Direct3D + mouse

[dixcie]

Hi,
a thing that I need for all the 3D games I make is the mouse, but I haven't found a good tutorial to combine Mouse and Graphics.

I'm using Direct3D7.
I wonder how I can transform the (x,y) coordinates of the mouse into 3D space. (And I know de Y-coordinate of the destination space, not X and Y)
(picking?)
Do I have to proceed like this?
transform Pmouse(x,y) 3 times:
1. projection matrix ^-1
2. view matrix ^-1
3. world matrix ^-1

Questions:
1) True?
2) Is there a function in VB to make the inverse matrix?
3) Why are there 4 rows and cols in those 3 matrices (x,y,z,?)
4) Is there a good tutorial for this problem (DX7! + VB6)

Thanx
DixCie

[Jacob Roman]

This should help you.

http://externalweb.exhedra.com/Direc...7/DI_Mouse.asp

[dixcie]

I can find the coords of the mouse easily, but I don't know how to transform those coordinates into Direct3D's 3D world. it's more a direct3d (7) question than a mouse question... (That tutorial had no relation with direct3d)
But thanks for the try!
DixCie

[Jacob Roman]

Like what do you mean? Using the Y coordinates as Z? It's that simple. Just use the Y coordinates as Z coordinates. You can't move the mouse in a 3-axis coordinate system (XYZ). It can only move 2-axises (XY). So yeah, you get the picture.

You can try downloading anyone of my projects at

www.angelfire.com/fl5/memorydll/index.html. That shows you controling the camera with the mouse as well as the numpad keys. For the Y axis in my code, just replace the Camera_Angle_X with Camera_Position_Z in the mouse event callback located in FrmMain.

[Edit]No better yet, something like this only it needs modified a bit to work with the mouse. I don't have VB in front of me so I can't test it now:

'Camera Forward
 
            Camera_Position_X = Camera_Position_X + Fast_Sin(Camera_Angle_Y) * Camera_Speed
            Camera_Position_Z = Camera_Position_Z - Fast_Cos(Camera_Angle_Y) * Camera_Speed

[Halsafar]

No JACOB, he wants to go from screen space (x,y) to world space (x,y,z)

It is done by extrapolating from screen space through the projection and camera matrix.

I had a function which did this very easily, I was going to share it but I cannot find it anywhere.

To determine the exact point, for example if you are clicking on a unit, you fire a ray from your screen space through to the 3d world and keep it going until it colldes with anything, that is where the user clicked...

Sorry I can't provide any tutorials or source.

[Jacob Roman]

Oh that's what he meant.

Slaps himself in the face

I already have something like that going in my DJ program to where I can scratch a 3D vinyl. Hang on I'll get the code.......

[Jacob Roman]

Ok in a module, declare this:

Public Direct3D_Viewport As D3DVIEWPORT8
 
'I know the data type is a vector, but it's actually returning vertices
 
Public Vertex_Out(4) As D3DVECTOR
Public Vertices(4) As D3DVECTOR

Next, over where you created your polygons, you do this:

'Here is where you have created your polygon
 
    Polygon_Vertex_List(0) = Create_Unlit_Vertex(-100, 0, -100, 0, 0, 0, 0, 0)
    Polygon_Vertex_List(1) = Create_Unlit_Vertex(100, 0, -100, 0, 0, 0, 1, 0)
    Polygon_Vertex_List(2) = Create_Unlit_Vertex(-100, 0, 100, 0, 0, 0, 0, 1)
    Polygon_Vertex_List(3) = Create_Unlit_Vertex(100, 0, -100, 0, 0, 0, 1, 0)
    Polygon_Vertex_List(4) = Create_Unlit_Vertex(100, 0, 100, 0, 0, 0, 1, 1)
    Polygon_Vertex_List(5) = Create_Unlit_Vertex(-100, 0, 100, 0, 0, 0, 0, 1)
    
    'This is the code that needs to be in there:
 
    [b]
    Vertices(0).X = Polygon_Vertex_List(0).X: Vertices(0).Y = Polygon_Vertex_List(0).Y: Vertices(0).Z = Polygon_Vertex_List(0).Z
    Vertices(1).X = Polygon_Vertex_List(1).X: Vertices(1).Y = Polygon_Vertex_List(1).Y: Vertices(1).Z = Polygon_Vertex_List(1).Z
    Vertices(2).X = Polygon_Vertex_List(4).X: Vertices(2).Y = Polygon_Vertex_List(4).Y: Vertices(2).Z = Polygon_Vertex_List(4).Z
    Vertices(3).X = Polygon_Vertex_List(5).X: Vertices(3).Y = Polygon_Vertex_List(5).Y: Vertices(3).Z = Polygon_Vertex_List(5).Z[/b]

Next, in your game loop, do this:

Direct3D_Device.GetViewport Direct3D_Viewport
 
'This converts World coordinates to screen coordinates
 
D3DXVec3Project Vertex_Out(0), Vertices(0), Direct3D_Viewport, Perspective_Transformation_Matrix, Camera_Transformation_Matrix, World_Transformation_Matrix
D3DXVec3Project Vertex_Out(1), Vertices(1), Direct3D_Viewport, Perspective_Transformation_Matrix, Camera_Transformation_Matrix, World_Transformation_Matrix
D3DXVec3Project Vertex_Out(2), Vertices(2), Direct3D_Viewport, Perspective_Transformation_Matrix, Camera_Transformation_Matrix, World_Transformation_Matrix
D3DXVec3Project Vertex_Out(3), Vertices(3), Direct3D_Viewport, Perspective_Transformation_Matrix, Camera_Transformation_Matrix, World_Transformation_Matrix

Using the Vertex_Out() variables, you test for a mouse click within that region by just using its X and Y variables. If it passes, you can have it do whatever you want.

Is this what you meant? Or was it the other way around. If it's the other way around, then you have to use DirectX's D3DXVec3Unproject with the Vertices() being your mouse coordinates.

[Halsafar]

Are you sure?
It cannot...

Where's all the math?

[Jacob Roman]

Using D3DXVec3Unproject, it'll just convert screen coordinates to World coordinates. No math needed. It does it for you. Same goes with D3DXVec3Project only you get opposite results.

[Jacob Roman]

I believe the math part you are talking about might have something to do with testing to see if the mouse is within the 3D coordinates after the conversion is made, right?

This is in OpenGL and in C++ but it might help. Converting it to DX and VB6 wouldn't be a problem.

[Jacob Roman]

Found this link in gamedev.net

http://www.mvps.org/directx/articles/rayproj.htm

[dixcie]

The mvps link seems nice!
Thanx!
I'll try to convert it into VB - code...

DixCie

[Halsafar]

Bingo Jacob!
That is the article I used to get it working the first time I did it, that is the math I saw missing.

Although I still forbid myself to use any D3DX functions unless I either know what they do or absolutly cannot find an alternative.

So the snippet I was trying to post was a D3DX free C++ version of what was in that article.

[dixcie]

Ok, first question:
how can I invert a matrix with VB and Direct3D7?

=> I want to put this 2 things into VB, but how (what is d3dmath?):

D3DMath_MatrixInvert(invMatrix,viewMatrix);
D3DMath_VectorMatrixMultiply(p1,p1,invMatrix);

[Jacob Roman]

Apparently I believe that D3DXVec3Unproject does the math for you. Look at the last post:

http://www.gamedev.net/community/for...opic_id=315786

[dixcie]

But I don't think D3DXVec3Project is compatible with DirectX7, is it?

[Jacob Roman]

Look in the object browser in VB and do a search. Remember, it's D3DXVec3Unproject that you are gonna need.

Yeah I don't have VB in front of me right now. I'm at work. So I can't tell just yet.

[Halsafar]

It sure aint, that is why I had to get all the math myself.
I'll see if I can help you.

[dixcie]

I found exactly the same tutorial for VB: http://gpwiki.org/index.php/VB:3Dmouse
but that is for DX7: I still can't do this:

D3DXMatrixInverse matInverse, 0, matView

(I tried to add DirectX8 to the project, but he gave an error on that line: 'Type d'argument ByRef incompatible' (FR) => any solutions?
=> matInverse is a D3Dmatrix, but maybe for DX7 instead of DX8???)

(but on this site I find the VectorMatrixMultiply function )

I still wonder how I can find matInverse...

[dixcie]

I did just make a Sub to resolve the problem, i think it's faster than searching on internet...

if someone is interested:

Sub InverseMatrix(srcMatrix As D3DMATRIX, destMatrix As D3DMATRIX)
 
'destination matrix = invMatrix
Dim rc(4, 8)
rc(1, 1) = srcMatrix.rc11
rc(1, 2) = srcMatrix.rc12
rc(1, 3) = srcMatrix.rc13
rc(1, 4) = srcMatrix.rc14
rc(2, 1) = srcMatrix.rc21
rc(2, 2) = srcMatrix.rc22
rc(2, 3) = srcMatrix.rc23
rc(2, 4) = srcMatrix.rc24
rc(3, 1) = srcMatrix.rc31
rc(3, 2) = srcMatrix.rc32
rc(3, 3) = srcMatrix.rc33
rc(3, 4) = srcMatrix.rc34
rc(4, 1) = srcMatrix.rc41
rc(4, 2) = srcMatrix.rc42
rc(4, 3) = srcMatrix.rc43
rc(4, 4) = srcMatrix.rc44
 
For i = 1 To 4
For j = 5 To 8
    If j = i + 4 Then
        rc(i, j) = 1
    Else
        rc(i, j) = 0
    End If
Next j
Next i
 
'1. Spil = 1,1
For i = 2 To 4
For j = 2 To 8
    rc(i, j) = rc(i, j) * rc(1, 1) - rc(i, 1) * rc(1, j)
Next j
    rc(i, 1) = 0
Next i
 
'2. Spil = 2,2
For i = 1 To 4
If i = 2 Then GoTo LineABCDE1
For j = 1 To 8
    If j <> 2 Then rc(i, j) = rc(i, j) * rc(2, 2) - rc(i, 2) * rc(2, j)
Next j
    rc(i, 2) = 0
LineABCDE1:
Next i
 
'3. Spil = 3,3
For i = 1 To 4
If i = 3 Then GoTo LineABCDE2
For j = 1 To 8
    If j <> 3 Then rc(i, j) = rc(i, j) * rc(3, 3) - rc(i, 3) * rc(3, j)
Next j
    rc(i, 3) = 0
LineABCDE2:
Next i
 
'4. Spil = 3,3
For i = 1 To 3
For j = 1 To 8
    If j <> 4 Then rc(i, j) = rc(i, j) * rc(4, 4) - rc(i, 4) * rc(4, j)
Next j
    rc(i, 4) = 0
Next i
 
destMatrix.rc11 = rc(1, 5) / rc(1, 1)
destMatrix.rc12 = rc(1, 6) / rc(1, 1)
destMatrix.rc13 = rc(1, 7) / rc(1, 1)
destMatrix.rc14 = rc(1, 8) / rc(1, 1)
destMatrix.rc21 = rc(2, 5) / rc(2, 2)
destMatrix.rc22 = rc(2, 6) / rc(2, 2)
destMatrix.rc23 = rc(2, 7) / rc(2, 2)
destMatrix.rc24 = rc(2, 8) / rc(2, 2)
destMatrix.rc31 = rc(3, 5) / rc(3, 3)
destMatrix.rc32 = rc(3, 6) / rc(3, 3)
destMatrix.rc33 = rc(3, 7) / rc(3, 3)
destMatrix.rc34 = rc(3, 8) / rc(3, 3)
destMatrix.rc41 = rc(4, 5) / rc(4, 4)
destMatrix.rc42 = rc(4, 6) / rc(4, 4)
destMatrix.rc43 = rc(4, 7) / rc(4, 4)
destMatrix.rc44 = rc(4, 8) / rc(4, 4)
 
End Sub

[Jacob Roman]

Are you sure that'll work?

[Halsafar]

I was attempting to proove it, but I cannot find the darn project I had which utilized a matrix inverse....

But, I did find something from a fairly useful matrix article, it is c++ but it is correct:

bool CMatrix4x4::inverseMatrix(CMatrix4x4 m)
{
   float tempMatrix[16] = {0};
   float d12, d13, d23, d24, d34, d41;

    d12 = m.matrix[2]  * m.matrix[7]  - m.matrix[3]  * m.matrix[6];
    d13 = m.matrix[2]  * m.matrix[11] - m.matrix[3]  * m.matrix[10];
    d23 = m.matrix[6]  * m.matrix[11] - m.matrix[7]  * m.matrix[10];
    d24 = m.matrix[6]  * m.matrix[15] - m.matrix[7]  * m.matrix[14];
    d34 = m.matrix[10] * m.matrix[15] - m.matrix[11] * m.matrix[14];
    d41 = m.matrix[14] * m.matrix[3]  - m.matrix[15] * m.matrix[2];

    tempMatrix[0] =   m.matrix[5] * d34 - m.matrix[9] * d24 + m.matrix[13] * d23;
    tempMatrix[1] = -(m.matrix[1] * d34 + m.matrix[9] * d41 + m.matrix[13] * d13);
    tempMatrix[2] =   m.matrix[1] * d24 + m.matrix[5] * d41 + m.matrix[13] * d12;
    tempMatrix[3] = -(m.matrix[1] * d23 - m.matrix[5] * d13 + m.matrix[9]  * d12);

    // Calculate the determinant.
    float determinant = m.matrix[0] * tempMatrix[0] + m.matrix[4] * tempMatrix[1] +
                       m.matrix[8] * tempMatrix[2] + m.matrix[12] * tempMatrix[3];

    // Clear if the determinant is equal to zero.  0 means matrix have no inverse.
    if(determinant == 0.0)
       {
           Clear();
         return false;
       }

    float invDeterminant = 1.0f / determinant;
    
    // Compute rest of inverse.
    tempMatrix[0] *= invDeterminant;
    tempMatrix[1] *= invDeterminant;
    tempMatrix[2] *= invDeterminant;
    tempMatrix[3] *= invDeterminant;

    tempMatrix[4] = -(m.matrix[4] * d34 - m.matrix[8] * d24 + m.matrix[12] * d23) * invDeterminant;
    tempMatrix[5] =   m.matrix[0] * d34 + m.matrix[8] * d41 + m.matrix[12] * d13  * invDeterminant;
    tempMatrix[6] = -(m.matrix[0] * d24 + m.matrix[4] * d41 + m.matrix[12] * d12) * invDeterminant;
    tempMatrix[7] =   m.matrix[0] * d23 - m.matrix[4] * d13 + m.matrix[8]  * d12  * invDeterminant;

    // Pre-compute 2x2 dets for first two rows when computing cofactors 
    // of last two rows.
    d12 = m.matrix[0]  * m.matrix[5]  - m.matrix[1]  * m.matrix[12];
    d13 = m.matrix[0]  * m.matrix[9]  - m.matrix[1]  * m.matrix[8];
    d23 = m.matrix[4]  * m.matrix[9]  - m.matrix[5]  * m.matrix[8];
    d24 = m.matrix[4]  * m.matrix[13] - m.matrix[5]  * m.matrix[12];
    d34 = m.matrix[8]  * m.matrix[13] - m.matrix[9]  * m.matrix[12];
    d41 = m.matrix[12] * m.matrix[1]  - m.matrix[13] * m.matrix[0];

    tempMatrix[8]  =   m.matrix[7] * d34 - m.matrix[11] * d24 + m.matrix[15] * d23 * invDeterminant;
    tempMatrix[9]  = -(m.matrix[3] * d34 + m.matrix[11] * d41 + m.matrix[15] * d13) * invDeterminant;
    tempMatrix[10] =   m.matrix[3] * d24 + m.matrix[7]  * d41 + m.matrix[15] * d12 * invDeterminant;
    tempMatrix[11] = -(m.matrix[3] * d23 - m.matrix[7]  * d13 + m.matrix[11] * d12) * invDeterminant;
    tempMatrix[12] = -(m.matrix[6] * d34 - m.matrix[10] * d24 + m.matrix[14] * d23) * invDeterminant;
    tempMatrix[13] =   m.matrix[2] * d34 + m.matrix[10] * d41 + m.matrix[14] * d13 * invDeterminant;
    tempMatrix[14] = -(m.matrix[2] * d24 + m.matrix[6]  * d41 + m.matrix[14] * d12) * invDeterminant;
    tempMatrix[15] =   m.matrix[2] * d23 - m.matrix[6]  * d13 + m.matrix[10] * d12 * invDeterminant;

   // Save the temp matrix to our matrix.
   matrix[0]  = tempMatrix[0];  matrix[1]  = tempMatrix[1];
   matrix[2]  = tempMatrix[2];  matrix[3]  = tempMatrix[3];
    matrix[4]  = tempMatrix[4];  matrix[5]  = tempMatrix[5];
   matrix[6]  = tempMatrix[6];  matrix[7]  = tempMatrix[7];
    matrix[8]  = tempMatrix[8];  matrix[9]  = tempMatrix[9];
   matrix[10] = tempMatrix[10]; matrix[11] = tempMatrix[11];
    matrix[12] = tempMatrix[12]; matrix[13] = tempMatrix[13];
   matrix[14] = tempMatrix[14]; matrix[15] = tempMatrix[15];

   return true;
}


bool CMatrix4x4::invertMatrix(CMatrix4x4 m)
{
   // Transpose rotation
  matrix[ 0] = m.matrix[ 0];  matrix[ 1] = m.matrix[ 4];  matrix[ 2] = m.matrix[ 8];
  matrix[ 4] = m.matrix[ 1];  matrix[ 5] = m.matrix[ 5];  matrix[ 6] = m.matrix[ 9];
  matrix[ 8] = m.matrix[ 2];  matrix[ 9] = m.matrix[ 6];  matrix[10] = m.matrix[10];
  
  // Clear shearing terms
  matrix[3] = 0.0f; matrix[7] = 0.0f; matrix[11] = 0.0f; matrix[15] = 1.0f;

  // Translation is minus the dot of tranlation and rotations
  matrix[12] = -(m.matrix[12]*m.matrix[0]) - (m.matrix[13]*m.matrix[1]) - (m.matrix[14]*m.matrix[ 2]);
  matrix[13] = -(m.matrix[12]*m.matrix[4]) - (m.matrix[13]*m.matrix[5]) - (m.matrix[14]*m.matrix[ 6]);
  matrix[14] = -(m.matrix[12]*m.matrix[8]) - (m.matrix[13]*m.matrix[9]) - (m.matrix[14]*m.matrix[10]);
   return true;
} 

[dixcie]

According to my TI 83+ calc., my method works (for 4x4 matrices)... I used the Gauss-Jordan method, but personalized a bit...

[Jacob Roman]

Hey, while we are on the subject of clicking polygons in 3D coordinates and all of that, what if you have a quad with a texture of a circle, such as a vinyl (which is what's in my DJ program) and you just wanted the the click to be within the radius of that polygon, no matter what angle it appears to be on? That way there, when you make it transparent to reveal just the vinyl, then you wouldn't see weird things like clicking within the invisible corners for when you are gonna scratch.

[Jacob Roman]

I found exactly the same tutorial for VB: http://gpwiki.org/index.php/VB:3Dmouse
but that is for DX7: I still can't do this:

D3DXMatrixInverse matInverse, 0, matView

(I tried to add DirectX8 to the project, but he gave an error on that line: 'Type d'argument ByRef incompatible' (FR) => any solutions?
=> matInverse is a D3Dmatrix, but maybe for DX7 instead of DX8???)

(but on this site I find the VectorMatrixMultiply function )

I still wonder how I can find matInverse...

Theres just one problem with that VB program. It only works with static non moving polygons, although the camera can move freely. That's why I made a thread on Getting Transformed Vertices so I can test that rather than testing the model vertices.

[dixcie]

(http://www.mvps.org/directx/articles/rayproj.htm)=> Is this tutorial working for camera's which source and dest vertices are not in the X,Y or Z axis?
(e.g.: Source(0,90,-30);Dest(0,0,0))

[Jacob Roman]

Could you possibly make a VB port from it? I was having a hardtime.