Webkit lags with large images in canvas - javascript

i have some problems with webkit (chrome/safari) and Canvas with large images.
For my customer we had to write an animated intro movie with changing images in 80ms per frame on a canvas.
There are 130 frames and we had to put the single images into 4 sprites to reduce the single-request load of the page.
Every sprite size is about 2,5MB and we just slice the part of the needed frame. So far so possible.
don't want to bother you with to much code. But in the end it's just about:
this.$context.drawImage(img, 0, top, img.width-1 ,(img.height / sequenceCount)-1, 0, 0, this.$canvas.width, this.$canvas.height);
witch is called within 80ms timeouts.
We did that and it works on all major browser. But with webkit every change to the next sprite causes a hard lag of about 400ms.
The same issue was with IE9 but it could be fixed with drawing every sprite once in the beginning
if (Browser.ie9) {
for(var x = 0; x < this.sequence[0].sprites.length; x++){
this.draw(this.sequence[0].sprites[x].obj, 0, 1);
}
this.$context.clearRect(0, 0, this.$canvas.width, this.$canvas.height);
}
(The draw function include just the drawImage function from the example before.)
But with webkit browsers i get still this lagging time of ~400ms.
Yes, the images are preloaded. So this is not the issue.
Any ideas what this could be?
Thanks for your help!

Every time you get the canvas or image width, you are accessing the DOM, which is generally slower than accessing JS memory. You might see an improvement if you store those values on resize, and use the stored value instead.
var canvasWidth, canvasHeight, imgWidth, imgHeight;

Related

Performance: Html5 Canvas / Javascript - Moving the background image using canvas.drawImage vs canvas.style

I have a background that scrolls up, down, left and right based on position, and was wondering if there is any significant performance difference into using drawImage() every time the position changes vs just drawing the entire background once and moving the canvas around using canvas.style
Using canvas.drawImage()
//cut piece and draw image to fill canvas every time position changes
canvas.drawImage(backgroundImg, 0 + positionX, 0 + positionY, c.width, c.height, 0, 0, c.width, c.height);
VS
Using canvas.style
canvas.width = widthofbackgroundImg;
canvas.height = heightofbackgroundImg;
//draw entire background only once
canvas.drawImage(backgroundImg, 0, 0, widthofbackgroundImg, heightofbackgroundImg);
..then
//move the entire canvas on every position change
canvas.style.left = positionX;
canvas.style.up = positionY;
Those are two different approaches which are depending on the browser processing (different browsers might set different priorities in rendering).
For sure, it does not make sense to move the canvas image by drawImage, if nothing else changes and there is just an initial draw.
Instead of going with canvas.style.left and canvas.style.right I would suggest you to use transform = translate3d(x, y, 0) and test the performance improvement with your dev tools. This will be rendered via the GPU and may increase speed. It will prevent reflow which is time consuming in general.
Also the reading from and writing to the DOM in alternation might slow down performance. It will help to first make all reading (get positions, scroll position or the like), save it in variables and then write everything.

How redraw on canvas resize without blurring?

When the canvas element is resized (via the style changing) I also want to scale the canvas' drawn image as well. I cannot just change the height/width as this causes the canvas to clear itself, so I do:
Create a temporary canvas element
Draw the current canvas' image onto that temporary canvas
Resize the current canvas
Draw the temp canvas' image back to the current canvas but scaled to the new size
This results in some blurring - very noticeable after many resizes (example: when dragging to resize). How would I do this without any blurring?
EDIT: Turning off image smoothing (context.webkitImageSmoothingEnabled = false;) does not fix the problem, it simply makes it redraw it more and more jagged until the image looks nothing like the original after a number of resizes.
Called on resize event:
var tmpCanvas = null;
//Make a temporary canvas
tmpCanvas = document.createElement( "canvas" );
//Set its size to be equal
tmpCanvas.height = originalCanvas.height;
tmpCanvas.width = originalCanvas.width;
//Draw our current canvas onto it
tmpCanvas.getContext( "2d" ).drawImage( originalCanvas, 0, 0 );
//Set new dimensions
originalCanvas.width = originalCanvas.offsetWidth;
originalCanvas.height = originalCanvas.offsetHeight;
var originalContext = originalCanvas.getContext( "2d" );
//Set background and colors
originalContext.fillStyle = "#ffffff";
originalContext.strokeStyle = "#000000";
//Set paintbrush
originalContext.lineWidth = 4;
originalContext.lineCap = "round";
//Fill background as white
originalContext.fillRect( 0, 0, originalCanvas.width, originalCanvas.height );
//We have a saved signature
if ( SignatureCanvas.hasSignature === true )
{
//Draw it back but scaled (results in blurred image)
originalContext.drawImage( tmpCanvas, 0, 0, tmpCanvas.width, tmpCanvas.height, 0, 0, originalCanvas.width, originalCanvas.height );
/**
* This results in a blurred image as well
//Draw it back but scaled
originalContext.scale( originalCanvas.width / tmpCanvas.width, originalCanvas.height / tmpCanvas.height );
originalContext.drawImage( tmpCanvas, 0, 0, tmpCanvas.width, tmpCanvas.height, 0, 0, tmpCanvas.width, tmpCanvas.height );
*/
}
Is there a way to get the strokes and "scale" all those points and redraw?
Instead of taking the rendered image from the original canvas, actually redraw the image. By that, I mean execute the same logic you executed against the original canvas, but with the points involved scaled to the new size.
If you can, think about using SVG instead. It scales well by its nature.
Edit: Another option I've thought of is to simply use a gigantic canvas to start with. Sizing down tends to look better than sizing up, especially with smoothing on.
Edit II: The original answer was irrelevant, though the comment I had made is relevant, and am now promoting it and editing it to be an answer...and the answer I had given was not all that great anyway **.
Of course if you scale up raster graphics, that is, from an image with a smaller pixel dimensions of pixels, create an image with higher pixel dimensions, you are going to get blurred images. By scaling up, you're making a low resolution picture high resolution, but without the high resolution details.
There's absolutely no way around that blurriness unless you make multiple additional assumptions about your raster image like the only gray you'd see is at an image edge, or corners can only occur at apparent inflection points where the angle between the tangents of the joined curves must be 100 degrees or less. Essentially, you'd have to give additional information so that your higher resolution image can have detail "filled in". It's not all that terribly different from reverse engineering an SVG from a raster.
So, you appear to want to emulate is scaling vector graphics, in which the only solution is to save the commands, draw a SVG, or draw to a bigger canvas like Stuart Branham suggested.
** I had originally proposed that invoking drawImage would distort the pixels even if it were not scaled, and that it would be better to work with the actual pixel data. If that's true, I can't find proof, but that's irrelevant, as he wanted his image scaled up, without blurring...which is impossible, as I just mentioned.

Drawing a Canvas into a smaller Canvas not working

For a Project I want to take the content of a canvas (Called SAVE_CV) and display it in another, smaller canvas.
Some things that I am aware of so far that could be causing me problems: resizing a canvas clears its content, the JS-size of a canvas is different from the CSS-size.
I want the smaller canvas to be 500px wide and appropriately high.
function restoreTaggingCV() {
var cv = document.getElementById( 'taggingCV' );
var ctx = cv.getContext( "2d" );
var styleHeight = SAVE_CV.height * 500 / SAVE_CV.width;
ctx.drawImage(SAVE_CV, 0, 0, cv.width, cv.height);
}
This is my Code so far. Whenever I try to resize the smaller canvas appropriately it only gives me a blank canvas with nothing in it. I tried to set the size with "cv.height = X" and "cv.style.height = styleHeight + 'px'" but neither worked. Also I would love to set the width of the canvas using CSS.
Appreciate any help.
EDIT
I want the image in a picture because later I want the user to mark areas in the smaller version which I then want to use to create individual imaged from the big version. I want to visualise thise area to the user. I probably could do all this by using an image and putting divs over it or something but I just fell more comfident using a canvas since I am pritty new to HTML and CSS.
Try using the CanvasRenderingContext2d.prototype.scale method. It sets the scale factor of the canvas and renders anything in the current state with it's dimensions multiplied by the factor.
So before you use the drawImage function, you scale the context appropriately (in this case, down). For example:
context.save();
context.scale(0.5, 0.5);
context.drawImage(canvas, 0, 0);
context.restore();
This would render the canvas on the context at 0.5 times it's current size. See in this fiddle how I used it to mirror a larger canvas onto a smaller, separate one.
Canvas objects don't like to be resised. After drawing Your image simply convert it toDataURL() and set as image source. They You may resize image as you want.
$img.attr('src',canvas.toDataURL());

Problems clearing canvas when animating a clipping region

I'm trying to accomplish an effect similar to what you might see on the cartoon Chowder (example link) , where shapes serve as masking layers for a texture underneath that stays static. I've begun playing around with this idea by creating a render loop that clears the canvas, saves it's state, then draws a rectangular clipping region, followed by drawing the background texture that occupies the entire width and height of the canvas.
Here's the draw function:
function draw()
{
context.clearRect(0,0, 640, 480);
context.save();
x += velocityX;
y += velocityY;
context.rect(x, y, 40, 40);
context.clip();
context.drawImage(image, 0,0, 640, 480);
context.restore();
}
Basically it just runs at 60 frames per second, updating the position of the rectangle and clipping a background image inside the clipping region. (I know the code isn't structured perfectly, but I was just experimenting to see if this effect was even possible on the canvas).
http://jsfiddle.net/JERje/86/
The problem I seem to be having is that the clipping area from the previous iteration of the loop hangs around creating the weird effect that you see in the fiddle above. I've tried reordering everything in the draw() step of the loop, but the only thing that seems to work is the canvas.width = canvas.width trick for clearing the screen. I'd like to avoid this method of clearing the screen, since it doesn't seem to work in IE, and it also destroys the canvas state. clearRect() should work to clear the screen. What am I doing wrong?
You're using the same HTML5 Canvas paperback I am aren't you.
If you set up an adhoc canvas as I did on your jsfiddle like so:
var newCanvas = document.createElement('canvas');
newCanvas.getContext("2d").drawImage(image,0,0);
A function such as this would be able to hack a section out of that canvas:
context.putImageData(newCanvas.getContext("2d").getImageData(x,y,40,40),x,y);
Thus giving you the chowder effect. Good show man, good luck. Pst me if it doesn't work
EDIT: However this solution will ignore some context scaling transformations. Just be smart about how you handle scale on your own (and you really should be anyways if you want the true "chowder" effect)
So, feel pretty dumb about this, but apparently when you call rect() you also have to make sure to call closePath afterwards in order to close the clipping area. Glad I figured it out finally, now on to adding multiple layers!
Here's the working fiddle: http://jsfiddle.net/JERje/129/

drawImage on canvas has weird aspect ratio in firefox and other problems

I am running firefox 3.5.6.
I want to display an image on a canvas and draw a couple lines on it. It needs to display properly in firefox and internet explorer (using excanvas).
Here is what I am getting:
The top image is what I see in IE8, the bottom is what I see in firefox.
IE seems to be a bit messed up as far as the canvas is the wrong size but firefox is going crazy! What gives with this aspect ratio? Why does the second half of my arc not appear?
Also, some times firefox just flat out doesn't show anything.
Here is my code by the way.
Aspect ratio problem
If you don't set a width on the canvas element, it defaults to 300x150. In your CSS, you set the style to 94x120, so it scales the image to that size. To fix it, you need to either set the width and height in the HTML, or with JavaScript.
In HTML:
<canvas id="c" width="94" height="120">Ugh, this just ain't gonna work</canvas>
In JavaScript (with jQuery):
$('canvas').attr('width', '94').attr('height', '120');
Internet Explorer's incorrect size
Adding the size to the canvas element should fix this problem too. Since IE is using VML instead of a canvas to render the image, the CSS rule for canvas won't apply. excanvas should see the specified size and apply it in IE.
Missing the second half of the arc
The simpleArc function doesn't work in Firefox when the amplitude is negative. The problem is that a negative amplitude results in a negative radius for the arc, which is illegal according to the canvas spec. It should actually throw an INDEX_SIZE_ERR exception, but Firefox just seems to ignore the call.
There are two possible solutions (basically; there are several ways you could accomplish either): when you pass a negative amplitude, either calculate the parameters for the arc taking into account the negative radius (with a different center point and angles, etc.), or change the sign and use transformations to rotate the arc. I implemented the second solution like this:
ctx.simpleArc = function(x,y, length, amplitude) {
var rotate = false;
// Check whether we need to rotate the image
if (amplitude < 0) {
rotate = true;
amplitude = -amplitude;
}
var radius = amplitude/2+ length*length/(8*amplitude);
var outerAngle = Math.asin((radius-amplitude)/radius);
var innerAngle = Math.PI - 2*outerAngle;
// The translate/rotate/translate steps could be combined into one matrix
// transformation, but I think this is clearer and less error-prone.
if (rotate) {
this.save(); // So we can easily undo the transformation
this.translate(x + length, y);
this.rotate(Math.PI);
this.translate(-length, -y);
}
this.arc(x+length/2, y+(radius-amplitude), radius, -(outerAngle+innerAngle), -outerAngle, false);
// Reset the transformation matrix to its original value
if (rotate) {
this.restore();
}
return this;
}
Firefox not showing anything
In your code, you create the image and set the source, but it may not be loaded before the rest of the code get's executed. The image loads asynchronously, and when you draw the image onto the canvas, it doesn't wait for it to finish. You will need to call the code that uses the image from an onload event.
var img = $('<img>');
img[0].onload = function() {
ctx.drawImage(img[0], 0, 0);
ctx.strokeStyle = "blue";
ctx.simpleStroke(function(ctx) { ctx.simpleArc(0, 70, img_w/2, 3)});
ctx.simpleStroke(function(ctx) { ctx.simpleArc(img_w / 2, 70, img_w/2, -3)});
};
// I moved this so it happens after you set the `onload` event, because I
// think IE won't call `onload` if it happens to already be loaded.
img.attr('src', 'shortcylinder.png');
You could also pre-load all the images you will need instead of creating them when you need them. You would still need to prevent the code from running until all the images are loaded.
I've recently noticed that using style to define width & height for canvas elements caused an issue like this. Taking from an earlier example
This works in FF 9.0.1 Mac
<canvas id="c" width="94" height="120">Ugh, this just ain't gonna work</canvas>
vs.
This had similar display issues as your example, in FF 9.0.1 Mac
<canvas id="c" style="width:94;height:120;">Ugh, this just ain't gonna work</canvas>
Maybe that's it?

Categories