A camera for LÖVE.
The first thing one needs to do to use gamera is to create it. You do so with gamera.new
. This function requires 4 numbers (left, top, width and height) defining the "world boundaries" for the camera.
local cam = gamera.new(0,0,2000,2000)
The left and top parameters are usually 0,0, but they can be anything, even negative numbers (see below).
You can update the world definition later on with setWorld
:
cam:setWorld(0,0,2000,2000)
By default gamera will use the whole screen to display graphics. You can restrict the amount of screen used with setWindow
:
cam:setWindow(0,0,800,600)
You can move the camera around by using setPosition
:
cam:setPosition(100, 200)
setPosition
takes into account the current window boundaries and world boundaries, and will keep the view inside the world. This means that if you try to look at something very close to the left border of the world, for example, the camera will not "scroll" to show empty space.
You can also zoom in and zoom out. This is done using the setScale
method. It's got a single parameter, which is used for increasing and decreasing the height and width. The default scale is 1.0.
cam:setScale(2.0) -- make everything twice as bigger. By default, scale is 1 (no change)
Take notice that gamera limits the amount of zoom out you can make; you can not "zoom out" to see the world edges. If you want to do this, make the world bigger first. For example, to give a 100-pixels border to a world defined as 0,0,2000,
, you can define it like -100,100,2100,2100
instead.
Finally, you can modify the angle with setAngle
:
cam:setAngle(newAngle) -- newAngle is in radians, by default the angle is 0
setAngle
will change both the scale and position of the camera to force you not to see the world borders. If you don't want this to happen, expand the world borders as mentioned above.
The camera has one method called "draw". It takes one function as a parameter. Here's one example:
local function drawCameraStuff(l,t,w,h)
-- draw stuff that will be affected by the camera, for example:
drawBackground(l,t,w,h)
drawTiles(l,t,w,h)
drawEntities(l,t,w,h)
end
...
-- pass your custom function to cam:draw when you want to draw stuff
cam:draw(drawCameraStuff)
Alternatively, you could create a custom anonymous function and pass it to cam:draw
, but in some extreme cases this could have a non-negligible performance impact.
cam:draw(function(l,t,w,h)
-- draw camera stuff here
end)
Anything drawn inside the custom function you pass to cam:draw
be modified by the camera (scaled, rotated, translated and cut) so that it appears as it should in the screen window.
Notice that the function takes 4 optional parameters. These parameters represent the area that the camera "sees" (same as calling cam:getVisible()
). They can be used to optimize the drawing, and not draw anything outside of those borders. Those borders are always axis-aligned. This means that when the camera is rotated, the area might include elements that are not strictly visible.
-
cam:getWorld()
returns the l,t,w,h of the world -
cam:getWindow()
returns the l,t,w,h of the screen window -
cam:getVisible()
returns the l,t,w,h of what is currently visible in the world, taking into account rotation, scale and translation. It coincides with the parameters of the callback function ingamera.draw
. It can contain more than what is necessary due to rotation. -
cam:getVisibleCorners()
returns the corners of the rotated rectangle that represent the exact region being seen by the camera, in the formx1,y1,x2,y2,x3,y3,x4,y4
-
cam:getPosition()
returns the coordinates the camera is currently "looking at", after it has been corrected so that the world boundaries are not visible, if possible. -
cam:getScale()
returns the current scaleX and scaleY parameters -
cam:getAngle()
returns the current rotation angle, in radians
cam:toWorld(x,y)
transforms screen coordinates into world coordinates, taking into account the window, scale, rotation and translation. Useful for mouse interaction, for example.cam:toScreen(x,y)
transforms given a coordinate in the world, return the real coords it has on the screen. Useul to represent icons in minimaps, for example.
Everything looks kindof "blurry" when I do zooms/rotations with this library. How do I prevent it?
LÖVE uses a default filter mode which makes images "blurry" when you make almost any transformation to them. To deactivate this behavior globally, you can add this at the beginning of your game, before you load anything (for example at the beginning of love.load
):
love.graphics.setDefaultFilter( 'nearest', 'nearest' )
You can also set the filter of each image you load individually:
local img = love.graphics.newImage('imgs/player.png')
img:setFilter('nearest', 'nearest')
I see "seams" around my tiles when I use this library. Why?
It is due to a combination of factors: how OpenGL textures work, how floating point numbers work and how LÖVE stores texture information in memory.
The end result is that sometimes, when drawing an image, "parts of the area around it" are also drawn. So if you draw a Quad of "earth", and immediately above it in your image you have a "bright lava" tile, when you draw the earth tile sometimes "a bit of the lava" is drawn near the top. If you are using images instead of quads, you can get seams too (either black or from other colors, depending on how the image is stored in memory).
To prevent this:
- Always use quads for tiles, never images.
- Add a 2-pixel border to each of your quads (So for a 32x32 quad, you use 36x36 space, with the 32x32 quad in the center and a 2-pixel border)
- Fill up the borders with the colors of the 32x32 quad. For example, if the earth quad is all brown on its left side, color the left border brown. If on its upper side is gray in the center and brown on the sides, color the upper border gray in the center and brown on the sides.
- The quads will still be 32x32 - they will just have some border in the image now.
- The "seams" will still appear, but now they will show the "colored borders" of the quads, so they will not be noticeable.
- Calculating the coordinates of the quads is a bit more complex than before. You can use anim8's "Grids" to simplify getting them:
local anim8 = require 'anim8'
...
local tiles = love.graphics.newImage('tiles.png')
local w,h = tiles:getDimensions()
local g = anim8.newGrid(32, 32, w, h, 0, 0, 2) -- tileWidth, tileHeight, imageW, imageH, left, top, border
local earth = g(1,1)[1] -- Get the first column, first row 32x32 quad, including border
local water = g(2,1)[1] -- Get the second column, first row quad
You can combine this with the previous FAQ and use a "nearest" filter instead of a linear one.
The camera is "clamping". I don't want it to clamp
Yes, by default gamera
cameras make sure that they always "stay in the world". They never "show black borders" around the scene.
There is no way to deactivate this behaviour. However, you can "mimic" it very well. It's actually very simple: give the camera a bigger world.
For example, if instead of doing this:
local cam = gamera.new(0,0,2000,2000)
You do this:
local cam = gamera.new(-200,-200,2200,2200)
You will give the world a 200 pixel "black border" which will be visible when the camera zooms out or moves to the left or right.
If you want to display the borders only sometimes, you can use setWorld
to activate/deactivate the zoom:
local cam = gamera.new(0,0,2000,2000)
... -- the camera "clamps" when drawing
cam:setWorld(-200,-200,2200,2200)
... -- Now the camera has a 200px border
cam:setWorld(0,0,2000,2000)
... -- now it clamps again
Just copy the gamera.lua file wherever you want it. Then require it where you need it:
local gamera = require 'gamera'
Please make sure that you read the license, too (for your convenience it's included at the beginning of the gamera.lua file).