Solving Resolution Independent Rendering And 2D Camera Using Monogame

As i promised in my previous post where i announced my Windows 8 game Marbles, i will try to tackle some of the common problems that beginner game developers encounter.

In this post i will jump ahead a little and talk about Resolution Independent Rendering using Monogame.

What the heck is Resolution Independent Rendering?

Well its fancy name for finding ways to not care about resolution during your game development.
Idea is that you render always using fixed (internal) virtual resolution you choose and then simply stretch or shrink all that for the real resolution of the current device you are using to display the game.
Sounds simple right?

Why solving Resolution Independent Rendering?

Because of multiple reasons:

• First of all its very important and difficult subject. You don’t want to care about resolution on which you are displaying your game. Game developer has enough problems to solve even without this.
• Every game developer that is targeting multiple devices/platforms has to solve this in one way or another
• I struggled until i found a solution for this and want to share it with the community
• many people lately asked me to share code for this and i kept sending them same email with zipped source code for this so it seems like a good idea for blog post 🙂

How are we going to solve it?

Firstly, we will decide on the internal resolution for our game, lets say 1366×768.

Then we will determine what is the real screen resolution of the device/monitor where we are rendering our game.

Then we will determine what is the best fit rectangle that we can get on this real screen that matches our virtual resolution (but always maintaining aspect ratio of the Virtual Resolution we decided so we don’t have stretching/shrinking artefacts).

For example if real screen resolution is bigger then our internal virtual resolution, we will create a best fit ‘viewport’ inside of that larger resolution by maintaining aspect ratio but stretching our viewport – therefore our game screens will be stretched and rendered inside this bigger viewport on the screen.

For smaller screen we will do the opposite – shrink the viewport and maintain aspect ratio of our virtual resolution – therefore our game gfx will first shrink to fit into the smaller viewport and then be rendered on the screen.

Since the viewport we decide on will often not fill the real screen around it we will fill it with some color – effect known in film industry as Letterboxing.

Once we know that centered viewport rectangle where we will render our stretched/shrinked screen we then create transformation matrix that we will use in all the rendering calls (by passing it to the SpriteBatch.Begin method).

Also we need to set the Mogame/XNA ViewPort property to that same viewport to limit rendering only to this region on the real screen.

To be honest, I’m not the one who invented this technique, first time i saw someone using GraphicDevice Viewport for this is David Amadors blog post on same subject where he shares example on how to achieve resolution independence with XNA.

But this was not enough for me i wanted to have also a Camera on top of that so that i can rotate zoom etc.

So i included camera that builds up on this resolution transformation matrix of the Resolution Renderer and creates its own view matrix (rotated, zoomed, translated etc).

Where is the code???

In the sample visual studio 2012 solution  you can see this in action where we are rendering a fixed image of 1366×768 to any screen you try it on, stretching and shrinking it accordingly without any changes in the code. Also you can use keyboard + and – to control camera zoom and Shift + and Shift – to rotate it.

And the screenshots?

Here are the screenshots of the sample code running on 3 different screens/devices:

1680×1050 monitor: image is stretched with little orange bar above

1280×1024: image is shrinked and we have strong pillarboxing effect

1024×768 – image is shrinked with some pillarboxing

Note: this orange color is just for the demo, code allows you to easily set what color will the Pillarbox be.
In the sample Visual Studio 2012 solution for Monogame 3.0.1 you will want to check out two important classes:

1. ResolutionIndependentRenderer – this class handles everything about resolutions and creates viewport transformation matrix for resolution independence

2. Camera2D – this class usess ResolutionIndependentRenderer and its matrix and creates final camera view matrix you should use for SpriteBatch and allows you to set camera position, zoom, rotate etc

Here is the code for ResolutionIndependentRenderer:

using Microsoft.Xna.Framework;
using Microsoft.Xna.Framework.Graphics;

namespace Roboblob.XNA.WinRT.ResolutionIndependence
{
    public class ResolutionIndependentRenderer
    {
        private readonly Game _game;
        private Viewport _viewport;
        private float _ratioX;
        private float _ratioY;
        private Vector2 _virtualMousePosition = new Vector2();

        public Color BackgroundColor = Color.Orange;

        public ResolutionIndependentRenderer(Game game)
        {
            _game = game;
            VirtualWidth = 1366;
            VirtualHeight = 768;

            ScreenWidth = 1024;
            ScreenHeight = 768;
        }

        public int VirtualHeight;

        public int VirtualWidth;

        public int ScreenWidth;
        public int ScreenHeight;

        public void Initialize()
        {
            SetupVirtualScreenViewport();

            _ratioX = (float)_viewport.Width / VirtualWidth;
            _ratioY = (float)_viewport.Height / VirtualHeight;

            _dirtyMatrix = true;
        }

        public void SetupFullViewport()
        {
            var vp = new Viewport();
            vp.X = vp.Y = 0;
            vp.Width = ScreenWidth;
            vp.Height = ScreenHeight;
            _game.GraphicsDevice.Viewport = vp;
            _dirtyMatrix = true;
        }

        public void BeginDraw()
        {
            // Start by reseting viewport to (0,0,1,1)
            SetupFullViewport();
            // Clear to Black
            _game.GraphicsDevice.Clear(BackgroundColor);
            // Calculate Proper Viewport according to Aspect Ratio
            SetupVirtualScreenViewport();
            // and clear that
            // This way we are gonna have black bars if aspect ratio requires it and
            // the clear color on the rest
        }

        public bool RenderingToScreenIsFinished;
        private static Matrix _scaleMatrix;
        private bool _dirtyMatrix = true;

        public Matrix GetTransformationMatrix()
        {
            if (_dirtyMatrix)
                RecreateScaleMatrix();

            return _scaleMatrix;
        }

        private void RecreateScaleMatrix()
        {
            Matrix.CreateScale((float)ScreenWidth / VirtualWidth, (float)ScreenWidth / VirtualWidth, 1f, out _scaleMatrix);
            _dirtyMatrix = false;
        }

        public Vector2 ScaleMouseToScreenCoordinates(Vector2 screenPosition)
        {
            var realX = screenPosition.X - _viewport.X;
            var realY = screenPosition.Y - _viewport.Y;

            _virtualMousePosition.X = realX / _ratioX;
            _virtualMousePosition.Y = realY / _ratioY;

            return _virtualMousePosition;
        }

        public void SetupVirtualScreenViewport()
        {
            var targetAspectRatio = VirtualWidth / (float) VirtualHeight;
            // figure out the largest area that fits in this resolution at the desired aspect ratio
            var width = ScreenWidth;
            var height = (int)(width / targetAspectRatio + .5f);

            if (height > ScreenHeight)
            {
                height = ScreenHeight;
                // PillarBox
                width = (int)(height * targetAspectRatio + .5f);
            }

            // set up the new viewport centered in the backbuffer
            _viewport = new Viewport
                            {
                                X = (ScreenWidth / 2) - (width / 2),
                                Y = (ScreenHeight / 2) - (height / 2),
                                Width = width,
                                Height = height
                            };

            _game.GraphicsDevice.Viewport = _viewport;
        }
    }
}

And then there is Camera2D that has all the usual camera properties that you can use to move it around and zoom rotate etc.

Here is the code for camera:

using Microsoft.Xna.Framework;
using Roboblob.XNA.WinRT.ResolutionIndependence;

namespace Roboblob.XNA.WinRT.Camera
{
    public class Camera2D
    {
        private float _zoom;
        private float _rotation;
        private Vector2 _position;
        private Matrix _transform = Matrix.Identity;
        private bool _isViewTransformationDirty = true;
        private Matrix _camTranslationMatrix = Matrix.Identity;
        private Matrix _camRotationMatrix = Matrix.Identity;
        private Matrix _camScaleMatrix = Matrix.Identity;
        private Matrix _resTranslationMatrix = Matrix.Identity;

        protected ResolutionIndependentRenderer ResolutionIndependentRenderer;
        private Vector3 _camTranslationVector = Vector3.Zero;
        private Vector3 _camScaleVector = Vector3.Zero;
        private Vector3 _resTranslationVector = Vector3.Zero;

        public Camera2D(ResolutionIndependentRenderer resolutionIndependence)
        {
            ResolutionIndependentRenderer = resolutionIndependence;

            _zoom = 0.1f;
            _rotation = 0.0f;
            _position = Vector2.Zero;
        }

        public Vector2 Position
        {
            get { return _position; }
            set
            {
                _position = value;
                _isViewTransformationDirty = true;
            }
        }

        public void Move(Vector2 amount)
        {
            Position += amount;
        }

        public void SetPosition(Vector2 position)
        {
            Position = position;
        }

        public float Zoom
        {
            get { return _zoom; }
            set
            {
                _zoom = value;
                if (_zoom < 0.1f)
                {
                    _zoom = 0.1f;
                }
                _isViewTransformationDirty = true;
            }
        }

        public float Rotation
        {
            get
            {
                return _rotation;
            }
            set
            {
                _rotation = value;
                _isViewTransformationDirty = true;
            }
        }

        public Matrix GetViewTransformationMatrix()
        {
            if (_isViewTransformationDirty)
            {
                _camTranslationVector.X = -_position.X;
                _camTranslationVector.Y = -_position.Y;

                Matrix.CreateTranslation(ref _camTranslationVector, out _camTranslationMatrix);
                Matrix.CreateRotationZ(_rotation, out _camRotationMatrix);

                _camScaleVector.X = _zoom;
                _camScaleVector.Y = _zoom;
                _camScaleVector.Z = 1;

                Matrix.CreateScale(ref _camScaleVector, out _camScaleMatrix);

                _resTranslationVector.X = ResolutionIndependentRenderer.VirtualWidth*0.5f;
                _resTranslationVector.Y = ResolutionIndependentRenderer.VirtualHeight * 0.5f;
                _resTranslationVector.Z = 0;

                Matrix.CreateTranslation(ref _resTranslationVector, out _resTranslationMatrix);

                _transform = _camTranslationMatrix *
                             _camRotationMatrix *
                             _camScaleMatrix *
                             _resTranslationMatrix *
                             ResolutionIndependentRenderer.GetTransformationMatrix();

                _isViewTransformationDirty = false;
            }

            return _transform;
        }

        public void RecalculateTransformationMatrices()
        {
            _isViewTransformationDirty = true;
        }
    }
}

How to use all this???

Its simple 🙂

In your Game create instance of ResolutionIndependentRenderer and SimpleCamera2D.

Camera accepts the instance of ResolutionIndependentRenderer in the constructor cause it uses its resolution Matrix to calculate its own view Matrix.

On start of your game (or whenever your real screen resolution changes) – user rotates the device, or moves the game to another monitor on desktop PC etc. – you must initialize the

ResolutionIndependentRenderer like this and also invalidate the camera matrix:

        private void InitializeResolutionIndependence(int realScreenWidth, int realScreenHeight)
        {
            _resolutionIndependence.VirtualWidth = 1366;
            _resolutionIndependence.VirtualHeight = 768;
            _resolutionIndependence.ScreenWidth = realScreenWidth;
            _resolutionIndependence.ScreenHeight = realScreenHeight;
            _resolutionIndependence.Initialize();

            _camera.RecalculateTransformationMatrices();
        }

As you see above we are passing to the ResolutionIndependentRenderer virtual resolution we want to use internally and the real resolution of the screen, and then we recalculate matrices in the camera. From that point we can use the same cached camera matrix until we change position or zoom etc (this is handled internally in the camera code).

Here exceptionally we manually have to notify the Camera class that we changed the ResolutionIndependentRenderer matrix so it can recalculate its own matrix.

Next stop is our Draw method of the game:

        protected override void Draw(GameTime gameTime)
        {
            _resolutionIndependence.BeginDraw();
            _spriteBatch.Begin(SpriteSortMode.Deferred, BlendState.AlphaBlend, SamplerState.LinearWrap, DepthStencilState.None, RasterizerState.CullNone, null, _camera.GetViewTransformationMatrix());
            _spriteBatch.Draw(_bkg, _bkgPos, Color.White);
            _spriteBatch.DrawString(_debugFont, string.Format("Translated Mouse Pos: x:{0:0}  y:{1:0}",_screenMousePos.X,_screenMousePos.Y), _mouseDrawPos, Color.Yellow);
            _spriteBatch.DrawString(_debugFont, _instructions, _instructionsDrawPos, Color.Yellow);
            _spriteBatch.End();

            // TODO: Add your drawing code here

            base.Draw(gameTime);
        }

Calling the BeginDraw method on ResolutionIndependentRenderer internally first sets the viewport to full screen, clear it with color that is set in its Background property and then sets the virtual resolution viewport to our virtual resolution so whatever we draw its drawn in the correct centered rectangle.

That way we get that color fill around our game.

Another thing to notice there is that we are passing a camera view matrix to the SpriteBatch.Begin call:

            _spriteBatch.Begin(SpriteSortMode.Deferred, BlendState.AlphaBlend, SamplerState.LinearWrap, DepthStencilState.None, RasterizerState.CullNone,
                null, _camera.GetViewTransformationMatrix());

This is very important cause if you don’t do this, you will still be drawing to the full screen and we don’t want that.
By passing the Matrix from our camera to the SpriteBatch we tell it to rotate, zoom and translate all our draws to the correct place

Benefit of all this is that we can setup camera and zoom and rotate our screen and move camera around.
At start of the sample code i center the camera to the middle of the virtual screen and set zoom to 1.0 but you can change/animate those values and have some nice effects.

Here is how the camera initialization code looks like:

            _camera = new Camera2D(_resolutionIndependence);
            _camera.Zoom = 1f;
            _camera.Position = new Vector2(_resolutionIndependence.VirtualWidth / 2, _resolutionIndependence.VirtualHeight / 2);

Download the sample visual studio 2012 solution and play with it.

Try to run it on multiple screen resolutions, it should always adapt as best possible and maintain aspect ratio of virtual resolution when stretching/shrinking it to the real screen.

Another cool feature of the ResolutionIndependentRenderer is the ability to convert our real screen mouse position to the virtual mouse position:

            _screenMousePos = _resolutionIndependence.ScaleMouseToScreenCoordinates(_inputHelper.MousePosition);

This is needed because our mouse events are happening in the real screen resolution and we need to convert (translate) them to the correct position in our virtual resolution space which is usually shifted (unless your screen is same size as your virtual resolution) so this method allows you to do that easily.

I hope this code will be of help to all the aspiring XNA/Monogame developers out there!

Check out the sample and let me know if it helps you!