GSAP, jQuery以及Zepto（CSS3）动画（渐变）性能测试

HTML5 Animation Speed Test: GSAP vs CSS Transitions (Zepto) vs jQuery 大体的比较结果是： 1. 当粒子数量少于100时，渲染效果类似； 2. 当粒子超过1000时，GSAP稍有延迟后正常渲染，而jQuery和CSS3已经很卡。

《three.js如何实现贴图模糊效果 - 使用canvas》申请公开！

第三方库
作品简介 23356
作品推荐
使用帮助
常用链接

HTML
CSS
JavaScript
Result

编译

JavaScript

/* http://www.greensock.com/js/speed.html */

var $start = $("#start"),
    $dotQtyInput = $("#dotQuantity"),
    $engineInput = $("#engine"),
    $propertiesInput = $("#properties"),
    $instructions = $("#instructions"),
    $field = $("#field"),
    $window = $(window),
    $inputs = $("select"),
    inProgress = false,
    tests = {},
    duration, radius, centerX, centerY, dots, rawDots, currentTest, startingCSS;

/**
 * The goal of this test is to compare how various animation engines perform under pressure, taking relatively common
 * animation tasks and running a lot of them at once to see raw performance. The goal is NOT to figure out the most
 * efficient way to move dots in a starfield pattern.
 *
 * The same code drives everything except the actual tweens themselves. Every test in the "tests"
 * object has 4 properties:
 *
 * 		- milliseconds [boolean] - true if the duration should be defined in milliseconds
 *
 * 		- wrapDot [function] - when each dot <img> is created, it is passed to the wrapDot() method
 * 							   and whatever is returned gets stored in the array of dots to tween. This
 * 							   is useful to improve performance of things like jQuery because
 * 							   instead of passing the dom element to the tween() method (which would require
 * 							   jQuery to then query the dom and wrap the element in an engine-specific object
 * 							   before calling animate() on it), a native object can be used. Basically it lets you
 * 							   cache the dot's wrapper for better performance.
 *
 * 		- tween [function] - This is the core of the whole test. tween() is called for each dot, and the dot is
 * 							 passed as a parameter. The tween() function should set the dot's cssText to the
 * 							 startingCSS value (which just places the dot in the middle of the screen and sets its
 * 							 width/height to 1px) and then after a random delay between 0 and the duration of the tween,
 * 							 it should tween the dot at a random angle, altering either the left/top values or transform
 * 							 translate() values accordingly, and adjust the size to 32px wide/tall using either
 * 							 width/height or transform:scale(). Then, after the tween is done, it should call the tween()
 * 							 method again for that dot. So the same dot will just continuously tween outward from the
 * 							 center at random angles and at random delay values.
 *
 * 		- stop [function] - This function is called when the user stops the test. The dot is passed as a parameter.
 * 							The function should immediately stop/kill the tween(s) of that dot (or all dots - that's fine too).
 *
 *  - nativeSize [Boolean] - true if the beginning width/height of the image should be its native size
 *                          (typically necessary for transforms, but not when we're animating width/height).
 *
 * I don't claim to be an expert at the various other animation engines out there, so if there are optimizations
 * that could be made to make them run better, please let me know. I tried to keep things as fair as possible.
 **/

//jQuery normal (top/left/width/height)
tests.jquery_normal = {
    milliseconds: true,
    wrapDot: function(dot) {
        return jQuery(dot); //wrap the dot in a jQuery object in order to perform better (that way, we don't need to query the dom each time we tween - we can just call animate() directly on the jQuery object)
    },
    tween: function(dot) {
        var angle = Math.random() * Math.PI * 2;
        dot[0].style.cssText = startingCSS;
        dot.delay(Math.random() * duration).animate({
            left: Math.cos(angle) * radius + centerX,
            top: Math.sin(angle) * radius + centerY,
            width: 32,
            height: 32
        }, duration, "cubicIn", function() {
            tests.jquery_normal.tween(dot)
        });
    },
    stop: function(dot) {
        dot.stop(true);
    },
    nativeSize: false
};

//GSAP normal (top/left/width/height)
tests.gsap_normal = {
    milliseconds: false,
    wrapDot: function(dot) {
        return dot; //no wrapping necessary
    },
    tween: function(dot) {
        var angle = Math.random() * Math.PI * 2;
        dot.style.cssText = startingCSS;
        TweenLite.to(dot, duration, {
            css: {
                left: Math.cos(angle) * radius + centerX,
                top: Math.sin(angle) * radius + centerY,
                width: 32,
                height: 32
            },
            delay: Math.random() * duration,
            ease: Cubic.easeIn,
            overwrite: "none",
            onComplete: tests.gsap_normal.tween,
            onCompleteParams: [dot]
        });
    },
    stop: function(dot) {
        TweenLite.killTweensOf(dot);
    },
    nativeSize: false
};

//GSAP transforms (translate()/scale())
tests.gsap_transforms = {
    milliseconds: false,
    wrapDot: function(dot) {
        return dot; //no wrapping necessary
    },
    tween: function(dot) {
        var angle = Math.random() * Math.PI * 2;
        TweenLite.set(dot, {
            css: {
                x: 0,
                y: 0,
                scale: 0.06
            },
            overwrite: "none"
        });
        TweenLite.to(dot, duration, {
            css: {
                x: (Math.cos(angle) * radius),
                y: (Math.sin(angle) * radius),
                scaleX: 2,
                scaleY: 2
            },
            delay: Math.random() * duration,
            ease: Cubic.easeIn,
            overwrite: "none",
            onComplete: tests.gsap_transforms.tween,
            onCompleteParams: [dot]
        });
    },
    stop: function(dot) {
        TweenLite.killTweensOf(dot);
    },
    nativeSize: true
};

//Zepto normal (top/left/width/height)
tests.zepto_normal = {
    milliseconds: true,
    wrapDot: function(dot) {
        return Zepto(dot); //wrap the dot in a jQuery object in order to perform better (that way, we don't need to query the dom each time we tween - we can just call animate() directly on the jQuery object)
    },
    tween: function(dot) {
        var angle = Math.random() * Math.PI * 2;
        dot[0].style.cssText = startingCSS;
        //Zepto's delay feature performs TERRIBLY under pressure, so we use a setTimeout() instead to improve performance.
        setTimeout(function() {
            if (!dot.isKilled) { //Zepto doesn't have a feature that allows us to kill tweens, so we simply set our own "isKilled" property to true when the tween is supposed to be killed and then stop the recursion thereafter which gives us a somewhat similar effect.
                dot.animate({
                    left: Math.cos(angle) * radius + centerX,
                    top: Math.sin(angle) * radius + centerY,
                    width: 32,
                    height: 32
                }, duration, "cubic-bezier(0.550, 0.055, 0.675, 0.190)", function() {
                    tests.zepto_normal.tween(dot)
                });
            }
        }, duration * Math.random());
    },
    stop: function(dot) {
        dot.isKilled = true;
    },
    nativeSize: false
};

//Zepto transforms (translate()/scale())
tests.zepto_transforms = {
    milliseconds: true,
    wrapDot: function(dot) {
        return Zepto(dot); //wrap the dot in a jQuery object in order to perform better (that way, we don't need to query the dom each time we tween - we can just call animate() directly on the jQuery object)
    },
    tween: function(dot) {
        //I couldn't just set the css() reliably with Zepto (it failed), so I was forced to use a zero-duration animate() call. It's fair, though, because we actually do a zero-duration tween for TweenLite too.
        dot.animate({
            translateX: "0px",
            translateY: "0px",
            rotateY: "0rad",
            rotateX: "0rad",
            scale: "0.06,0.06"
        }, 0);
        //Zepto's delay feature performs TERRIBLY under pressure, so we use a setTimeout() instead to improve performance.
        setTimeout(function() {
            if (!dot.isKilled) { //Zepto doesn't have a feature that allows us to kill tweens, so we simply set our own "isKilled" property to true when the tween is supposed to be killed and then stop the recursion thereafter which gives us a somewhat similar effect.
                var angle = Math.random() * Math.PI * 2;
                dot.animate({
                    translateX: (Math.cos(angle) * radius) + "px",
                    translateY: (Math.sin(angle) * radius) + "px",
                    scale: "2,2",
                    delay: duration * Math.random()
                }, duration, "cubic-bezier(0.550, 0.055, 0.675, 0.190)", function() {
                    tests.zepto_transforms.tween(dot);
                });
            }
        }, duration * Math.random());
    },
    stop: function(dot) {
        dot.isKilled = true;
    },
    nativeSize: true
};

} else {
        $start.html(" START ");
        $start.css({
            backgroundColor: "#9af600",
            background: "linear-gradient(to bottom, #9af600 0%,#71B200 100%"
        });
        TweenLite.to($instructions, 0.7, {
            autoAlpha: 1,
            delay: 0.2
        });
        $inputs.prop("disabled", false);
        calibrateInputs();
        $field.css({
            pointerEvents: "auto"
        });
        //stop the tweens and remove the dots.
        i = dots.length;
        while (--i > -1) {
            currentTest.stop(dots[i]);
            $field[0].removeChild(rawDots[i]); //removes dot(s)
        }
        dots = null;
        rawDots = null;
    }
}

function createDots() {
    var i = parseInt($dotQtyInput.val()),
        dot;
    dots = [];
    rawDots = [];
    while (--i > -1) {
        dot = document.createElement("img");
        dot.src = "https://s3-us-west-2.amazonaws.com/s.cdpn.io/16327/dot.png";
        dot.width = 1;
        dot.height = 1;
        dot.id = "dot" + i;
        dot.style.cssText = startingCSS;
        $field[0].appendChild(dot);
        rawDots.push(dot);
        dots.push(currentTest.wrapDot(dot));
    }
}

$start.click(toggleTest);
$inputs.change(calibrateInputs);

jQuery.easing.cubicIn = $.easing.cubicIn = function(p, n, firstNum, diff) { //we need to add the standard CubicIn ease to jQuery
    return firstNum + p * p * p * diff;
}
jQuery.fx.interval = 16; //ensures that jQuery refreshes at roughly 60fps like GSAP and the others to be more even/fair.
$propertiesInput.prop("disabled", true);

编译

CSS

body {
    background-color: black;
    margin: 0;
    padding: 0;
    color: #eee;
    font-family: Signika Negative, sans-serif;
    font-weight: 300;
    font-size: 1.15em;
    user-select: none;
    -webkit-user-select: none;
}
html,
body {
    height: 100%;
    overflow: hidden;
}
h1 {
    font-weight: 400;
    font-size: 2em;
    line-height: 1em;
    margin-bottom: 0.1em;
    color: white;
}
a,
a:hover,
a:visited {
    color: #71b200;
}
#controls {
    display: table-row;
    background-color: #555;
    background: linear-gradient(to bottom, #777 0%, #444 100%);
    padding: 10px 10px 10px 5px;
    z-index: 1000;
}
#controls form li {
    display: table-cell;
    padding: 12px 6px 10px 6px;
    vertical-align: middle;
    text-shadow: 1px 1px 1px #000;
}
#instructions {
    width: 82%;
    margin-left: 8%;
    padding-top: 1em;
    line-height: 1.5em;
    color: #ccc;
}
#demo {
    display: table;
    width: 100%;
    height: 100%;
}
#field {
    position: relative;
    display: table-cell;
    width: 100%;
    height: 100%;
    overflow: hidden;
    z-index: -100;
    border-top: 1px solid #777;
}
#start {
    color: black;
    border-radius: 6px;
    padding: 5px 18px;
    border: 2px solid black;
    background: #9af600;
    background: linear-gradient(to bottom, #9af600 0%, #71b200 100%);
    cursor: pointer;
    text-shadow: none;
    font-weight: 400;
}

HTML

HTML5 Animation Speed Test: GSAP vs CSS Transitions (Zepto) vs jQuery 大体的比较结果是： 1. 当粒子数量少于100时，渲染效果类似； 2. 当粒子超过1000时，GSAP稍有延迟后正常渲染，而jQuery和CSS3已经很卡。

编辑器支持Emmet

快捷键

HTML5 Animation Speed Test: GSAP vs CSS Transitions (Zepto) vs jQuery 大体的比较结果是： 1. 当粒子数量少于100时，渲染效果类似； 2. 当粒子超过1000时，GSAP稍有延迟后正常渲染，而jQuery和CSS3已经很卡。

three.js webgpu渲染器加载gltf模型代码示例（Chrome 113版本以上）

使用three.js和PlaneBufferGeometry构建的3d交互波浪背景动画

编辑器支持Emmet

快捷键

语言：

语言：

导入代码模板：