x, paths[i][j].y);
context.stroke();
}
context.restore();
};

How to make a signature pad in AngularJS or pure javascript without Jquery?

I'm making a mobile app and I need the simplest possible signature pad for javascript for implementing with AngularJS but I cant find any resources anywhere.
All the plugins use jquery which I refuse to load since it's a mobile app and jquery would just bloat the app. Does anyone know of a tutorial I could use or could give me basic directions how would I do this, what elements do i need... I'd really appriciate it!
UPDATE: I just wrote something basic as it was answered here. But it doesn't work. Can anyone indentifiy why? I have a directive that i inject into html as an element (). Console is not returning any exceptions.
sig.directive("signatureDirective", function () {
return {
template: '<canvas id="canvas" width="500" height="100" style="border: 1px solid #ccc;"></canvas>',
restrict: 'E',
link: function (scope, element, attrs) {
var canvas = $(element);
var context = canvas.getContext("2d");
var clickX = new Array();
var clickY = new Array();
var clickDrag = new Array();
var paint;
$(element).addEventListener("mousedown", mouseDown, false);
$(element).addEventListener("mousemove", mouseXY, false);
document.body.addEventListener("mouseup", mouseUp, false);
$(element).addEventListener("touchstart", mouseDown, false);
$(element).addEventListener("touchmove", mouseXY, true);
$(element).addEventListener("touchend", mouseUp, false);
document.body.addEventListener("touchcancel", mouseUp, false);
function draw() {
context.clearRect(0, 0, 500, 100);
context.strokeStyle = "#000000";
context.lineJoin = "miter";
context.lineWidth = 2;
for (var i = 0; i < clickX.length; i++) {
context.beginPath();
if (clickDrag[i] && i) {
context.moveTo(clickX[i - 1], clickY[i - 1]);
} else {
context.moveTo(clickX[i] - 1, clickY[i]);
}
context.lineTo(clickX[i], clickY[i]);
context.stroke();
context.closePath();
}
}
function mouseDown(e) {
var mouseX = e.pageX - this.offsetLeft;
var mouseY = e.pageY - this.offsetTop;
paint = true;
addClick(e.pageX - this.offsetLeft, e.pageY - this.offsetTop);
draw();
}
function mouseUp() {
paint = false;
}
function mouseXY(e) {
if (paint) {
addClick(e.pageX - this.offsetLeft, e.pageY - this.offsetTop, true);
draw();
}
}
}
};
});

Take a look at the canvas element. You can directly manipulate the pixel data and draw primitive shapes with a canvas and as a HTML element, it supports all of the mouse manipulation events that you will need out of the box.
<canvas id='signature' width='300' height='50'></canvas>
Something like this:
var canvas, context;
var pen = {};
canvas = document.getElementById('signature');
context = canvas.getContext('2d');
canvas.addEventListener('mousemove', function(e) {
pen.x = e.pageX;
pen.y = e.pageY;
});
canvas.addEventListener('mousedown', function(e) {
context.fillRect(pen.x, pen.y, 1, 1);
});
// when you're done, you can get a base 64 encoded png version
// of the image data.
context.getImageData();
This is a pretty primitive way of doing it, using the fillRect method to simulate pixels. If you wanted to look at heavier solutions, you could try doing something with click points and bezier curves.
If you're using Angular, then you'll almost certainly want to wrap this up in a directive as well.
EDIT
First off, you're using a strange mix of jQuery and plain js. All of the lines that look like this, won't work.
$(element).addEventListener("mousedown", mouseDown, false);
Should look like this:
$(element).on("mousedown", mouseDown, false);
Next off, you seem to be calling a method called addClick which isn't actually there.

Tutorial on isPointinPath not copying to my computer

I found this helpful tutorial
http://www.rgraph.net/blog/2013/february/an-example-of-the-html5-canvas-ispointinpath-function.html
i copied it into my own text editor and nothing happens when I open it. I changed it by adding a declaration
<!DOCTYPE html>
<html lang="en">
<head>
<title>exampleMouseOver</title>
</head>
<script>
window.onload = function (e)
{
var canvas = document.getElementById('cvs');
var context = canvas.getContext('2d');
// Draw the rectangle
context.beginPath();
context.rect(50,50,100,100);
context.fill();
context.fillStyle = 'red';
// Draw the circle
context.beginPath();
context.arc(450,175, 50, 0,2 * Math.PI, false);
context.fill();
context.fillStyle = 'green';
// Draw the shape
context.beginPath();
context.moveTo(250,100);
context.lineTo(350,175);
context.lineTo(325,215);
context.lineTo(185,195);
context.fill();
canvas.onmousemove = function (e)
{
var canvas = e.target;
var context = canvas.getContext('2d');
// This gets the mouse coordinates (relative to the canvas)
var mouseXY = RGraph.getMouseXY(e);
var mouseX = mouseXY[0];
var mouseY = mouseXY[1];
// Replay the rectangle path (no need to fill() it) and test it
context.beginPath();
context.rect(50,50,100,100);
if (context.isPointInPath(mouseX, mouseY)) {
canvas.style.cursor = 'pointer';
return;
}
///////////////////////////////////////////////////////////////
// Replay the circle path (no need to fill() it) and test it
context.beginPath();
context.arc(450,175, 50, 0,2 * Math.PI, false);
if (context.isPointInPath(mouseX, mouseY)) {
canvas.style.cursor = 'pointer';
return;
}
///////////////////////////////////////////////////////////////
// Replay the irregular shape path (no need to fill() it) and test it
context.beginPath();
context.moveTo(250,100);
context.lineTo(350,175);
context.lineTo(325,215);
context.lineTo(185,195);
if (context.isPointInPath(mouseX, mouseY)) {
canvas.style.cursor = 'pointer';
return;
}
///////////////////////////////////////////////////////////////
// Return the cursor to the default style
canvas.style.cursor = 'default';
}
}
</script>
</html>

You'll need a body element and a canvas element. Also your script element needs to be inside either your head element or your body element.
The following is what the sample was using, but did not include in their sample code:
<body>
<canvas id="cvs" width="600" height="250" style="border: 1px solid gray; cursor: pointer;">[No canvas support]</canvas>
</body>
Edit: Additionally the code is calling "RGraph.getMouseXY(e)", which is in a library file that you are not referencing. You can either add a reference to that library or get the mouse position yourself.
If you want to use other parts of the RGraph library, for drawing charts, you should add the library. To add the library you should follow the instructions on the RGraph site related to downloading and starting with RGraph (http://www.rgraph.net/docs/starting-with-rgraph.html).
If this was just a sample that happened to do what you wanted to do, you should get the mouse position yourself. You can do this by changing these lines:
var mouseXY = RGraph.getMouseXY(e);
var mouseX = mouseXY[0];
var mouseY = mouseXY[1];
to these:
var mouseX = e.clientX - canvas.getBoundingClientRect().left;
var mouseY = e.clientY - canvas.getBoundingClientRect().top;
This may not be the most robust solution, but it should suffice for your purposes. Essentially you are getting the mouse position in the window, then subtracting the top-left of the canvas in the window, so that you are left with the mouse position in the canvas.