Furthermore, individual primitives (polygons) are also rendered in the order they are stored in the indice buffer:
Now if you were making a 2D game with 1000 zombie sprites onscreen you would undoubtedly want to use 3D-in-2D rendering with an orthographic camera. Batching and depth discard would give you much faster performance when the number of objects goes up. However, the 2D aspect of most games is relatively simple, with only a dozen or so 2D sprites making up the user interface. Given that 2D graphics are not normally going to be much of a bottleneck, and that the biggest performance savings we have achieved was in making text a static object, I decided to rework the 2D rendering system into something that was a little simpler to use.
Sprites are no longer a 3D entity, but are a new type of pure 2D object. They act in a similar way as entities with position, rotation, and scale commands, but they only use 2D coordinates:
//Create a sprite auto sprite = CreateSprite(world,100,100); //Make blue sprite->SetColor(0,0,1); //Position in upper-left corner of screen sprite->SetPosition(10,10) Sprites have a handle you can set. By default this is in the upper-left corner of the sprite, but you can change it to recenter them. Sprites can also be rotated around the Z axis:
//Center the handle sprite->SetHandle(0.5,0.5); //Rotation around center sprite->SetRotation(45); SVG vector images are great for 2D drawing and GUIs because they can scale for different display resolutions. We support these as well, with an optional scale value the image can be rasterized at.
auto sprite = LoadSprite(world, "tiger.svg", 0, 2.0);
Text is now just another type of sprite:
auto text = CreateSprite(world, font, L"Hello, how are you today?\nI am fine.", 72, TEXT_LEFT); These sprites are all displayed within the same world as the 3D rendering, so unlike what I previously wrote about...
You do not have to create extra cameras or worlds just to draw 2D graphics. (If you are doing something advanced then the multi-camera method I previously described is a good option, but you have to have very demanding needs for it to make a difference.) Regular old screen coordinates you are used to will be used (coordinate [0,0] is top-left). By default sprites will be drawn in the order they are created. However, I definitely see a need for additional control here and I am open to ideas. Should there be a sprite order value, a MoveToFront() method, or a system of different layers? I'm not sure yet.
I'm also not sure how per-camera sprites will be controlled. At this time sprites are stored in a per-world list, but we will want some 2D elements to only appear on some cameras. I am not sure yet how this will be controlled.
I am going to try to get an update out soon with these features so you can try them out yourself.