Search Results

Search found 2362 results on 95 pages for 'fear matrix'.

Page 10/95 | < Previous Page | 6 7 8 9 10 11 12 13 14 15 16 17  | Next Page >

  • D3DXMatrixDecompose gives different quaternion than D3DXQuaternionRotationMatrix

    - by Fraser
    In trying to solve this problem, I tracked down the problem to the conversion of the rotation matrix to quaternion. In particular, consider the following matrix: -0.02099178 0.9997436 -0.008475631 0 0.995325 0.02009799 -0.09446743 0 0.09427284 0.01041905 0.9954919 0 0 0 0 1 SlimDX.Quaternion.RotationMatrix (which calls D3DXQuaternionRotationMatrix gives a different answer than SlimDX.Matrix.Decompose (which uses D3DXMatrixDecompose). The answers they give (after being normalized) are: X Y Z W Quaternion.RotationMatrix -0.05244324 0.05137424 0.002209336 0.9972991 Matrix.Decompose 0.6989997 0.7135442 -0.03674842 -0.03006023 Which are totally different (note the signs of X, Z, and W are different). Note that these aren't q/-q (two quaternions that represent the same rotation); they face completely different directions. I've noticed that with matrices for rotations very close to that one (successive frames in the animation) that the Matrix.Decompose version gives a solution that flips around wildly and occasionally goes into the desired position, while the Quaternion.RotationMatrix version gives solutions that are stable but go in the wrong direction. This is only for the right arm in my animation -- for the left arm, both functions give the correct solution, which is the same quaternion within error tolerances. This makes me think that there's some sort of numeric instability or weird stuff with signs going on. I tried implementing this and then this, but both gave me a completely incorrect solution (even for the matricies where the SlimDX ones were working correctly) -- maybe the rows and columns are flipped?

    Read the article

  • Rotation matrix for a 3D vector

    - by Shashwat
    I have a direction vector on which I have to apply some rotation to align it to positive z-axis. To use Matrix.CreateRotationX(angle) of XNA, I need the angle for which I'd have to compute cos or tan inverse. I think this is a complex task to do. Also, eventually those are also converted to sin(angle) and cos(angle) in the matrix. Is there any inbuilt way to create rotation matrix from a 3D vector? However, I can write the function but still asking if there is one already there.

    Read the article

  • Translating an object along its heading

    - by Kuros
    I am working on a simulation that requires me to have several objects moving around in 3D space (text output of their current position on the grid and heading is fine, I do not need graphics), and I am having some trouble getting objects to move along their relative headings. I have a basic understanding of vectors and matrices. I am using a vector to represent their position, and I am also using Euler Angles. I can translate one of my entities with a matrix along whatever axis, and I can alter their heading. For example, if I have an entity at (order is XYZ) 1, 1, 1, with a heading of 0, I can apply a translation matrix to get them to talk to 1, 1, 2 fine. However, if I change their heading to 270, they still walk to 1, 1, 3, instead of 2, 1, 2 as I desire. I have a feeling that my problem lies in not translating my matrix from world space to object space, but I am not sure how to go about that. How can I do this? Addition: I am using 3D vectors to represent their current position and their heading (using the three euler angles). For now, all I want to do is have an entity walk in a square, reporting their current position at each step. So, assuming it starts at 10, 10, 10 I want it to walk as follows: 10,10,10 -> 10, 10, 15 10, 10, 15 -> 5, 10, 15 5, 10, 15 -> 5, 10, 10 5, 10, 10 -> 10, 10, 10 My 1 Z unit translation matrix is as follows: [1 0 0 0] [0 1 0 0] [0 0 1 1] [0 0 0 1] My rotation matrix is as follows: [0 0 1 0] [0 1 0 0] [-1 0 0 0] [0 0 0 1]

    Read the article

  • Matrix Math in SQL

    Relational Datbases have tables as data structures, not arrays. This makes it tricky and slow to do matrix operations, but it doesn't mean it is impossible to do. Joe gives the Celko Slant on how to go about doing Matrix Math in SQL. 12 essential tools for database professionalsThe SQL Developer Bundle contains 12 tools designed with the SQL Server developer and DBA in mind. Try it now.

    Read the article

  • Android Rotate Matrix

    - by jitm
    Hello, I have matrix. This matrix represents array x and y coordinates. For example float[] src = {7,1,7,2,7,3,7,4}; I need to rotate this coordinates to 90 degrees. I use android.graphics.Matrix like this: float[] src = {7,1,7,2,7,3,7,4}; float[] dist = new float[8]; Matrix matrix = new Matrix(); matrix.preRotate(90.0f); matrix.mapPoints(dist,src); after operation rotate I have array with next values -1.0 7.0 -2.0 7.0 -3.0 7.0 -4.0 7.0 Its is good for area with 360 degrees. And how do rotate in area from 0 to 90? I need set up center of circle in this area but how ? Thanks.

    Read the article

  • How to browse a Matrix returned by the OpenCV2.0 Canny() function ?

    - by user290613
    Hi all, I've been browsing the web to get an answer but no way to put my hand on it. I call the Canny function that fills me a matrix Canny(src, dst, 220, 299, 3 ); the dst variable is now a Matrix that is according to me of type CV_8UC1, 8bit, 1channel. Now I would like to browse this matrix according to rows and columns, so there was a trick that I've seen on stackoverflow dst.at<uchar*>(3, 5) In my case it crashes. (Assertion Failed, unknown function, cxmat.hpp line 450) My matrix information are myMatrix.width() = 181 myMatrix.height() = 65 One more thing, when I do std::cout << dst.depth() << std::endl; It always retrieves me 0 and not 8 (which I guess Im suppose to get) Thank you very much for all who will reply,

    Read the article

  • Simple matrix example using C++ template class

    - by skyeagle
    I am trying to write a trivial Matrix class, using C++ templates in an attempt to brush up my C++, and also to explain something to a fellow coder. This is what I have som far: template class<T> class Matrix { public: Matrix(const unsigned int rows, const unsigned int cols); Matrix(const Matrix& m); Matrix& operator=(const Matrix& m); ~Matrix(); unsigned int getNumRows() const; unsigned int getNumCols() const; template <class T> T getCellValue(unsigned int row, unsigned col) const; template <class T> void setCellValue(unsigned int row, unsigned col, T value) const; private: // Note: intentionally NOT using smart pointers here ... T * m_values; }; template<class T> inline T Matrix::getCellValue(unsigned int row, unsigned col) const { } template<class T> inline void Matrix::setCellValue(unsigned int row, unsigned col, T value) { } I'm stuck on the ctor, since I need to allocate a new[] T, it seems like it needs to be a template method - however, I'm not sure I have come accross a templated ctor before. How can I implemnt the ctor?

    Read the article

  • How can I find the dimensions of a matrix in Python?

    - by PBD10017
    How can I find the dimensions of a matrix in Python. Len(A) returns only one variable. Edit: Hi Thanks. close = dataobj.get_data(timestamps, symbols, closefield) Is (I assume) generating a matrix of integers (less likely strings). I need to find the size of that matrix, so I can run some tests without having to iterate through all of the elements. As far as the data type goes, I assume it's an array of arrays (or list of lists).

    Read the article

  • averaging matrix efficiently

    - by user248237
    in Python, given an n x p matrix, e.g. 4 x 4, how can I return a matrix that's 4 x 2 that simply averages the first two columns and the last two columns for all 4 rows of the matrix? e.g. given: a = array([[1, 2, 3, 4], [5, 6, 7, 8], [9, 10, 11, 12], [13, 14, 15, 16]]) return a matrix that has the average of a[:, 0] and a[:, 1] and the average of a[:, 2] and a[:, 3]. I want this to work for an arbitrary matrix of n x p assuming that the number of columns I am averaging of n is obviously evenly divisible by n. let me clarify: for each row, I want to take the average of the first two columns, then the average of the last two columns. So it would be: 1 + 2 / 2, 3 + 4 / 2 <- row 1 of new matrix 5 + 6 / 2, 7 + 8 / 2 <- row 2 of new matrix, etc. which should yield a 4 by 2 matrix rather than 4 x 4. thanks.

    Read the article

  • how to retain the animated position in opengl es 2.0

    - by Arun AC
    I am doing frame based animation for 300 frames in opengl es 2.0 I want a rectangle to translate by +200 pixels in X axis and also scaled up by double (2 units) in the first 100 frames Then, the animated rectangle has to stay there for the next 100 frames. Then, I want the same animated rectangle to translate by +200 pixels in X axis and also scaled down by half (0.5 units) in the last 100 frames. I am using simple linear interpolation to calculate the delta-animation value for each frame. Pseudo code: The below drawFrame() is executed for 300 times (300 frames) in a loop. float RectMVMatrix[4][4] = {1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1 }; // identity matrix int totalframes = 300; float translate-delta; // interpolated translation value for each frame float scale-delta; // interpolated scale value for each frame // The usual code for draw is: void drawFrame(int iCurrentFrame) { // mySetIdentity(RectMVMatrix); // comment this line to retain the animated position. mytranslate(RectMVMatrix, translate-delta, X_AXIS); // to translate the mv matrix in x axis by translate-delta value myscale(RectMVMatrix, scale-delta); // to scale the mv matrix by scale-delta value ... // opengl calls glDrawArrays(...); eglswapbuffers(...); } The above code will work fine for first 100 frames. in order to retain the animated rectangle during the frames 101 to 200, i removed the "mySetIdentity(RectMVMatrix);" in the above drawFrame(). Now on entering the drawFrame() for the 2nd frame, the RectMVMatrix will have the animated value of first frame e.g. RectMVMatrix[4][4] = { 1.01, 0, 0, 2, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1 };// 2 pixels translation and 1.01 units scaling after first frame This RectMVMatrix is used for mytranslate() in 2nd frame. The translate function will affect the value of "RectMVMatrix[0][0]". Thus translation affects the scaling values also. Eventually output is getting wrong. How to retain the animated position without affecting the current ModelView matrix? =========================================== I got the solution... Thanks to Sergio. I created separate matrices for translation and scaling. e.g.CurrentTranslateMatrix[4][4], CurrentScaleMatrix[4][4]. Then for every frame, I reset 'CurrentTranslateMatrix' to identity and call mytranslate( CurrentTranslateMatrix, translate-delta, X_AXIS) function. I reset 'CurrentScaleMatrix' to identity and call myscale(CurrentScaleMatrix, scale-delta) function. Then, I multiplied these 'CurrentTranslateMatrix' and 'CurrentScaleMatrix' to get the final 'RectMVMatrix' Matrix for the frame. Pseudo Code: float RectMVMatrix[4][4] = {0}; float CurrentTranslateMatrix[4][4] = {0}; float CurrentScaleMatrix[4][4] = {0}; int iTotalFrames = 300; int iAnimationFrames = 100; int iTranslate_X = 200.0f; // in pixels float fScale_X = 2.0f; float scaleDelta; float translateDelta_X; void DrawRect(int iTotalFrames) { mySetIdentity(RectMVMatrix); for (int i = 0; i< iTotalFrames; i++) { DrawFrame(int iCurrentFrame); } } void getInterpolatedValue(int iStartFrame, int iEndFrame, int iTotalFrame, int iCurrentFrame, float *scaleDelta, float *translateDelta_X) { float fDelta = float ( (iCurrentFrame - iStartFrame) / (iEndFrame - iStartFrame)) float fStartX = 0.0f; float fEndX = ConvertPixelsToOpenGLUnit(iTranslate_X); *translateDelta_X = fStartX + fDelta * (fEndX - fStartX); float fStartScaleX = 1.0f; float fEndScaleX = fScale_X; *scaleDelta = fStartScaleX + fDelta * (fEndScaleX - fStartScaleX); } void DrawFrame(int iCurrentFrame) { getInterpolatedValue(0, iAnimationFrames, iTotalFrames, iCurrentFrame, &scaleDelta, &translateDelta_X) mySetIdentity(CurrentTranslateMatrix); myTranslate(RectMVMatrix, translateDelta_X, X_AXIS); // to translate the mv matrix in x axis by translate-delta value mySetIdentity(CurrentScaleMatrix); myScale(RectMVMatrix, scaleDelta); // to scale the mv matrix by scale-delta value myMultiplyMatrix(RectMVMatrix, CurrentTranslateMatrix, CurrentScaleMatrix);// RectMVMatrix = CurrentTranslateMatrix*CurrentScaleMatrix; ... // opengl calls glDrawArrays(...); eglswapbuffers(...); } I maintained this 'RectMVMatrix' value, if there is no animation for the current frame (e.g. 101th frame onwards). Thanks, Arun AC

    Read the article

  • Find points whose pairwise distances approximate a given distance matrix

    - by Stephan Kolassa
    Problem. I have a symmetric distance matrix with entries between zero and one, like this one: D = ( 0.0 0.4 0.0 0.5 ) ( 0.4 0.0 0.2 1.0 ) ( 0.0 0.2 0.0 0.7 ) ( 0.5 1.0 0.7 0.0 ) I would like to find points in the plane that have (approximately) the pairwise distances given in D. I understand that this will usually not be possible with strictly correct distances, so I would be happy with a "good" approximation. My matrices are smallish, no more than 10x10, so performance is not an issue. Question. Does anyone know of an algorithm to do this? Background. I have sets of probability densities between which I calculate Hellinger distances, which I would like to visualize as above. Each set contains no more than 10 densities (see above), but I have a couple of hundred sets. What I did so far. I did consider posting at math.SE, but looking at what gets tagged as "geometry" there, it seems like this kind of computational geometry question would be more on-topic here. If the community thinks this should be migrated, please go ahead. This looks like a straightforward problem in computational geometry, and I would assume that anyone involved in clustering might be interested in such a visualization, but I haven't been able to google anything. One simple approach would be to randomly plonk down points and perturb them until the distance matrix is close to D, e.g., using Simulated Annealing, or run a Genetic Algorithm. I have to admit that I haven't tried that yet, hoping for a smarter way. One specific operationalization of a "good" approximation in the sense above is Problem 4 in the Open Problems section here, with k=2. Now, while finding an algorithm that is guaranteed to find the minimum l1-distance between D and the resulting distance matrix may be an open question, it still seems possible that there at least is some approximation to this optimal solution. If I don't get an answer here, I'll mail the gentleman who posed that problem and ask whether he knows of any approximation algorithm (and post any answer I get to that here).

    Read the article

  • Data Movement and the Decision Matrix

    - by BuckWoody
    Maybe it’s my military background, or maybe I’ve always had this predilection, but I like to use two devices when I need to make a complex decision: A checklist and a decision matrix. I like to use a checklist because it ensures that I remember the big bits of what I need to do, and brings up questions or areas that I didn’t think about when evaluating options for the decision. And the decision matrix – that’s the thing I use to actually lay out those options. It’s simply a spreadsheet-like grid (I use Excel, but paper and pencil works as well) that lays out the requirements or advantages for the decision across the top, and the options I have on the left-hand side. Then in the “cells” I put whether or not that option on the left will meet the requirement in that column. I then simply “weight” each cell to organize the choices by best-fit. The right answer (or answers) will float right to the top. I was asked yesterday about options for moving data in SQL Server to another system. There are just dozens of ways to do this, from bcp to Replication, each with certain advantages and costs. But asking the questions for the top row first helped me show the person that it isn’t a particular technology that is important, it’s laying out those requirements and thinking about which elements are more important than the other. For instance, is it more important to have the data moved all the time, or is it OK if that happens once in a while? Does the data have to move in two directions or just one? All of these will help that answer jump right out. Try it sometime – it’s a great learning exercise, since it will force you to focus on filling out the matrix. The answer is out there, Neo. Share this post: email it! | bookmark it! | digg it! | reddit! | kick it! | live it!

    Read the article

  • Binding BoundingSpheres to a world matrix in XNA

    - by NDraskovic
    I made a program that loads the locations of items on the scene from a file like this: using (StreamReader sr = new StreamReader(OpenFileDialog1.FileName)) { String line; while ((line = sr.ReadLine()) != null) { red = line.Split(','); model = row[0]; x = row[1]; y = row[2]; z = row[3]; elements.Add(Convert.ToInt32(model)); data.Add(new Vector3(Convert.ToSingle(x), Convert.ToSingle(y), Convert.ToSingle(z))); sfepheres.Add(new BoundingSphere(new Vector3(Convert.ToSingle(x), Convert.ToSingle(y), Convert.ToSingle(z)), 1f)); } I also have a list of BoundingSpheres (called spheres) that adds a new bounding sphere for each line from the file. In this program I have one item (a simple box) that moves (it has its world matrix called matrixBox), and other items are static entire time (there is a world matrix that holds those elements called simply world). The problem i that when I move the box, bounding spheres move with it. So how can I bind all BoundingSpheres (except the one corresponding to the box) to the static world matrix so that they stay in their place when the box moves?

    Read the article

  • Migrating SQL Server Compact Edition (SQL CE) database to SQL Server using Web Matrix

    - by Harish Ranganathan
    One of the things that is keeping us busy is the Web Camps we are delivering across 5 cities.  If you are a reader of this blog, and also attended one of these web camps, there is a good chance that you have seen me since I was there in all the places, so far.  The topics that we cover include Visual Studio 2010 SP1, SQL CE, ASP.NET MVC & HTML5.  Whenever I talk about SQL CE, the immediate response is that, people are wow that Microsoft has shipped a FREE compact edition database, which is an embedded database that can be x-copy deployed.  If you think, well didn’t Microsoft ship SQL Express which is FREE?  The difference is that, SQL Express runs as a service in the machine (if you open SQL Configuration Manager, you can notice that SQL Express is running as a service along with your SQL Server Engine (if you have installed ).  This makes it that, even if you are willing to use SQL Express when you deploy your application, it needs to be installed on the production machine (hosting provider) and it needs to run as a service.  Many hosters don’t allow such services to run on their space. SQL CE comes as a x-Copy deploy-able database with just a few DLLs required to run it on the machine and they don’t even need to be installed in GAC on the production machine.  In fact, if you have Visual Studio 2010 SP1 installed, you can use the “Add Deployable Dependencies” option in Project-Properties and it would detect that SQL CE is something you would probably want to add as a deploy-able dependency for your project.  With that, it bundles the required DLLs as a part of the “_bin_deployableAssemblies” folder.  So your project can be x-Copy deployed and just works fine. However, SQL CE has the limit of 4GB storage space.  Real world applications often require more than just 4GB of data storage and it often turns out that people would like to use SQL CE for development/ramp up stages but would like to migrate to full fledged SQL Server after a while.  So, its only natural that the question arises “How do I move my SQL CE database to SQL Server”  And honestly, it doesn’t come across as a straight forward support.  I was talking to Ambrish Mishra (PM in SQL CE Team, Hyderabad) since I got this question in almost all the places where we talked about SQL CE.   He was kind enough to demonstrate how this can be accomplished using Web Matrix.  Open Web Matrix (Web Matrix can be installed for free from www.microsoft.com/web) and click on “Site from Template” Click on the “Bakery” template (since by default it uses a SQL CE database and has all the required sample data) and click “Ok”. In the project, you can navigate to the Database tab and will be able to find that the Bakery site uses a SQL CE database “bakery.sdf” Select the “bakery.sdf” and you will be able to see the “Migrate” button on the top right Once you click on the “Migrate” button, you will notice that the popup wizard opens up and by default is configured for SQL Express.  You can edit the same to point to your local SQL Server instance, or a remote server. Upon filling in the Server Name, Username and Password, when you click “Ok”, couple of things happen.  1. The database is migrated to SQL Server (local or remote – subject to permissions on remote server).   You can open up SQL Server Management Studio and connect to the server to verify that the “bakery” database exists under “Databases” node. 2. You can also notice that in Web Matrix, when you navigate to the “Files” tab and open up the web.config file, connection string now points to the SQL Server instance (yes, the Migrate button was smart enough to make this change too ) And there it is, your SQL Server Compact Edition database, now migrated to SQL Server!! In a future post, I would explain the steps involved when using Visual Studio. Cheers !!!

    Read the article

  • Movement on the X an Z axis are combined?

    - by Magicaxis
    This is probably a stupid question, but I'm trying to simply move a 3D object up, down, left, and right (Not forward or backward). The Y axis works fine, but when I increment the object's X position, the object moves BOTH right and backwards! when I decrement X, left and forwards! setPosition(getPosition().X + 2/*times deltatime*/, getPosition().Y, getPosition().Z); I was astonished that XNA doesnt have its own setPosition function, so I made a parent class for all objects with a setPosition and Draw function. Setposition simply edits a variable "mPosition" and passes it to the common draw function: // Copy any parent transforms. Matrix[] transforms = new Matrix[block.Bones.Count]; block.CopyAbsoluteBoneTransformsTo(transforms); // Draw the model. A model can have multiple meshes, so loop. foreach (ModelMesh mesh in block.Meshes) { // This is where the mesh orientation is set, as well // as our camera and projection. foreach (BasicEffect effect in mesh.Effects) { effect.EnableDefaultLighting(); effect.World = transforms[mesh.ParentBone.Index] * Matrix.CreateRotationY(MathHelper.ToRadians(mOrientation.Y)) * Matrix.CreateTranslation(mPosition); effect.View = game1.getView(); effect.Projection = game1.getProjection(); } // Draw the mesh, using the effects set above. mesh.Draw(); } I tried to work it out by attempting to increment and decrement the Z axis, but nothing happens?! So using the X axis changes the objects x and z axis', but changing the Z does nothing. Great. So how do I seperate the X and Z axis movement?

    Read the article

  • RTTI Delphi Create as TValue an n-dimensional matrix.

    - by user558126
    Good day, I had tried to make recurrent function to return a TValue as a n-dimensional. matrix(2D, 3D, 4D...) for example, this procedure will show a n-dimensional matrix(it will list all elements from a n-dimensional matrix as TValue variable): Procedure Show(X:TValue); var i:integer; begin if x.IsArray then begin for i:=0 to x.GetArrayLength-1 do show(x.GetArrayElement(i)); writeln; end else write(x.ToString,' '); end; I don't understand how to create a function to create from a TValue an n-dimensional matrix. For example i need a Function CreateDynArray(Dimensions:array of integer; Kind:TTypeKind):TValue; and the function will return a TValue which is a dynamic array how contain the dimenssions for example: Return=CreateDynArray([2,3],tkInteger); will return a TValue as tkDynArray and if i will show(Return) will list 0 0 0 0 0 0 Thank you very much, and have a nice day!

    Read the article

  • Camera Projection back Into 3D world, offset error

    - by Anthony
    I'm using XNA to simulate a robot in a 3D world and then do image analysis on what the camera sees. I have my camera looking down in front of the direction that the robot is going, and I have the robot detecting white pixels. I'm trying to take the white pixels that it finds and project them back into the 3D world so that I can see if it is actually detecting the correct pixels. I almost have it working, but there is an offset between where the white is in in the World and were I put my orange triangles (which represent what the robot things is white). /// <summary> /// Takes a bool map of and makes vertex positions based on the map. /// </summary> /// <param name="c"> The bool map</param> private void ProjectBoolMapOnGroundAnthony2(bool[,] c) { float triangleSize = 0.04f; // Point of interest in World W cordinate system. Vector3 pointOfInterest_W = Vector3.Zero; // Point of interest in Robot Cordinate system R Vector3 pointOfInterest_R = Vector3.Zero; // alpha is the angle from the robot camera to where it is looking in the center. //double alpha = Math.Atan(1.8f / 1); /// Matrix representation of the view determined by the position, target, and updirection. Matrix View = ((SimulationMain)Game).mainRobot.robotCameraView.View; /// Matrix representation of the view determined by the angle of the field of view (Pi/4), aspectRatio, nearest plane visible (1), and farthest plane visible (1200) Matrix Projection = ((SimulationMain)Game).mainRobot.robotCameraView.Projection; /// Matrix representing how the real world cordinates differ from that of the rendering by the camera. Matrix World = ((SimulationMain)Game).mainRobot.robotCameraView.World; Plane groundPlan = new Plane(Vector3.UnitZ, 0.0f); for (int x = 0; x < this.screenWidth; x++) { for (int y = 0; y < this.screenHeight; ) { if (c[x, y] == true && this.count1D < 62000) { int j = 1; Vector3 nearPlanePoint = Game.GraphicsDevice.Viewport.Unproject(new Vector3(x, y, 0), Projection, View, World); Vector3 farPlanePoint = Game.GraphicsDevice.Viewport.Unproject(new Vector3(x, y, 1), Projection, View, World); //Vector3 pointOfInterest_W = Vector3.in Ray ray = new Ray(nearPlanePoint, farPlanePoint); pointOfInterest_W = ray.Position + ray.Direction * (float) ray.Intersects(groundPlan); this.vertexArray2[this.count1D + 0].Position.X = pointOfInterest_W.X - triangleSize; this.vertexArray2[this.count1D + 0].Position.Y = pointOfInterest_W.Y - triangleSize * j; this.vertexArray2[this.count1D + 0].Position.Z = pointOfInterest_W.Z; this.vertexArray2[this.count1D + 0].Color = Color.DarkOrange; // Put another vertex a the position but +1 in the X direction triangleSize //this.vertexArray2[this.count1D + 1].Position.X = pointOnGroud.X + 3; //this.vertexArray2[this.count1D + 1].Position.Y = pointOnGroud.Y + j; this.vertexArray2[this.count1D + 1].Position.X = pointOfInterest_W.X; this.vertexArray2[this.count1D + 1].Position.Y = pointOfInterest_W.Y + triangleSize * j; this.vertexArray2[this.count1D + 1].Position.Z = pointOfInterest_W.Z; this.vertexArray2[this.count1D + 1].Color = Color.Red; // Put another vertex a the position but +1 in the X direction //this.vertexArray2[this.count1D + 0].Position.X = pointOnGroud.X; //this.vertexArray2[this.count1D + 0].Position.Y = pointOnGroud.Y + 3 + j; this.vertexArray2[this.count1D + 2].Position.X = pointOfInterest_W.X + triangleSize; this.vertexArray2[this.count1D + 2].Position.Y = pointOfInterest_W.Y - triangleSize * j; this.vertexArray2[this.count1D + 2].Position.Z = pointOfInterest_W.Z; this.vertexArray2[this.count1D + 2].Color = Color.Orange; this.count1D += 3; y += j; } else { y++; } } } } The world is a grass texture with lines on it. The world plane is normal at (0,0,1). Any ideas on why there is an offset? Any Ideas? Thanks for the help, Anthony G.

    Read the article

  • Question about Target parameter of Matrix.CreateLookAt

    - by manning18
    I have a newbie question that's causing me a little bit of confusion when experimenting with cameras and reading other peoples implementations - does this parameter represent a point or a vector? In some examples I've seen people treat it like a specific point they are looking at (eg a position in the world), other times I see people caching the orientation of the camera in a rotation matrix and simply using the Matrix.Forward property as the "target", and other times it's a vector that's the result of targetPos - camPos and also I saw a camPos + orientation.Forward I was also just playing around with hard-coded target positions with same direction eg 1 to 10000 with no discernible difference in what I saw in the scene. Is the "Target" parameter actually a position or a direction (irrespective of magnitude)? Are there any subtle differences in behaviors, common mistakes or gotchas that are associated with what values you provide, or HOW you provide this paramter? Are all the methods I mentioned above equivalent? (sorry, I've only recently started and my math is still catching up)

    Read the article

  • Cannot compute wNear and wFar from projection matrix

    - by DeadMG
    I've got the following error from Direct3D when attempting to render in 3D: Direct3D9: (WARN) :Cannot compute WNear and WFar from the supplied projection matrix Direct3D9: (WARN) :Setting wNear to 0.0 and wFar to 1.0 My projection matrix is as follows: D3DXMatrixPerspectiveFovLH( &Projection, D3DXToRadian(90), (float)GetDimensions().x / (float)GetDimensions().y, NearPlane, FarPlane ); D3DCALL(device->SetTransform( D3DTS_PROJECTION, &Projection )); The NearPlane is 0.1f, the FarPlane is 40.0f, and the dimensions are 1920x1018. This code was working earlier but I appear to have broken it, and I'm not sure where the fault is. Previously I've only encountered it if NearPlane was 0, and Google hasn't suggested any other causes either. Any suggestions?

    Read the article

  • Managing Matrix Relationships: Organization Visualization and Navigation

    - by Nancy Estell Zoder
    v\:* {behavior:url(#default#VML);} o\:* {behavior:url(#default#VML);} w\:* {behavior:url(#default#VML);} .shape {behavior:url(#default#VML);} Normal 0 false false false false EN-US X-NONE X-NONE /* Style Definitions */ table.MsoNormalTable {mso-style-name:"Table Normal"; mso-tstyle-rowband-size:0; mso-tstyle-colband-size:0; mso-style-noshow:yes; mso-style-priority:99; mso-style-qformat:yes; mso-style-parent:""; mso-padding-alt:0in 5.4pt 0in 5.4pt; mso-para-margin:0in; mso-para-margin-bottom:.0001pt; mso-pagination:widow-orphan; font-size:11.0pt; font-family:"Calibri","sans-serif"; mso-ascii-font-family:Calibri; mso-ascii-theme-font:minor-latin; mso-fareast-font-family:"Times New Roman"; mso-fareast-theme-font:minor-fareast; mso-hansi-font-family:Calibri; mso-hansi-theme-font:minor-latin; mso-bidi-font-family:"Times New Roman"; mso-bidi-theme-font:minor-bidi;} Oracle is pleased to announce the posting of our latest feature, Matrix Relationship Administration. Our continued investment in our Organization Visualization and Navigation solution is demonstrated with the release of Matrix Relationship Administration as well as the enhancements made to our Org Viewer capabilities. Some of those enhancements include the ability to export to Excel and Visio, Search, Zoom, as well as the addition of Manager Self Service transactions. Matrix relationships are relationships defined by rules or ad hoc. These relationships can include, but are not limited to, product or project affiliations, functional groups including multi dimensional relationships such as when the product, region or even the customer is the profit center. The PeopleSoft solution will enable you to configure how you work in this multi dimensional world to ensure you have the tools to be productive……. For more information, please check out the datasheet available on oracle.com, video on the feature on YouTube or contact your sales representative.

    Read the article

  • Split vector vs matrix notation for transformation

    - by seahorse
    Some rendering engines like Ogre prefer to use a individual vector based notation for transformations like the following Split vector notation: Net Transformation is represented by Scale vector = sx, sy, sz Transformation vector = tx, ty, tz Rotation Quaternion Vector = w,x,y,z Matrix notation: There are other engines which simply use a net combined transformation matrix. What are the advantages of the first notation over the second? Also for animation interpolation does it work in the first notation that we interpolate across the individual components and use the interpolated parts to get the net transformation? Is this another advantage?

    Read the article

< Previous Page | 6 7 8 9 10 11 12 13 14 15 16 17  | Next Page >