The problem: I'm trying to create a simple drawing app using p5.js. Instead of the standard cursor image, I'd like to show a circle at my cursor location that represents the size of the drawing brush.
Potential solution 1: Replace the cursor using the cursor() function native to p5.
Why it doesn't work: The p5 cursor function only takes the following parameters:
ARROW, CROSS, HAND, MOVE, TEXT, or WAIT, or path for image
As such, there's no native way to replace the cursor using the ellipse class.
Potential solution 2: Use the noCursor() function and then draw the circle at the cursor location, while also drawing the background, as such:
var brushSize = 50;
function setup() {
createCanvas(1080,720);
noCursor();
}
function draw() {
background(100);
ellipse(mouseX,mouseY,brushSize);
}
Why it doesn't work: While this solution gets the desired effect i.e. replacing the cursor with a circle the size of the brush, the constantly updating background prevents me from actually drawing to the canvas with the brush when I want to.
Is there some way I can replace the cursor without actually drawing the ellipse to the canvas? Is there any way to save and then instantly reload a canvas in p5? I couldn't find such a method searching through the API docs. Any hints are appreciated.
According to the reference, you can pass a URL into the cursor() function to set an image.
If you want to use an image that you draw, you're going to have to draw them ahead of time and save them to files, and then use those files. Something like this:
cursor('images/ellipse-15.png');
Where ellipse-15.png is an image that you generated ahead of time, to match when brushSize is 15, for example.
Btw P5.js is just setting the cursor CSS property. You can read more about it here.
If you want to go with the noCursor() approach and draw the ellipse yourself, you could draw your drawing to a buffer (the createGraphics() function is your friend) and then draw the ellipse on top of that every frame. I'd still probably use a cross cursor just because there's going to be some annoying lag if you draw it yourself.
Create a circular DIV inside the canvas container and show it on top of the actual canvas.
I have two canvases. I have made them circular using border-radius. The 2nd is positioned inside the first one (using absolute position).
I have click events on both circles. If you click on inside canvas, the color at the point of the click is loaded in the outside canvas with opacity varying from white to the picked color and finally to black. If you click on outer canvas the exact color value at that point is loaded in the text-box at the bottom
I am unable to click in red zones (as shown in figure below) of the outer canvas when using chrome. I tried z-idex, arcs but nothing is helping me. But In Firefox everything is working fine.
Note: You can drag the picker object in the outer circle. But if you leave it in red zones, you would not be able to click it again in Chrome. Clicking in green zone will get you its control again
Code in this JSFiddle
Edit
I excluded all irrelevant code to make it easy. Now there is only a container having two canvas.
Filled simply with two distinct colors. Open following fiddle link in both chrome and firefox. Click on both cirles in different zones and see difference in chrome and firefox. I want them to behave in chrome as they do in firefox
Note I will ultimately draw an image in inner canvas.
Updated Fiddle Link
-
Your problem is because canvases currently are always rectangular, even if they don't look rectangular. Border radius makes the edges except the circle transparent, but it still doesn't stop events in Chrome on the corner areas. This is why you cannot click the bottom circle in those areas
I even tried putting it inside of a container that had a border-radius instead but the click event still goes through
With that being said, you have two options. You could either change your code to only use one canvas with the same type of layout, just drawing the background circle before the other each time. Essentially you'd draw a circle, draw your black to color to white gradient, use the xor operation to combine the two into one circle, then do the same with the rainbox gradient. You must draw the background circle first because canvas paints over the old layers every time
or
You could use javascript to only detect clicks in the circular area which takes just a little bit of math (: This solution is featured in edit below
In the future, CSS Shapes may allow canvases to be non-rectangular elements to be used, I'm actually not sure, but we don't have that capability yet at least
Edit
Alright, so after going through your code a bit it seems there are some things I should cover before I offer a solution
Setup all your finite variables outside of the functions that run every time. This means you don't put them (like radiuses, offsets, etc.) in the click function or something that runs often since they don't change
Your "radius"es are actually "diameter"s. The format of .rect goes .rect(x, y, width (diameter of circle), height (diameter of circle))
Almost always when overlaying canvases like you are you want to make them equal dimensions and starting position to prevent calculation error. In the end it makes it easier, doing all relative positioning with javascript instead of mixing it with CSS. In this case, however, since you're using border-radius instead of arc to make a circle, keep it like it is but position it using javascript ....
jQuery isn't needed for something this simple. If you're worried about any load speed I'd recommend doing it in vanilla javascript, essentially just changing the .click() functions into .onclick functions, but I left jQuery for now
You can declare multiple variables in a row without declaring var each time by using the following format:
var name1 = value1,
name2 = value2;
Variables with the same value you can declare like so:
var name1 = name2 = sameValue;
When children have position:absolute and you want it to be positioned relative to the parent, the parent can have position:relative, position:fixed, or position:absolute. I would think you'd want position:relative in this case
When you don't declare var for a variable it becomes global (unlessed chained with a comma like above). For more on that read this question
Now, onto the solution.
After talking with a friend I realized I could sort do the math calculation a lot easier than I originally thought. We can just calculate the center of the circles and use their radiuses and some if statements to make sure the clicks are in the bounds.
Here's the demo
After everything is set up correctly, you can use the following to detect whether or not it's in the bounds of each
function clickHandler(e, r) {
var ex = e.pageX,
ey = e.pageY,
// Distance from click to center
l = Math.sqrt(Math.pow(cx - ex, 2) + Math.pow(cy - ey, 2));
if(l > r) { // If the distance is greater than the radius
if(r === LARGE_RADIUS) { // Outside of the large
// Do nothing
} else { // The corner area you were having a problem with
clickHandler(e, LARGE_RADIUS);
}
} else {
if(r === LARGE_RADIUS) { // Inside the large cirle
alert('Outer canvas clicked x:' + ex + ',y:' + ey);
} else { // Inside the small circle
alert('Inner canvas clicked x:' + ex + ',y:' + ey);
}
}
}
// Just call the function with the appropriate radius on click
$(img_canvas).click(function(e) { clickHandler(e, SMALL_RADIUS); });
$(wheel_canvas).click(function(e) { clickHandler(e, LARGE_RADIUS); });
Hopefully the comments above and code make enough sense, I tried to clean it up as best as I could. If you have any questions don't hesitate to ask!
The basis for the code is from John E. Graham's blog http://johnegraham2.com/blog/2010/09/25/project-javascript-2d-tile-engine-with-html5-canvas-part-4-using-zones-for-further-optimization/
It works perfectly for drawing a screen's worth of tiles, but I cannot for the life of me figure out how to adjust it 1 row/column at a time based on pressing up, down, left, or right keys.
Here is an example with the transparency to help visualize the zones http://simplehotkey.com/Javascript/canvas.html (loading positions of 1,188 tiles but only draws a couple hundred to fill the screen) I had it loading an array with 70,000 entries and it was still quick because it's only drawing whats on the screen, but cannot figure out how to slide everything based on input...
I've come up with a couple ideas and am not sure what's the best way.
One screen worth of tiles is shown here:
tilesArray = [
0,0,0,0,0,0,0,1,2,9,6,0,0,7,0,0,1,0,
0,9,9,9,9,9,9,9,9,9,9,9,9,9,9,9,9,0,
0,9,9,9,9,9,9,9,9,9,9,9,9,9,9,0,9,0,
0,9,9,9,9,9,9,9,9,9,9,9,9,9,9,0,9,0,
0,9,9,9,9,9,9,0,7,2,0,0,0,0,0,1,2,0,
0,9,9,9,9,9,9,9,9,9,9,9,9,9,9,9,9,0,
0,9,9,9,9,9,9,9,9,9,9,9,9,9,9,9,9,0,
0,9,0,7,2,9,9,9,9,9,9,9,9,9,9,9,9,0,
0,9,9,9,9,9,9,9,9,9,9,9,9,9,9,9,9,0,
0,9,9,9,9,9,9,9,9,9,9,9,9,9,9,9,9,0,
0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
];
Where 0 is a wall tile (around perimeter), 9 a floor tile, 7 a door and a couple other random tiles.
That is exactly what is loaded to the screen, but I cannot figure out how to shift everything 1 tile in either direction based on input, up, down, left, right.
The one idea I'm leaning towards now, is just to use that array above as the basis for rendering, and somehow feeding the new values into it based on keyboard input, from another array. Maybe slicing from another, much larger array (holding all the tiles for the entire level) and using that slice to populate the array that's actually rendered???
That's replacing every tile every frame though...
for getting player input I was using:
//Key listener
document.onkeydown = function(e){
e = e?e:window.event;
console.log(e.keyCode + "Down");
switch (e.keyCode){
case 38:
//UP KEY
Game.inputReaction('up');
//Game.moveDir('up');
break;
case 40:
//DOWN KEY
//Game.inputReaction(40);
//Game.moveDir('down');
break;
case 37:
//Left Key
//Game.inputReaction(37);
break;
}
}
The other alternative is to try to adjust the tiles already on the screen and add new tiles but this engine isn't using global variables so I'm not sure how to affect the tile engine programatically based on input....like I can add another method (inputReaction(num)) and trigger some actions from my keyboard input (console.log()) but I can't access the other methods actually drawing the tiles. Or maybe I have to make a copy of the object, change it and return it? but it's pretty complex.
I think it might be easier to adjust the array values that are being fed into the "engine" (array above) rather than changing around how the engine is calculating what's being drawn. Can you confirm this?
Add a camera abstraction that you can move around on the map, then shift the drawing positions according to the camera position. When the camera moves south 10px, all tiles move north 10px, same with east and west. Since you only draw the tiles that are visible, there won't be much of a performance loss.
The renderer looks at the camera to figure out what needs to be drawn and you can expose the camera object to the outside to manipulate it. That way you only need to change the camera position to change what is shown on the screen.
I did this in a proof of concept tiling engine a year ago and I was able to smoothly scroll and scale huge tilemaps.
If you start changing the array itself, your performance will suffer and you won't be able to scroll smoothly since you can only go in steps of one tile and not one pixel.