I am trying to make a 1D fractal noise function. I have a function generating every single individual graph, but am struggling with how to add them together. I am following this tutorial
https://web.archive.org/web/20160530124230/http://freespace.virgin.net/hugo.elias/models/m_perlin.htm
Here is my code for my final noise function
(I am using sfml, which is what the sf::vector2f are. It's just a vector of two floats, representing a coordinate.)
void fractalNoise() {
std::vector<sf::Vector2f> allGraphs;
std::vector<sf::Vector2f> singleNoise;
float persistance = 0.8; //represents the decrease of amplitude with frequency.
//The closer to one, the less the amplitude decreases each iteration
int nOOPM1 = 10; //number of iterations
for (int i = 0; i < nOOPM1; i++) {
float frequency = pow(2, i);
float amplitude = pow(persistance, I);
//generate a random plots of noise, equidistant on the x, and random on the Y.
//the 3 is the interpolation method(ignore this), and the 1000 is how many points to draw
singleNoise = this->interpolateNoise(
this->generateNoise(frequency, 300 * amplitude), 3, 1000);
between each point.
allGraphs.insert(allGraphs.end(), singleNoise.begin(), singleNoise.end());
}
this->noiseGenerated = allGraphs;
//every pixel stored in noiseGenerated is rendered to a window
};
I understand that the allGraphs.insert is just putting the next graph after the current one, but I am unsure how to add each graph together. Because of the nature of fractal noise, and the fact my frequencies are always changing, I can't just add the noise points before interpolating them, as they will mostly have different x values
Any help would be appreciated
Related
In my program, I am downscaling an image of 500px or larger to an extreme level of approx 16px-32px. The source image is user-specified so I do not have control over its size. As you can imagine, few pixel interpolations hold up and inevitably the result is heavily aliased.
I've tried bilinear, bicubic and square average sampling. The square average sampling actually provides the most decent results but the smaller it gets, the larger the sampling radius has to be. As a result, it gets quite slow - slower than the other interpolation methods.
I have also tried an adaptive square average sampling so that the smaller it gets the greater the sampling radius, while the closer it is to its original size, the smaller the sampling radius. However, it produces problems and I am not convinced this is the best approach.
So the question is: What is the recommended type of pixel interpolation that is fast and works well on such extreme levels of downscaling?
I do not wish to use a library so I will need something that I can code by hand and isn't too complex. I am working in C++ with VS 2012.
Here's some example code I've tried as requested (hopefully without errors from my pseudo-code cut and paste). This performs a 7x7 average downscale and although it's a better result than bilinear or bicubic interpolation, it also takes quite a hit:
// Sizing control
ctl(0): "Resize",Range=(0,800),Val=100
// Variables
float fracx,fracy;
int Xnew,Ynew,p,q,Calc;
int x,y,p1,q1,i,j;
//New image dimensions
Xnew=image->width*ctl(0)/100;
Ynew=image->height*ctl(0)/100;
for (y=0; y<image->height; y++){ // rows
for (x=0; x<image->width; x++){ // columns
p1=(int)x*image->width/Xnew;
q1=(int)y*image->height/Ynew;
for (z=0; z<3; z++){ // channels
for (i=-3;i<=3;i++) {
for (j=-3;j<=3;j++) {
Calc += (int)(src(p1-i,q1-j,z));
} //j
} //i
Calc /= 49;
pset(x, y, z, Calc);
} // channels
} // columns
} // rows
Thanks!
The first point is to use pointers to your data. Never use indexes at every pixel. When you write: src(p1-i,q1-j,z) or pset(x, y, z, Calc) how much computation is being made? Use pointers to data and manipulate those.
Second: your algorithm is wrong. You don't want an average filter, but you want to make a grid on your source image and for every grid cell compute the average and put it in the corresponding pixel of the output image.
The specific solution should be tailored to your data representation, but it could be something like this:
std::vector<uint32_t> accum(Xnew);
std::vector<uint32_t> count(Xnew);
uint32_t *paccum, *pcount;
uint8_t* pin = /*pointer to input data*/;
uint8_t* pout = /*pointer to output data*/;
for (int dr = 0, sr = 0, w = image->width, h = image->height; sr < h; ++dr) {
memset(paccum = accum.data(), 0, Xnew*4);
memset(pcount = count.data(), 0, Xnew*4);
while (sr * Ynew / h == dr) {
paccum = accum.data();
pcount = count.data();
for (int dc = 0, sc = 0; sc < w; ++sc) {
*paccum += *i;
*pcount += 1;
++pin;
if (sc * Xnew / w > dc) {
++dc;
++paccum;
++pcount;
}
}
sr++;
}
std::transform(begin(accum), end(accum), begin(count), pout, std::divides<uint32_t>());
pout += Xnew;
}
This was written using my own library (still in development) and it seems to work, but later I changed the variables names in order to make it simpler here, so I don't guarantee anything!
The idea is to have a local buffer of 32 bit ints which can hold the partial sum of all pixels in the rows which fall in a row of the output image. Then you divide by the cell count and save the output to the final image.
The first thing you should do is to set up a performance evaluation system to measure how much any change impacts on the performance.
As said precedently, you should not use indexes but pointers for (probably) a substantial
speed up & not simply average as a basic averaging of pixels is basically a blur filter.
I would highly advise you to rework your code to be using "kernels". This is the matrix representing the ratio of each pixel used. That way, you will be able to test different strategies and optimize quality.
Example of kernels:
https://en.wikipedia.org/wiki/Kernel_(image_processing)
Upsampling/downsampling kernel:
http://www.johncostella.com/magic/
Note, from the code it seems you apply a 3x3 kernel but initially done on a 7x7 kernel. The equivalent 3x3 kernel as posted would be:
[1 1 1]
[1 1 1] * 1/9
[1 1 1]
I am using cocos2d-x 3.8.
I try to create two polygon sprites with the following code.
I know we can detect intersect with BoundingBox but is too rough.
Also, I know we can use Cocos2d-x C++ Physics engine to detect collisions but doesn't it waste a lot of resource of the mobile device? The game I am developing does not need physics engine.
is there a way to detect the intersect of polygon sprites?
Thank you.
auto pinfoTree = AutoPolygon::generatePolygon("Tree.png");
auto treeSprite= Sprite::create(pinfoTree);
treeSprite-> setPosition(width / 4 * 3 - 30 , height / 2 - 200);
this->addChild(treeSprite);
auto pinfoBird = AutoPolygon::generatePolygon("Bird.png");
auto Bird= Sprite::create(pinfoTree);
Bird->setPosition(width / 4 * 3, height / 2);
this->addChild(Bird)
This is a bit more complicated: AutoPolygon gives you a bunch of triangles - the PhysicsBody::createPolygon requires a convex polygon with clockwise winding… so these are 2 different things. The vertex count might even be limited. I think Box2d’s maximum count for 1 polygon is 8.
If you want to try this you’ll have to merge the triangles to form polygons. An option would be to start with one triangle and add more as long as the whole thing stays convex. If you can’t add any more triangles start a new polygon. Add all the polygons as PhysicsShapes to your physics body to form a compound object.
I would propose that you don’t follow this path because
Autopolygon is optimized for rendering - not for best fitting
physics - that is a difference. A polygon traced with Autopolygon will always be bigger than the original sprite - Otherwise you would see rendering artifacts.
You have close to no control over the generated polygons
Tracing the shape in the app will increase your startup time
Triangle meshes and physics outlines are 2 different things
I would try some different approach: Generate the collision shapes offline. This gives you a bunch of advantages:
You can generate and tweak the polygons in a visual editor e.g. by
using PhysicsEditor
Loading the prepares polygons is way faster
You can set additional parameters like mass etc
The solution is battle proven and works out of the box
But if you want to know how polygon intersect work. You can look at this code.
// Calculate the projection of a polygon on an axis
// and returns it as a [min, max] interval
public void ProjectPolygon(Vector axis, Polygon polygon, ref float min, ref float max) {
// To project a point on an axis use the dot product
float dotProduct = axis.DotProduct(polygon.Points[0]);
min = dotProduct;
max = dotProduct;
for (int i = 0; i < polygon.Points.Count; i++) {
flaot d = polygon.Points[i].DotProduct(axis);
if (d < min) {
min = dotProduct;
} else {
if (dotProduct> max) {
max = dotProduct;
}
}
}
}
// Calculate the distance between [minA, maxA] and [minB, maxB]
// The distance will be negative if the intervals overlap
public float IntervalDistance(float minA, float maxA, float minB, float maxB) {
if (minA < minB) {
return minB - maxA;
} else {
return minA - maxB;
}
}
// Check if polygon A is going to collide with polygon B.
public boolean PolygonCollision(Polygon polygonA, Polygon polygonB) {
boolean result = true;
int edgeCountA = polygonA.Edges.Count;
int edgeCountB = polygonB.Edges.Count;
float minIntervalDistance = float.PositiveInfinity;
Vector edge;
// Loop through all the edges of both polygons
for (int edgeIndex = 0; edgeIndex < edgeCountA + edgeCountB; edgeIndex++) {
if (edgeIndex < edgeCountA) {
edge = polygonA.Edges[edgeIndex];
} else {
edge = polygonB.Edges[edgeIndex - edgeCountA];
}
// ===== Find if the polygons are currently intersecting =====
// Find the axis perpendicular to the current edge
Vector axis = new Vector(-edge.Y, edge.X);
axis.Normalize();
// Find the projection of the polygon on the current axis
float minA = 0; float minB = 0; float maxA = 0; float maxB = 0;
ProjectPolygon(axis, polygonA, ref minA, ref maxA);
ProjectPolygon(axis, polygonB, ref minB, ref maxB);
// Check if the polygon projections are currentlty intersecting
if (IntervalDistance(minA, maxA, minB, maxB) > 0)
result = false;
return result;
}
}
The function can be used this way
boolean result = PolygonCollision(polygonA, polygonB);
I once had to program a collision detection algorithm where a ball was to collide with a rotating polygon obstacle. In my case the obstacles where arcs with certain thickness. and where moving around an origin. Basically it was rotating in an orbit. The ball was also rotating around an orbit about the same origin. It can move between orbits. To check the collision I had to just check if the balls angle with respect to the origin was between the lower and upper bound angles of the arc obstacle and check if the ball and the obstacle where in the same orbit.
In other words I used the various constrains and properties of the objects involved in the collision to make it more efficient. So use properties of your objects to cause the collision. Try using a similar approach depending on your objects
Basically, I want to detect a fault in an image using logistic regression. I'm hoping to get so feedback on my approach, which is as follows:
For training:
Take a small section of the image marked "bad" and "good"
Greyscale them, then break them up into a series of 5*5 pixel segments
Calculate the histogram of pixel intensities for each of these segments
Pass the histograms along with the labels to the Logistic Regression class for training
Break the whole image into 5*5 segments and predict "good"/"bad" for each segment.
Using the sigmod function the linear regression equation is:
1/ (1 - e^(xθ))
Where x is the input values and theta (θ) is the weights. I use gradient descent to train the network. My code for this is:
void LogisticRegression::Train(float **trainingSet,float *labels, int m)
{
float tempThetaValues[m_NumberOfWeights];
for (int iteration = 0; iteration < 10000; ++iteration)
{
// Reset the temp values for theta.
memset(tempThetaValues,0,m_NumberOfWeights*sizeof(float));
float error = 0.0f;
// For each training set in the example
for (int trainingExample = 0; trainingExample < m; ++trainingExample)
{
float * x = trainingSet[trainingExample];
float y = labels[trainingExample];
// Partial derivative of the cost function.
float h = Hypothesis(x) - y;
for (int i =0; i < m_NumberOfWeights; ++i)
{
tempThetaValues[i] += h*x[i];
}
float cost = h-y; //Actual J(theta), Cost(x,y), keeps giving NaN use MSE for now
error += cost*cost;
}
// Update the weights using batch gradient desent.
for (int theta = 0; theta < m_NumberOfWeights; ++theta)
{
m_pWeights[theta] = m_pWeights[theta] - 0.1f*tempThetaValues[theta];
}
printf("Cost on iteration[%d] = %f\n",iteration,error);
}
}
Where sigmoid and the hypothesis are calculated using:
float LogisticRegression::Sigmoid(float z) const
{
return 1.0f/(1.0f+exp(-z));
}
float LogisticRegression::Hypothesis(float *x) const
{
float z = 0.0f;
for (int index = 0; index < m_NumberOfWeights; ++index)
{
z += m_pWeights[index]*x[index];
}
return Sigmoid(z);
}
And the final prediction is given by:
int LogisticRegression::Predict(float *x)
{
return Hypothesis(x) > 0.5f;
}
As we are using a histogram of intensities the input and weight arrays are 255 elements. My hope is to use it on something like a picture of an apple with a bruise and use it to identify the brused parts. The (normalized) histograms for the whole brused and apple training sets look somthing like this:
For the "good" sections of the apple (y=0):
For the "bad" sections of the apple (y=1):
I'm not 100% convinced that using the intensites alone will produce the results I want but even so, using it on a clearly seperable data set isn't working either. To test it I passed it a, labeled, completely white and a completely black image. I then run it on the small image below:
Even on this image it fails to identify any segments as being black.
Using MSE I see that the cost is converging downwards to a point where it remains, for the black and white test it starts at about cost 250 and settles on 100. The apple chuncks start at about 4000 and settle on 1600.
What I can't tell is where the issues are.
Is, the approach sound but the implementation broken? Is logistic regression the wrong algorithm to use for this task? Is gradient decent not robust enough?
I forgot to answer this... Basically the problem was in my histograms which when generated weren't being memset to 0. As to the overall problem of whether or not logistic regression with greyscale images was a good solution, the answer is no. Greyscale just didn't provide enough information for good classification. Using all colour channels was a bit better but I think the complexity of the problem I was trying to solve (bruises in apples) was a bit much for simple logistic regression on its own. You can see the results on my blog here.
I have a particle simulation program I'm currently working on that generate random particle positions in a rectangular shape. This works for now but it's not the best solution, often particles overlap and there are small holes in the rectangle. How would I initialize the particles in the shape of a rectangle?
My current loop:
for(auto i=0; i<MAXPARTICLES; i++){
int particleIndex = FindUnusedParticle(); //grab the index to give a particle life
ParticlesContainer[particleIndex].life = 100.0f; //Long particle life
//generate random positions for particles in the shape of a box with random patterns
ParticlesContainer[particleIndex].pos = glm::vec3((rand()%50)/5.0,(rand()%50)/5.0,-50.0);
// Very bad way to generate a random color
ParticlesContainer[particleIndex].r = 255;
ParticlesContainer[particleIndex].g = 0;
ParticlesContainer[particleIndex].b = 0;
ParticlesContainer[particleIndex].a = 255;
ParticlesContainer[particleIndex].size = .2f;
}
It's a more complicated problem than just using uniform distribution over two dimensions separately. If your "evenliness" is the distance between two closest particles, then there are exactly two ways to achieve that: equilateral triangles and squares (well, three if you count hexagons). The only random thing there is the position of one seed and the "direction" in which that will go.
Refer to the image:
That will give you a very regular, grid-like look, though. You could try applying random, minuscule changes to particles distributed that way, and see how that would work.
I'm trying to simulate waves by numerically integrating the wave equation using euler integration (just until I get the kinks worked out, then I'll switch to runge-kutta). I'm using an array of floats as a grid. Then I create a disturbance by changing the value of the grid at one point. Now, instead of radiating in all directions away from this point, the wave only travels in one direction, towards the upper-left, i.e. towards decreasing x and y. So, my question is how do I make the wave radiate out?
Here's my code
void Wave::dudx(float *input,float *output) //calculate du/dx
{
for(int y=0;y<this->height;y++)
{
for(int x=0;x<this->width;x++)
{
output[x+y*this->width]=(this->getPoint((x+1)%this->width,y)-this->getPoint(x,y)); //getPoint returns the value of the grid at (x,y)
}
}
}
void Wave::dudy(float *input,float *output) //calculate du/dy
{
for(int x=0;x<this->width;x++)
{
for(int y=0;y<this->height;y++)
{
output[x+y*this->width]=(this->getPoint(x,(y+1)%this->height)-this->getPoint(x,y));
}
}
}
void Wave::simulate(float dt)
{
float c=6.0f;
//calculate the spatial derivatives
this->dudx(this->points,this->buffer);
this->dudx(this->buffer,this->d2udx2);
this->dudy(this->points,this->buffer);
this->dudy(this->buffer,this->d2udy2);
for(int y=0;y<this->height;y++)
{
for(int x=0;x<this->width;x++)
{
this->points[x+y*this->width]+=c*c*(this->d2udx2[x+y*this->width]+this->d2udy2[x+y*this->width])*dt*dt; //I know that I can calculate c*c and dt*dt once, but I want to make it clear what I'm doing.
}
}
}
Just for the sake of somebody else coming here for the same problem. The usual way to convert the Laplacian to a finite difference expression on a regular grid is:
∆u(x,y) -> idx2*[u(x+1,y) + u(x-1,y) - 2*u(x,y)] +
idy2*[u(x,y+1) + u(x,y-1) - 2*u(x,y)]
where idx2 and idy2 are the inverse squares of the grid spacing in dimension x and y respectively. In the case when the grid spacing in both dimensions is the same, this simplifies to:
∆u(x,y) -> igs2*[u(x+1,y) + u(x-1,y) + u(x,y+1) + u(x,y-1) - 4*u(x,y)]
The multiplicative coefficient can be removed by hiding it inside other coefficients, e.g. c, by changing their units of measurement:
∆u(x,y) -> u(x+1,y) + u(x-1,y) + u(x,y+1) + u(x,y-1) - 4*u(x,y)
By the way, there cannot be 2D spherical waves since spheres are 3D objects. 2D waves are called circular waves.