Skip to main content

Gift Wrapping Convex Hull Algorithm With Unity Implementation


Convex Hull Algorithm

Convex Hull algorithms are one of those algorithms that keep popping up from time to time in seemingly unrelated fields from big data to image processing to collision detection in physics engines, It seems to be all over the place. Why should you care? Cus you can do magic with it and it seems so simple to implement when you first hear about it, but when you start thinking about it, you will realize why it's not such a straightforward thing to do.
Now that I got you interested (hopefully) and now we will see just what a convex hull is.
a set of points enclosed by a convex hull
As you may have noticed a perimeter was made with the same points that was given and these perimeter points enclose the entire set of points.
Now we have to clear up the term 'Convex'.
Convex means no part of the object is caved inwards or that none of the internal angles made by the points exceed 180 degrees.
An example of concave shape

In this example of a concave shape internal angles go beyond 180 degrees.
What are those red lines for? Well... concave shapes have a feature that can easily differentiate it from a convex one, in  the case of a  concave shape at least one line from one point to another will travel outside the shape or a line made by any two points will intersect with another set of lines making up the shape.
Those conditions don't apply to a convex shape.
An example of convex shape



All the points can connect with any other point without intersecting any other line in the shape.
The 2D implementation of the Gift Wrapping algorithm is called 'Jarvis March'.
The image above describes how the algorithm goes about creating the convex hull. We will look at some pseudo code( based on the one given in Wikipedia )
CreateConvexHull(SetOfPoints)  
   pointOnHull = leftmost point in SetOfPoints
   i = 0  
   repeat  
      P[i] = pointOnHull  
      endpoint = SetOfPoints[0] // initial endpoint for a candidate edge on the hull  
      for j from 1 to |SetOfPoints|  
        if (endpoint == pointOnHull) or (SetOfPoints[j] is on left of line from P[i] to endpoint)  
        endpoint = SetOfPoints[j]  // found greater left turn, update endpoint  
      i = i+1  
      pointOnHull = endpoint  
   until endpoint == P[0]   // wrapped around to first hull point  
You may have noticed that there are sections in the pseudo code regarding finding or checking if a point lies on the left.We will take a look at those instances.
pointOnHull = leftmost point in SetOfPoints
This line says that we should find the left most point, so if we are using the X-Y co-ordinate plane the left most point will be the point with the smallest X co-ordinate.
if (endpoint == pointOnHull) or (SetOfPoints[j] is on left of line from P[i] to endpoint)  
We are checking if the point 'SetOfPoints[j]' is on the left of the line from 'P[i]' to 'endpoint'.
How do we checking if a point is on the left or right side of a line?
Here's how:
(Bx - Ax) * (Cy - Ay) - (By - Ay) * (Cx - Ax) 
'A' and 'B' are the points making up the line and C is the point we are checking whether it's to the right or left or if it's a point lying along line (collinear).
If we get the result as 0, then the point is collinear.
If we get the result as +ve, then the point is to the left
If we get the result as -ve, then the point is to the right.
This is end result we will be getting.
We are going to be making this so that I can be run in the editor, so we are going to be making use of Gizmos (debugging tool in unity) to draw the spheres and lines.
So now create a class called JarvisMarchConvexHull( very creative name 👏).
These are the member variables.
public Transform[] points;  
[Range(0.05f,1.5f)]  
public float size;  
public bool drawIt;  
private HashSet<Transform> result;  
'points' - the transforms that are going to be passed in from the inspector.
'size' - the size if the sphere to be drawn
'drawIt' - whether or not I want to display the spheres and lines.
'result' - this is a HashSet which means only unique items can be added to the list, It's useful when we have to avoid duplication or when we are trying to sound smart about data structures 😛.
*Warning Big Wall Of Text Incoming, Don't worry, we are breaking everything down into simple chunks below.
Let's looks at the Update function.
void Update()  
{  
     result = new HashSet<Transform>();  
     int leftMostIndex = 0;  
     for (int i = 1; i < points.Length; i++)  
     {  
       if (points[leftMostIndex].position.x > points[i].position.x)  
         leftMostIndex = i;  
     }  
     result.Add(points[leftMostIndex]);  
     List<Transform> collinearPoints = new List<Transform>();  
     Transform current = points[leftMostIndex];  
     while (true)  
     {  
       Transform nextTarget = points[0];  
       for (int i = 1; i < points.Length; i++)  
       {  
         if (points[i] == current)  
           continue;  
         float x1, x2, y1, y2;  
         x1 = current.position.x - nextTarget.position.x;  
         x2 = current.position.x - points[i].position.x;  
         y1 = current.position.y - nextTarget.position.y;  
         y2 = current.position.y - points[i].position.y;  
         float val = (y2 * x1) - (y1 * x2);  
         if (val > 0)  
         {  
           nextTarget = points[i];  
           collinearPoints = new List<Transform>();  
         }  
         else if (val == 0)  
         {  
           if (Vector2.Distance(current.position, nextTarget.position) < Vector2.Distance(current.position, points[i].position))  
           {  
             collinearPoints.Add(nextTarget);  
             nextTarget = points[i];  
           }  
           else  
             collinearPoints.Add(points[i]);            
         }  
       }  
       foreach (Transform t in collinearPoints)  
         result.Add(t);        
       if (nextTarget == points[leftMostIndex])  
         break;  
       result.Add(nextTarget);  
       current = nextTarget;  
     }  
}
We will go through all the important sections one by one.

int leftMostIndex = 0;  
for (int i = 1; i < points.Length; i++)  
  {  
    if (points[leftMostIndex].position.x > points[i].position.x)  
       leftMostIndex = i;  
  }  
result.Add(points[leftMostIndex]);  
Here we are finding the left most point and adding it to the result which is the HashSet that holds all the points of the convex hull.
List<Transform> collinearPoints = new List<Transform>();  
Transform current = points[leftMostIndex];
We need list of points that are collinear to the line we are checking currently and then do some stuff with them - remember this, we will come to this later.
And we are creating a new variable called 'current' which will be used in the loop to signify which point we are on right now. We assign current to the left most point in the beginning.
Now the loop.
while (true)  
{  
    Transform nextTarget = points[0];  
    for (int i = 1; i < points.Length; i++)  
    {  
       if (points[i] == current)  
          continue;  
       float x1, x2, y1, y2;  
       x1 = current.position.x - nextTarget.position.x;  
       x2 = current.position.x - points[i].position.x;  
       y1 = current.position.y - nextTarget.position.y;  
       y2 = current.position.y - points[i].position.y;  
       float val = (y2 * x1) - (y1 * x2);  
       if (val > 0)  
       {  
          nextTarget = points[i];  
          collinearPoints = new List<Transform>();  
       }  
       else if (val == 0)  
       {  
          if (Vector2.Distance(current.position, nextTarget.position) < Vector2.Distance(current.position, points[i].position))  
          {  
             collinearPoints.Add(nextTarget);  
             nextTarget = points[i];  
          }  
          else  
             collinearPoints.Add(points[i]);            
       }  
 }  
Time to break down the loop.
Transform nextTarget = points[0];  
This the point that keeps going around all the other points other than the current point.
The pair of 'current' and 'nextTarget' points make up the line that we use to check if a point is on the left or not.
float x1, x2, y1, y2;  
x1 = current.position.x - nextTarget.position.x;  
x2 = current.position.x - points[i].position.x;  
y1 = current.position.y - nextTarget.position.y;  
y2 = current.position.y - points[i].position.y;  
float val = (y2 * x1) - (y1 * x2);  
This is basically checking if the point ( 'points[i]' ) is to the left of line made by 'current'  & 'nextTarget'.
if (val > 0)  
{  
    nextTarget = points[i];  
    collinearPoints = new List<Transform>();  
}  
else if (val == 0)  
{  
   if (Vector2.Distance(current.position, nextTarget.position) < Vector2.Distance(current.position, points[i].position))  
   {  
       collinearPoints.Add(nextTarget);  
       nextTarget = points[i];  
   }  
   else  
       collinearPoints.Add(points[i]);            
}
If  'val' > 0, then it means the point is on the left and we update 'nextTarget' to be 'point[i]' as 'point[i]' was lying to the left of the line.
We also clear up the 'collinearPoints' list. You will understand why as we go on.
When 'val' is equal to 0 that means 'point[i]' was lying along the line made by 'current' & 'nextTarget'.
The point that we add to 'collinearPoints' must be which ever point is closer to the 'current' point ( either 'nextTarget' or 'points[i]').
foreach (Transform t in collinearPoints)  
       result.Add(t);        
if (nextTarget == points[leftMostIndex])  
       break;  
result.Add(nextTarget);  
current = nextTarget; 
You might have been wondering what we were gonna do with those 'collinearPoints', well now you know.
We add all of them to the result.
The reason we have to do this is because a line made by 'current' & 'nextTarget' might get mutiple set of points that are lying along the line ( line extends to infinity ), we should only take in the collinear point that is closest to 'current' point.
if (nextTarget == points[leftMostIndex])  
       break;  
If we have completely checked the entire set of points we will reach back to the left most point that we started which, so we exit out of the loop.
result.Add(nextTarget);  
current = nextTarget; 
If we haven't exited the loop then we add 'nextTarget' to the result as well as make 'current point' as next target, so that checking for leftmost point will start from the point the was recently added to the result.
We are essentially done here, you can stick around for the OnDrawGizmos stuff though.
void OnDrawGizmos()  
{  
   if (drawIt)  
   {  
      if (result != null)  
      {  
         List<Transform> outter = new List<Transform>();  
         foreach (var item in result)  
           outter.Add(item);        
         for (int i = 0; i < outter.Count - 1; i++)  
           Gizmos.DrawLine(outter[i].position, outter[i + 1].position);        
         Gizmos.DrawLine(outter[0].position, outter[outter.Count - 1].position);  
      }  
      for (int i = 0; i < points.Length; i++)  
      {  
         Gizmos.color = Color.cyan;  
         if (result != null)  
         {  
           if (result.Contains(points[i]))  
             Gizmos.color = Color.yellow;  
         }  
         Gizmos.DrawSphere(points[i].position, size);  
       }  
   }  
}  
Nothing too magical happening, but still...
List<Transform> outter = new List<Transform>();  
foreach (var item in result)  
      outter.Add(item);        
for (int i = 0; i < outter.Count - 1; i++)  
      Gizmos.DrawLine(outter[i].position, outter[i + 1].position);        
Gizmos.DrawLine(outter[0].position, outter[outter.Count - 1].position);  
We need a list to iterate over the values that's why I am taking the values from the HashSet and putting it into the 'outter' list. Then I'm just drawing them.
Gizmos.color = Color.cyan;  
if (result != null)  
   {  
      if (result.Contains(points[i]))  
          Gizmos.color = Color.yellow;  
   }  
Gizmos.DrawSphere(points[i].position, size);  
Then I'm drawing the spheres either cyan or yellow depending on whether the point is the convex hull or not.
For more cool algorithm implementations go HERE.
You can get the source code HERE.
Get Unity package HERE.
Support Bitshift Programmer by leaving a like on Bitshift Programmer Facebook Page and be updated as soon as there is a new blog post.
If you have any questions that you might have about shaders or unity development in general don't be shy and leave a message on my facebook page or down in the comments.

Assets Worth Checking Out

POPULAR POSTS

Introduction To Replacement Shaders & Shader Keywords

What is a replacement shader? A replacement shader is a shader that gets applied to every object being rendered.
Since the camera determines what objects end up being shown on screen, The functionality for setting up replacement shaders are in the camera class as well.

A good use case of a replacement shader would be in making effects like SSAO.
Here we need access to the normals and the depth information so a replacement shader that displays only the normals can be rendered ( stored in a render texture ) and then another shader that displays the depth information ( stored in a render texture ) and then the final image is rendered with the SSAO effect by taking the two render textures as input and doing a bunch of calculations.

Another use case would be to visualize the environment differently for various reasons like how they did in City Skylines.

The function that performs shader replacement is:
Camera.SetReplacementShader( Shader shader, string replacementTag ) Takes in a shader as …

Curved Surface Shader [ Unity Implementation ]

Curved Surface Shader This is the shader that we will be having at the end of this tutorial.
 The curved surface shader is capable of achieving really varied visual effects from showing space-time curve due to gravity to a generic curved world shader that is seen in endless runners like Subway Surfers.
The concepts that you learn here can open you up to a new way of looking at shaders and if you didn't think they were the coolest thing ever already, hopefully let this be the turning point.😝.

Both the examples show above use the same exact material is just that different values have been passed to the shader.
Start by creating a new unlit shader in Unity and we will work our way from there.
First we define what the properties are:
_MainTex("Texture", 2D) = "white" {} _BendAmount("Bend Amount", Vector) = (1,1,1,1) _BendOrigin("Bend Origin", Vector) = (0,0,0,0) _BendFallOff("Bend Falloff", float) = 1.0 _BendFallOffStr("Falloff s…

Noise Generation Algorithms : White Noise With Shader Example

White Noise Many of you may know what white noise is and how it can be used in code.
But there are a spectrum of  'Coloured Noises' as well these include ones like brown noise, pink, blue and violet noise all these have slightly different properties and can be very useful in creating different effects. More details on coloured noises can be found HERE.
A lot of movies as well as video games like City Skylines use white noise to add a feeling of realism and texture to the on screen content, it's also referred to as a grain filter or overlay. It is a really subtle effect and should not be over used.
Creating white noise boils down to the random function used.
While writing C++ code or C# code we have access to good pseudo-random functions from their standard libraries... but while writing code in shaders we usually have to create our own pseudo-random number generator.
One really good way of making one is by using the frac( in Unity & HLSL ) or fract( in GLSL ) functio…

Alto's Adventure Style Procedural Surface Generation Part 1

Alto's Adventure Style - Procedural Surface Generation This game appears to be a strictly 2D game but if you have played it enough you will notice that some of the art assets used look like it's 3D ( I don't know if they are tho ). If you haven't played the game you are missing out on one the most visually pleasing and calming games out there ( There is literally a mode called Zen mode in the game ).
Anyway, I am going to show you how to make a procedural 2D world ( without the trees, buildings and background ) like in Alto's Adventure.
But you may notice I have a plane which is in in the Z-axis giving a depth to the surface which is not there in Alto's Adventure but if you want to know how to do it then that will be in part 2.
To achieve the same effect of Alto's Adventure ( I'm leaving that up to you ) only minimal changes are needed to the code that I am going to explain.
We are going to be using the plane mesh in unity for creating the 2D surface as th…

Making An Audio Visualizer In Unity With Blend Shapes & Audio API - Part 1

Audio Visualizer with Unity Audio API & Blend Shapes This tutorial is going to be split into two parts, The first part (this one) we are going to be covering Blendshapes, what they are, what it's used for and how to use them in Unity and in Part 2 we will cover integrating that with Unity's audio API to get the desired result.
So now you got a feel as to what we will be making.
Blendshapes are just like transitions that the vertices go through to go from being in one position in space to another with the help of a transition value. All the major 3d modelling software from Blender to Maya has it.
As you can see that the model has 2 shape keys, one called MinValue and another called MaxValue, these names don't carry over into unity but the index of the shape key( blend shape ) matters.
We will be manipulating the MaxValue to make changes to the mesh.
This is how the model looks when it has it's MaxValue set at zero.
You may have guessed now that it is just a basic linear i…