Monday, May 16, 2016

CSS3 Smoke Animation Effect

The comic today uses quite a bit of CSS3 animation. This is a rather new thing for me - I've been using mostly JavaScript-powered animations on Amphibian.com since it started. But after I did the rain animation using CSS back in March (see the comic here), I've been warming up to the idea of more CSS and less JavaScript to move things around.

For this task, I wanted to make some animated smoke come out of the frogs' rocket ship before the launch. My style is mostly just simple geometric shapes arranged to look like things, so using circular DIV elements to look like puffs of smoke was fine with me. I found a great starting point by Andrea Verlicchi on CodePen, and then modified it for my comic.

Here's the basic idea - puffs of smoke emanate from a given source element. They move downward and off to the side while fading away.

The puffs of smoke will be represented by rounded SPAN elements, with this CSS applied to them:

span.smokepuff {
    display: block;
    position: absolute;
    bottom: -35px;
    left: 50%;
    margin-left: -20px;
    height: 0px;
    width: 0px;
    border: 35px solid #4b4b4b;
    border-radius: 35px;
    left: -14px;
    opacity: 0;
    transform: scale(0.2);
}

The above styling just makes them round, grey, and positioned absolutely in their container. I also have two animation keyframes defined, one for the down-and-left movement and one for the down-and-right movement:

@keyframes smokeL {
    0% {
        transform: scale(0.2) translate(0, 0);
    }
    10% {
        opacity: 1;
        transform: scale(0.2) translate(0, 5px);
    }
    100% {
        opacity: 0;
        transform: scale(1) translate(-50px, 80px);
    }
}

@keyframes smokeR {
    0% {
        transform: scale(0.2) translate(0, 0);
    }
    10% {
        opacity: 1;
        transform: scale(0.2) translate(0, 5px);
    }
    100% {
        opacity: 0;
        transform: scale(1) translate(50px, 80px);
    }
}

The above keyframe definitions define an animation that will move the smoke puffs lower by 80 pixels and 50 pixels to either side while at the same time scaling them up and fading them out. It defines 3 steps: 0% (the start), 10% (moved a little down), and 100% (moved completely down and over). There's one for the left, smokeL, and one for the right, smokeR.

Note: if you care about being compatible with slightly older browsers, you would want copies of these with @-moz-keyframes and @-webkit-keyframes as the names as well as adding -moz-transform and -webkit-transform to them all! I left that out here to keep the example simpler!

I said this was pure CSS3 animation, but there's still a little JavaScript involved. It doesn't really do the animating, but I use some code to generate the puffs in the first place. Something like this:

function createSmoke(time, num) {

    var timeGap = (time / num); 

    for( var i = 0; i < num; i++) {

        var delay = (timeGap * i) + 's';

        var aniName = "smokeL";
        if (((i+1) % 2) == 0) {
            aniName = "smokeR";
        }

        var aniStyle = "animation: " + aniName + " " + duration + " " + delay + " infinite";
        $('#smoker').append('<span class="smokepuff" style="' + aniStyle + '"></span>');

    }

}

When this function is called, you give it the length of the animation and the number of puffs of smoke you want. It figures out how much of a delay there should be between each puff's animation starting based on those two values. For example, if you want the animation to run 5 seconds and have 10 puffs of smoke, the first puff would have no delay, the second would have a delay of ( 5 / 10 ) * 1, the third a delay of (5 / 10 ) * 2, and so on. In this example, that just means add a half-second delay for each puff you generate. Also, each time through the loop, it alternates between the smokeL and smokeR animations so that every other puff moves in the opposite direction. One final piece of the total animation style is to set the repeat-count to infinite, so the puffs keep on coming! The function generates new SPAN tags with these animation styles applied and appends them to the parent element, which here is named smoker. It's just a DIV somewhere on the page - all the puffs of smoke will appear to come out of it.

The finished animation. Don't miss it!

I think the effect turned out great. Even though I used a little JavaScript to create the elements, doing all the animation with JavaScript would have been much, much more complicated. I may consider replacing all the old JavaScript-powered animation on Amphibian.com with CSS animation, and fix up my animation editor to go with it!

Now don't miss the smoke effect in today's comic! If you're reading this on the publication day (16 May 2016) the countdown to launch will be live! That means the last frame of the comic will change and do different things right up to the launch time! Keep watching it!

Amphibian.com comic for 16 May 2016

Monday, May 9, 2016

Add Gamepad Support to a Phaser Game

It seems like I haven't written a blog post in forever! It's actually been less than 2 weeks. Right before I took a break, I wrote a little about my new NES-style Bluetooth gamepad. I've been trying off-and-on ever since then to get my 8-bit style platformer to work with it, and today I finally had some success!

First of all, it needs to be said that support for gamepads in the browser is very inconsistent. The W3C's Gamepad API document is still a working draft after all these years (I first read about it and tried it out in 2013). It seems as though Mozilla and Google have some different opinions on how it should work, because the way you interact with the devices varies significantly between Firefox and Chrome. Phaser provides gamepad support through the Gamepad object, but the documentation carries a warning about the volatility of the specification.

Here's what I learned when I tried to use it...

I started with some of the examples on Phaser's site. They worked, most of the time. Let me explain. In theory, working with a gamepad in Phaser is simple. You get a gamepad object, setup a callback to handle the detection of a gamepad device, and bind to buttons in that callback. Then you start the gamepad polling.

function create() {

    // ... setup stuff ...

    var jumpButton = null;

    controller = game.input.gamepad.pad1;

    controller.addCallbacks(this, {
        onConnect: function() {
            // you could use a different button here if you want...
            jumpButton = controller.getButton(Phaser.Gamepad.BUTTON_1);
        }
    });

    game.input.gamepad.start();

    // ... other stuff ...

}

function update() {

    // ... other stuff ...

    if (jumpButton.isDown) {
        // jump code goes here!
    }

    // ... other stuff ...

}

Much like you do for keyboard input, you can set up the buttons you want to listen for in create and then perform actions based on their state in update. And this works pretty well - in Firefox. Chrome, on the other hand, has some issues. Phaser's example code, much like my example above, works most of the time in Chrome when the code gets executed very quickly after the page loads. But if you put enough setup code in front of your gamepad initialization you'll be wondering, like I was, why your gamepad never connects.

I had to dig into the Phaser code in order to figure this out. It works consistently in Firefox because Firefox waits until the first time a button is pressed on a gamepad before it emits a gamepadconnected event from the window object. Phaser catches that and sets everything up, calling the onConnect function when complete. In Chrome, however, gamepads just show up magically at some point after the page is loaded, in an array-like object accessed by calling navigator.getGamepads(). Phaser checks this list constantly, and when things appear for the first time, it makes all the internal setup calls. And right there's the problem! If the gamepads appear BEFORE my onConnect callback function is set up, I missed the boat. A default, no-op callback got executed instead and my gamepad buttons never get set up!

There was no work-around for this that I felt was acceptable, so I actually forked Phaser and fixed the problem in the Gamepad object's code. It was a fairly simple fix - I just don't start polling for those gamepad objects until after the call has been made to game.input.gamepad.start().

I forked off of version 2.4.7 and submitted a pull request, so hopefully my fix makes it in to the next Phaser release and the rest of you won't have to deal with this problem like I had to! If you can't wait, try using my fork and gamepad branch.
UPDATE: My pull request was merged, but not in time for 2.4.8. Look for this fix in the 2.4.9 release!
If you're just interesting in playing the game I've been working on, you can do that here: http://amphibian.com/eight-bit. The full source code is available on GitHub. If you're just interested in viewing today's comic, you can do that here:

Amphibian.com comic for 9 May 2016