# OpenGL OpenGL Frustum Culling Problem

This topic is 3143 days old which is more than the 365 day threshold we allow for new replies. Please post a new topic.

## Recommended Posts

Hello everybody. I have a rather complex scene I want to cull via the frustum. However, for some strange reason my implementation is not working (nothing is drawn). I have looked at *many* of other people implementations, and it seems as if they are identical to mine. Here are my various data structures:
struct Plane
{
f32 a, b, c, d;
};

struct Matrix4x4
{
f32 m[16];
};

struct vector3df
{
f32 x, y, z;
};

Here are various mathematical functions:
FASTCALL struct Matrix4x4 multiplyMatrix4x4(struct Matrix4x4 a, struct Matrix4x4 b)
{
struct Matrix4x4 c;

c.m[0]  = a.m[0] * b.m[0] + a.m[1] * b.m[4] + a.m[2] * b.m[8] + a.m[3] * b.m[12];
c.m[1]  = a.m[0] * b.m[1] + a.m[1] * b.m[5] + a.m[2] * b.m[9] + a.m[3] * b.m[13];
c.m[2]  = a.m[0] * b.m[2] + a.m[1] * b.m[6] + a.m[2] * b.m[10] + a.m[3] * b.m[14];
c.m[3]  = a.m[0] * b.m[3] + a.m[1] * b.m[7] + a.m[2] * b.m[11] + a.m[3] * b.m[15];

c.m[4]  = a.m[4] * b.m[0] + a.m[5] * b.m[4] + a.m[6] * b.m[8] + a.m[7] * b.m[12];
c.m[5]  = a.m[4] * b.m[1] + a.m[5] * b.m[5] + a.m[6] * b.m[9] + a.m[7] * b.m[13];
c.m[6]  = a.m[4] * b.m[2] + a.m[5] * b.m[6] + a.m[6] * b.m[10] + a.m[7] * b.m[14];
c.m[7]  = a.m[4] * b.m[3] + a.m[5] * b.m[7] + a.m[6] * b.m[11] + a.m[7] * b.m[15];

c.m[8]  = a.m[8] * b.m[0] + a.m[9] * b.m[4] + a.m[10] * b.m[8] + a.m[11] * b.m[12];
c.m[9]  = a.m[8] * b.m[1] + a.m[9] * b.m[5] + a.m[10] * b.m[9] + a.m[11] * b.m[13];
c.m[10] = a.m[8] * b.m[2] + a.m[9] * b.m[6] + a.m[10] * b.m[10] + a.m[11] * b.m[14];
c.m[11] = a.m[8] * b.m[3] + a.m[9] * b.m[7] + a.m[10] * b.m[11] + a.m[11] * b.m[15];

c.m[12] = a.m[12] * b.m[0] + a.m[13] * b.m[4] + a.m[14] * b.m[8] + a.m[15] * b.m[12];
c.m[13] = a.m[12] * b.m[1] + a.m[13] * b.m[5] + a.m[14] * b.m[9] + a.m[15] * b.m[13];
c.m[14] = a.m[12] * b.m[2] + a.m[13] * b.m[6] + a.m[14] * b.m[10] + a.m[15] * b.m[14];
c.m[15] = a.m[12] * b.m[3] + a.m[13] * b.m[7] + a.m[14] * b.m[11] + a.m[15] * b.m[15];

return c;
}

FASTCALL void normalizePlane(struct Plane * plane)
{
f32 mag = sqrtf(plane->a * plane->a + plane->b * plane->b + plane->c * plane->c);

if(mag)
{
plane->a /= mag;
plane->b /= mag;
plane->c /= mag;
plane->d /= mag;
}
}

Here is the global frustum planes declaration:
struct Plane frustumPlane[6];

Here is the frustum extraction function:
FASTCALL void updateFrustum()
{
struct Matrix4x4 proj, modl, clip;
struct Plane * p;

glPushMatrix();

glGetFloatv(GL_PROJECTION_MATRIX, &proj.m[0]);
glGetFloatv(GL_MODELVIEW_MATRIX, &modl.m[0]);

glPopMatrix();

clip = multiplyMatrix4x4(proj, modl);

printf("{%f, %f, %f, %f\n %f, %f, %f, %f\n %f, %f, %f, %f\n %f, %f, %f, %f}\n\n",
clip.m[0], clip.m[1], clip.m[2], clip.m[3],
clip.m[4], clip.m[5], clip.m[6], clip.m[7],
clip.m[8], clip.m[9], clip.m[10], clip.m[11],
clip.m[12], clip.m[13], clip.m[14], clip.m[15]);

p = &frustumPlane;
p->a = clip.m[3]  - clip.m[0];
p->b = clip.m[7]  - clip.m[4];
p->c = clip.m[11] - clip.m[8];
p->d = clip.m[15] - clip.m[12];

p = &frustumPlane;
p->a = clip.m[3]  + clip.m[0];
p->b = clip.m[7]  + clip.m[4];
p->c = clip.m[11] + clip.m[8];
p->d = clip.m[15] + clip.m[12];

p = &frustumPlane[BOTTOM];
p->a = clip.m[3]  + clip.m[1];
p->b = clip.m[7]  + clip.m[5];
p->c = clip.m[11] + clip.m[9];
p->d = clip.m[15] + clip.m[13];

p = &frustumPlane[TOP];
p->a = clip.m[3]  - clip.m[1];
p->b = clip.m[7]  - clip.m[5];
p->c = clip.m[11] - clip.m[9];
p->d = clip.m[15] - clip.m[13];

p = &frustumPlane[BACK];
p->a = clip.m[3]  - clip.m[2];
p->b = clip.m[7]  - clip.m[6];
p->c = clip.m[11] - clip.m[10];
p->d = clip.m[15] - clip.m[14];

p = &frustumPlane[FRONT];
p->a = clip.m[3]  + clip.m[2];
p->b = clip.m[7]  + clip.m[6];
p->c = clip.m[11] + clip.m[10];
p->d = clip.m[15] + clip.m[14];

u8 i;
for (i = 0; i < 6; i++)
normalizePlane(&frustumPlane);
}

Here is the function to test if a vertex is in the frustum:
FASTCALL bool isVector3dfInFrustum(struct vector3df d)
{
if(frustumPlane.a * d.x + frustumPlane.b * d.y + frustumPlane.c * d.z + frustumPlane.d > 0)
return false;

if(frustumPlane.a * d.x + frustumPlane.b * d.y + frustumPlane.c * d.z + frustumPlane.d > 0)
return false;

if(frustumPlane[BACK].a * d.x + frustumPlane[BACK].b * d.y + frustumPlane[BACK].c * d.z + frustumPlane[BACK].d > 0)
return false;

if(frustumPlane[FRONT].a * d.x + frustumPlane[FRONT].b * d.y + frustumPlane[FRONT].c * d.z + frustumPlane[FRONT].d > 0)
return false;

if(frustumPlane[TOP].a * d.x + frustumPlane[TOP].b * d.y + frustumPlane[TOP].c * d.z + frustumPlane[TOP].d > 0)
return false;

if(frustumPlane[BOTTOM].a * d.x + frustumPlane[BOTTOM].b * d.y + frustumPlane[BOTTOM].c * d.z + frustumPlane[BOTTOM].d > 0)
return false;

return true;
}

And a simple test case scenario:
    glBegin(GL_POINTS);
glPointSize(5.0f);

s32 x, y, z;
u32 count = 0;
for(x = -100; x < 100; x++)
for(y = -100; y < 100; y++)
for(z = -100; z < 100; z++)
if(isVector3dfInFrustum(makeVector3df(x, y, z)))
{
glVertex3i(x, y, z);
count++;
}

glEnd();

printf("%lu\n", count);

This test case scenario renders 0 of the points, regardless of the camera orientation. Any ideas on what I might be doing wrong? [Edited by - n0obAtroN on June 10, 2009 2:39:47 AM]

##### Share on other sites
Edited last post, added test case scenario.

Anyone?

##### Share on other sites
Couple of things:

1. In updateFrustum(), you're not setting the matrix mode before pushing the matrix stack, so it's not clear which stack is being manipulated. Also, since you're loading the identity matrix before querying for the projection and modelview matrices, I would think that one of the two would always come back as identity (which would definitely be wrong in the case of the projection matrix, and is likely to be wrong in the case of the modelview matrix).

2. I'd recommend creating a separate function for testing a point against a plane rather than writing the test out manually each time; this will help avoid typos and copy-and-paste errors.

Frustum culling code (especially based on this particular algorithm), as well as math library code in general, is difficult to debug just by looking at it; a single typo (such as a wrong index in a matrix multiplication function) can throw the whole thing off.

Testing the functions and components individually can help. With frustum culling, I recommend confirming visually that the frustum planes are correct before moving on to culling. It can be a bit of a pain to set up, but a fairly straightforward way to generate debug graphics for the frustum is to intersect the planes in sets of three to yield the corners of the frustum, and then render the 6 quadrilaterals making up the frustum. (You'll most likely need to attach the frustum to an object other than the camera in order to see the debug graphics clearly.)

##### Share on other sites
Hi all!
I have a similar problem:
I want to calculate the points of the View frustum (after the transformations of course).

I just don't know why it isn't working.

Calculating code:
float MVM[16];
void Calc_Frustum()
{
float pt[16] = {0,0,0,1, 0,0,0,1, 0,0,0,1, 1,1,1,1};
float out[16];
float inv_MVM[16];

glGetFloatv(GL_MODELVIEW_MATRIX, MVM);
MatrixTranspose(MVM,inv_MVM);

for( int i = 0; i < 5; i++ )
{
pt[0] = Frustum_Points[0];
pt[1] = Frustum_Points[1];
pt[2] = Frustum_Points[2];

MatrixMult(inv_MVM,pt,out);

Frustum_Tarnsformed_Points[0] = out[0];
Frustum_Tarnsformed_Points[1] = out[1];
Frustum_Tarnsformed_Points[2] = out[2];
}
}

void MatrixMult(float *in,float *in2, float *out)
{
out[ 0] = in[0]*in2[0]+in[4]*in2[1]+in[8]*in2[2]+in[12]*in2[3];
out[ 1] = in[1]*in2[0]+in[5]*in2[1]+in[9]*in2[2]+in[13]*in2[3];
out[ 2] = in[2]*in2[0]+in[6]*in2[1]+in[10]*in2[2]+in[14]*in2[3];
out[ 3] = in[3]*in2[0]+in[7]*in2[1]+in[11]*in2[2]+in[15]*in2[3];
out[ 4] = in[0]*in2[4]+in[4]*in2[5]+in[8]*in2[6]+in[12]*in2[7];
out[ 5] = in[1]*in2[4]+in[5]*in2[5]+in[9]*in2[6]+in[13]*in2[7];
out[ 6] = in[2]*in2[4]+in[6]*in2[5]+in[10]*in2[6]+in[14]*in2[7];
out[ 7] = in[3]*in2[4]+in[7]*in2[5]+in[11]*in2[6]+in[15]*in2[7];
out[ 8] = in[0]*in2[8]+in[4]*in2[9]+in[8]*in2[10]+in[12]*in2[11];
out[ 9] = in[1]*in2[8]+in[5]*in2[9]+in[9]*in2[10]+in[13]*in2[11];
out[10] = in[2]*in2[8]+in[6]*in2[9]+in[10]*in2[10]+in[14]*in2[11];
out[11] = in[3]*in2[8]+in[7]*in2[9]+in[11]*in2[10]+in[15]*in2[11];
out[12] = in[0]*in2[12]+in[4]*in2[13]+in[8]*in2[14]+in[12]*in2[15];
out[13] = in[1]*in2[12]+in[5]*in2[13]+in[9]*in2[14]+in[13]*in2[15];
out[14] = in[2]*in2[12]+in[6]*in2[13]+in[10]*in2[14]+in[14]*in2[15];
out[15] = in[3]*in2[12]+in[7]*in2[13]+in[11]*in2[14]+in[15]*in2[15];
} This works well in any other places.

void MatrixTranspose(float *in, float *out)
{
out[ 0] = in[0];
out[ 1] = in[4];
out[ 2] = in[8];
out[ 3] = in[12];

out[ 4] = in[1];
out[ 5] = in[5];
out[ 6] = in[9];
out[ 7] = in[13];

out[ 8] = in[2];
out[ 9] = in[6];
out[10] = in[10];
out[11] = in[14];

out[12] = in[3];
out[13] = in[7];
out[14] = in[11];
out[15] = in[15];
}

The Calc_Frustum() is called at the proper place in the main code, look at the glLoadMatrixf(MVM) for the debugging.
...
SetupView();
Calc_Frustum();
//drawing stuff
...

Drawing the calculated frustum points:
...
Draw_Frustum();
...

There are no transformations in Draw_Frustum(), just drawing the lines.
If the calculation is correct, the frustum should appear as a screen sized rectangle, regardless of the camera orientation.
The rotations apply, but the translations don't, so I see the rectangle, but its position is changing.
Maybe matrix representation of the points are wrong? Or I shouldn't invert the modelview matrix? Or something totally wrong here...

Thanks!

##### Share on other sites
The Modelview matrix isn't orthogonal.
So invert != transposed matrix
SORRY!

##### Share on other sites
The matrix stack being manipulated in updateFrustum() is the modelview matrix (GL_MODELVIEW).

##### Share on other sites
You should try to draw the frustum. This isn't easy, try to draw the normals of the
side planes, then draw the front and back planes as Quads with different colors and backface culling enabled. So you can determine if the orientations are good.

Or you could try (if you haven't yet) to comment all the statements in the isVector3dfInFrustum() function, so the points always pass, then uncomment the statements one by one, and just one at a time. Because if one of them is incorrect, the points will always fail.

The glLoadIdentity() in the updateFrustum() isn't clear for me.

##### Share on other sites
I have updated the code to try to eliminate typos. Here it is:

#define calculateFrustumPlane(a, b, c)     for(i = 0; i < 4; i++)         p->d =  clip.m[a + i * 4] b clip.m[c + i * 4]FASTCALL void updateFrustum(){    struct Matrix4x4 proj, modl, clip;    struct Plane * p;    u8 i; // used in calculateFrustumPlane() macro    glMatrixMode(GL_MODELVIEW);    glPushMatrix();    glGetFloatv(GL_PROJECTION_MATRIX, &proj.m[0]);    glGetFloatv(GL_MODELVIEW_MATRIX, &modl.m[0]);    glPopMatrix();    clip = multiplyMatrix4x4(proj, modl);    p = &frustumPlane;    calculateFrustumPlane(3, -, 0);    p = &frustumPlane;    calculateFrustumPlane(3, +, 0);    p = &frustumPlane[BOTTOM];    calculateFrustumPlane(3, +, 1);    p = &frustumPlane[TOP];    calculateFrustumPlane(3, -, 1);    p = &frustumPlane[BACK];    calculateFrustumPlane(3, -, 2);    p = &frustumPlane[FRONT];    calculateFrustumPlane(3, +, 2);    for (i = 0; i < 6; i++)        normalizePlane(&frustumPlane);}FASTCALL bool isVector3dfInFrustum(struct vector3df d){    u8 i;    for(i = 0; i < 6; i++)        if(frustumPlane[BOTTOM].d[0] * d.x + frustumPlane[BOTTOM].d[1] * d.y + frustumPlane[BOTTOM].d[2] * d.z + frustumPlane[BOTTOM].d[3] > 0)            return false;    return true;}

##### Share on other sites
Now you are only testing against a single frustum plane...

##### Share on other sites
As noted above, you're testing against the same plane each time through the loop. Also, you should really make the point-plane test a function; using a loop to eliminate some of the redundant code is a good start, but it would make much more sense to make this a separate function that can be tested in isolation and used elsewhere in your code if needed.

I would also advise against making calculateFrustumPlane a macro. It looks like you're programming in pure C (where use of macros is a little more defensible than in C++), but there are still better, safer alternatives to using a macro (for example, you could use a function rather than a macro, and use a function pointer to abstract away the summation operation).

##### Share on other sites
Thank you for your help. However, I prefer C over C++ for many reasons, and will stick to using macros as it saves a function call (much the way an inline function does).

I have come here for help in debugging frustum culling code, not for the criticism of my coding technique.

The code that has been posted on this forum is a quick test case scenario, that has been broken down into smaller, stringier chunks to make it easier for online debugging here on the forums.

Hence, I openly welcome help on matters pertaining to why the frustum is not being extracted properly.

Thank you.

##### Share on other sites
Quote:
 Thank you for your help. However, I prefer C over C++ for many reasons, and will stick to using macros as it saves a function call (much the way an inline function does).I have come here for help in debugging frustum culling code, not for the criticism of my coding technique.
For the most part, my comments on your coding style were entirely relevant. The types of things you were doing in your originally posted code (typing things out by hand, overlooking opportunities to refactor) are *exactly* the kinds of things that make complex code like this difficult to debug. So I stand by my recommendations: cleaning up the code and addressing technical and stylistic issues can often be an important (sometimes even necessary) first step in fixing this sort of problem.

As for the macro, in most cases, frustum plane extraction code will be called once per frame, maybe a couple of times if you're doing picking or something of that sort. Do you really think the performance difference (if any) between a macro and a function call is going to be measurable in this case? Or is this a premature optimization? Have you profiled the application? If so, does the frustum plane extraction code even show up in the profile?

The reason I mention the macro thing is that in some cases using macros can introduce subtle bugs that can be avoided by using real functions. As such, using a function rather than a macro here could eliminate a possible source of error, thereby making it easier to debug the algorithm as a whole. (Also, note that I said nothing about using C++ rather than C, so I'm not sure what your comment about preferring C is in relation to.)

Again, stylistic and technical analysis of code such as this is often entirely relevant, even when it's not the stated subject of the post. Sorry the suggestions bug you, but they really are intended to help.

##### Share on other sites
The application I am writing is the basis of a multi-computer multi-threaded rendering farm. The culling code is re-used for every refraction rendering pass, every reflection rendering pass, and every normal rendering pass.

To set up a test case scenario to show how much juice the frustum culling code is taking I will (arbitrarily) create a 3D scene of a single house. This house has water running in the sink, 5 candles burning, and lets say 1 mirror, and 1 window. We will render 1 minute of video at 25 FPS.

The flames of each candle is composed of about ~5000 particles. Each particle has a unique light source attached to it, and affects the lighting of objects around it. It also affects the shadow projections of the objects around it. Thats 5x5000 light sources, and 5x5000 textured quads. For each of the particles the scene must be translated to its location, and the scene re-rendered for the shadow pass of the particles light source. Each time the scene is rendered, the culling code is called. Thats approximately 5x5000 calls to eh frustum culling code.

The water in the sink has to refract the scene, and reflect the scene. It must re-render all the geometry for each pass. This multiplies the frustum calls by 3 (original pass, reflection pass, and refraction pass). That is approximately 3x5x5000 calls to the frustum code.

The mirror only reflects the scene, and hence multiplies the frustum calls by 2. That is approximately 2x2x5x5000 calls to the frustum code.

The window both refracts the scene and reflects the scene, and multiplies the number of calls by 3. That is approximately 3x2x2x5x5000 calls to the frustum code.

That is a total of 300,000 calls to the frustum code in a single frame. Running at 25 FPS for 60 seconds, it is a grand total of 450,000,000 calls to the frustum code.

Rendering a scene of a house, with a window, a mirror, and 5 candles for 1 minute has called the frustum code 450,000,000 times. This is not a realistic scenario, as that would include a lot more in the scene and a lot more calls to the frustum code.

As you can see the frustum code may not be the biggest juice sucker, but still uses a massive amount of juice. This is why I am counting the number of function calls I use, limit the number of variables (this includes variables used by loops), and using FASTCALL calling conventions. This results in much stringier code, yes, but also much faster in the long run.

I would love to write it in a less-stringy way, but its not an option.

With that said, any ideas why the frustum is not being extracted correctly?

##### Share on other sites
Quote:
 The application I am writing is the basis of a multi-computer multi-threaded rendering farm. The culling code is re-used for every refraction rendering pass, every reflection rendering pass, and every normal rendering pass. To set up a test case scenario to show how much juice the frustum culling code is taking I will (arbitrarily) create a 3D scene of a single house. This house has water running in the sink, 5 candles burning, and lets say 1 mirror, and 1 window. We will render 1 minute of video at 25 FPS.The flames of each candle is composed of about ~5000 particles. Each particle has a unique light source attached to it, and affects the lighting of objects around it. It also affects the shadow projections of the objects around it. Thats 5x5000 light sources, and 5x5000 textured quads. For each of the particles the scene must be translated to its location, and the scene re-rendered for the shadow pass of the particles light source. Each time the scene is rendered, the culling code is called. Thats approximately 5x5000 calls to eh frustum culling code.The water in the sink has to refract the scene, and reflect the scene. It must re-render all the geometry for each pass. This multiplies the frustum calls by 3 (original pass, reflection pass, and refraction pass). That is approximately 3x5x5000 calls to the frustum code.The mirror only reflects the scene, and hence multiplies the frustum calls by 2. That is approximately 2x2x5x5000 calls to the frustum code.The window both refracts the scene and reflects the scene, and multiplies the number of calls by 3. That is approximately 3x2x2x5x5000 calls to the frustum code.That is a total of 300,000 calls to the frustum code in a single frame. Running at 25 FPS for 60 seconds, it is a grand total of 450,000,000 calls to the frustum code.
So, you're saying the frustum code is called a lot? (j/k :)

I didn't spot anything, but here are some things you might double-check:

1. Did you already fix the loop that was only testing against the bottom frustum plane? (Just checking...)

2. There are two forms of the frustum extraction algorithm you're using: one for D3D-style projection matrices, and one for OpenGL-style projection matrices (they differ in that D3D uses a distance of 0 for the near plane of the canonical view volume, while OpenGL uses a distance of -1). You might check to make sure you're using the right form of the algorithm, just in case.

I'm still not clear on how you're handling the matrices exactly (for example, right now you have a matrix push-pop in your code that doesn't actually do anything), so I can't really comment on that part of the code.

Lastly, have you tried attaching the frustum to a 'test' object of some sort and rendering it so that you can see if it's correct? That might help you nail down the source of the problem (if the frustum is obviously wrong, for example, that will point towards the frustum extraction or matrix transform code rather than the culling code).

##### Share on other sites
As I have now idea whats wrong, I would suggest you start by culling against one of the planes at a time. Also, try to cull against a different camera than your actual openGL view camera, then you can render the extracted planes as huge squares and literally see whats going one.

As long as it doesn't work, you should not care about speed. As soon as you get it working, you can refactor the code to your preferred lightning fast equivalent.

##### Share on other sites
Quote:
 Original post by szecsYou should try to draw the frustum. This isn't easy, try to draw the normals of the side planes, then draw the front and back planes as Quads with different colors and backface culling enabled. So you can determine if the orientations are good.Or you could try (if you haven't yet) to comment all the statements in the isVector3dfInFrustum() function, so the points always pass, then uncomment the statements one by one, and just one at a time. Because if one of them is incorrect, the points will always fail.

These ideas are totally wrong?
Or haven't you read this comment?
Or maybe my English :D
It's maybe 5 minutes of work and will work.
If it's bullshit than Sorry!

##### Share on other sites
#1:
Expanding your calculateFrustumPlane() macro reveals your method to be exactly like mine (with one exception covered in #2). This is a problem, since my matrices are row-major and yours are column-major. Transpose your matrix before using this routine, or rewrite the routine to work with column-major matrices.

#2:
Plane distances are always stored negative. This reduces a lot of otherwise redundant negations in plane equations.
This is the difference between my routine and yours.

A snippet of mine:
		// Left clipping plane.		pLeftPlane.n.x = _mMatrix._14 + _mMatrix._11;		pLeftPlane.n.y = _mMatrix._24 + _mMatrix._21;		pLeftPlane.n.z = _mMatrix._34 + _mMatrix._31;		pLeftPlane.dist = -(_mMatrix._44 + _mMatrix._41);		// Right clipping plane.		pRightPlane.n.x = _mMatrix._14 - _mMatrix._11;		pRightPlane.n.y = _mMatrix._24 - _mMatrix._21;		pRightPlane.n.z = _mMatrix._34 - _mMatrix._31;		pRightPlane.dist = -(_mMatrix._44 - _mMatrix._41);

And the associated way to compare points:
	// Determine on which side of the plane is given point is.	CPlane3::PLANE_INTERSECT LSM_CALL CIntersection::ClassifyPoint( const CPlane3 &_pPlane, const CVector3 &_vPoint ) {		FXREAL fDist = _pPlane.n.Dot( _vPoint ) - _pPlane.dist;		if ( fDist > FX_PLANE_THICKNESS ) { return CPlane3::PI_FRONT; }		if ( fDist < -FX_PLANE_THICKNESS ) { return CPlane3::PI_BACK; }		return CPlane3::PI_COPLANAR;	}

#3:
Extending off the above, you can see that it is insufficient to simply return true or false; there are 3 possible cases, and for numerical robustness each must be correctly handled.

#4:
You have already stressed how performance-demanding your routine needs to be. Fixing your plane (and vector) normalization yields the following:
    f32 mag = plane->a * plane->a + plane->b * plane->b + plane->c * plane->c;    if(mag)    {        mag = 1.0f / sqrt( mag );        plane->a *= mag;        plane->b *= mag;        plane->c *= mag;        plane->d *= mag;    }

Never divide multiple numbers by the same denominator. Get the reciprocal via one divide and then multiply your values by that number.

Storing plane distances as negatives is another optimization, and one assumed by many textbooks (which means the formulas they give will not work if your plane is not negated).

#5:
As already mentioned, your new version of isVector3dfInFrustum() uses only the BOTTOM plane.
Surely this is not the code you are using now, but I am covering every ground here.

L. Spiro

##### Share on other sites
Thank you YogurtEmperor, your post was very helpful. I had forgoten that division is slower than multiplication. I have re-written my frustum code as follows:
#define calculateFrustumPlane(a, b, c)     p->d[0] =  clip.m[a + 0 * 4] b clip.m[c + 0 * 4];     p->d[1] =  clip.m[a + 1 * 4] b clip.m[c + 1 * 4];     p->d[2] =  clip.m[a + 2 * 4] b clip.m[c + 2 * 4];     p->d[3] =  -(clip.m[a + 3 * 4] b clip.m[c + 3 * 4]);     normalizePlane(p);FASTCALL void updateFrustum(){    static struct Matrix4x4 proj, modl, clip;    static struct Plane * p;    glMatrixMode(GL_MODELVIEW);    glPushMatrix();    glGetFloatv(GL_PROJECTION_MATRIX, &proj.m[0]);    glGetFloatv(GL_MODELVIEW_MATRIX, &modl.m[0]);    glPopMatrix();    proj = getTransposedMatrix4x4(proj);    modl = getTransposedMatrix4x4(modl);    clip = multiplyMatrix4x4(proj, modl);    p = &frustumPlane;    calculateFrustumPlane(3, -, 0);    p = &frustumPlane;    calculateFrustumPlane(3, +, 0);    p = &frustumPlane[BOTTOM];    calculateFrustumPlane(3, +, 1);    p = &frustumPlane[TOP];    calculateFrustumPlane(3, -, 1);    p = &frustumPlane[BACK];    calculateFrustumPlane(3, -, 2);    p = &frustumPlane[FRONT];    calculateFrustumPlane(3, +, 2);}// because I am only doing point based culling (no bbox, or bsphere) if a point is on a frustum plane it would not be seen anyways, so just return false.FASTCALL bool isVector3dfInFrustum(struct vector3df d){    static u8 i;    for(i = 0; i < 6; i++)        if(frustumPlane.d[0] * d.x + frustumPlane.d[1] * d.y + frustumPlane.d[2] * d.z - frustumPlane.d[3] <= 0)            return false;    return true;}FASTCALL struct Matrix4x4 getTransposedMatrix4x4(struct Matrix4x4 a){    struct Matrix4x4 b;    b.m[0]  = a.m[0];    b.m[1]  = a.m[4];    b.m[2]  = a.m[8];    b.m[3]  = a.m[12];    b.m[4]  = a.m[1];    b.m[5]  = a.m[5];    b.m[6]  = a.m[9];    b.m[7]  = a.m[13];    b.m[8]  = a.m[2];    b.m[9]  = a.m[6];    b.m[10] = a.m[10];    b.m[11] = a.m[14];    b.m[12] = a.m[3];    b.m[13] = a.m[7];    b.m[14] = a.m[11];    b.m[15] = a.m[15];    return b;}

It is now clipping some things, but not others. Moving the camera down the X-axis culls points in the distance down the Z-axis, rather than on the X-axis.

##### Share on other sites
You should swap the order of the operands in multiplyMatrix4x4() and transpose the clip matrix again, or modify the multiplyMatrix4x4()
function by indexing it like I did it in my first reply as YogurtEmperor told about matrix indexing.
So you don't have to transpose before or after the multiplication.

And I think you should invert the modelview matrix before the operations:

glMatrixMode(GL_MODELVIEW);f32 inv_modl[16];    glPushMatrix(); //these are totally useless here    glGetFloatv(GL_PROJECTION_MATRIX, &proj.m[0]);    glGetFloatv(GL_MODELVIEW_MATRIX, &modl.m[0]);    glPopMatrix(); //these are totally useless hereinvert(&modl.m[0], inv_modl);    //inv_modl = getInvertMatrix4x4(modl);    clip = multiplyMatrix4x4(proj, inv_modl);//the modified func

.....

I had the same problem as I mentioned before, and that was the solution.

Invertion code
/*	| a1 a2 |	| b1 b2 | calculate the determinent of a 2x2 matrix*/double det2x2(const double a1, const double a2,			const double b1, const double b2) {	return a1*b2 - b1*a2;}/*	| a1 a2 a3 |	| b1 b2 b3 |	| c1 c2 c3 | calculate the determinent of a 3x3 matrix*/double det3x3(const double a1, const double a2, const double a3,			  const double b1, const double b2, const double b3,			  const double c1, const double c2, const double c3) {	return a1*det2x2(b2,b3,c2,c3) - b1*det2x2(a2,a3,c2,c3) +			c1*det2x2(a2,a3,b2,b3);}void invert(float *output, float *i) {	double a11 =  det3x3(i[5],i[6],i[7],i[9],i[10],i[11],i[13],i[14],i[15]);	double a21 = -det3x3(i[1],i[2],i[3],i[9],i[10],i[11],i[13],i[14],i[15]);	double a31 =  det3x3(i[1],i[2],i[3],i[5],i[6],i[7],i[13],i[14],i[15]);	double a41 = -det3x3(i[1],i[2],i[3],i[5],i[6],i[7],i[9],i[10],i[11]);	double a12 = -det3x3(i[4],i[6],i[7],i[8],i[10],i[11],i[12],i[14],i[15]);	double a22 =  det3x3(i[0],i[2],i[3],i[8],i[10],i[11],i[12],i[14],i[15]);	double a32 = -det3x3(i[0],i[2],i[3],i[4],i[6],i[7],i[12],i[14],i[15]);	double a42 =  det3x3(i[0],i[2],i[3],i[4],i[6],i[7],i[8],i[10],i[11]);	double a13 =  det3x3(i[4],i[5],i[7],i[8],i[9],i[11],i[12],i[13],i[15]);	double a23 = -det3x3(i[0],i[1],i[3],i[8],i[9],i[11],i[12],i[13],i[15]);	double a33 =  det3x3(i[0],i[1],i[3],i[4],i[5],i[7],i[12],i[13],i[15]);	double a43 = -det3x3(i[0],i[1],i[3],i[4],i[5],i[7],i[8],i[9],i[11]);	double a14 = -det3x3(i[4],i[5],i[6],i[8],i[9],i[10],i[12],i[13],i[14]);	double a24 =  det3x3(i[0],i[1],i[2],i[8],i[9],i[10],i[12],i[13],i[14]);	double a34 = -det3x3(i[0],i[1],i[2],i[4],i[5],i[6],i[12],i[13],i[14]);	double a44 =  det3x3(i[0],i[1],i[2],i[4],i[5],i[6],i[8],i[9],i[10]);	double det = (i[0]*a11) + (i[4]*a21) + (i[8]*a31) + (i[12]*a41);	double oodet = 1/det;	output[ 0] = a11*oodet;	output[ 1] = a21*oodet;	output[ 2] = a31*oodet;	output[ 3] = a41*oodet;	output[ 4] = a12*oodet;	output[ 5] = a22*oodet;	output[ 6] = a32*oodet;	output[ 7] = a42*oodet;	output[ 8] = a13*oodet;	output[ 9] = a23*oodet;	output[10] = a33*oodet;	output[11] = a43*oodet;	output[12] = a14*oodet;	output[13] = a24*oodet;	output[14] = a34*oodet;	output[15] = a44*oodet;}

and another thing: you should pass the matrices to the functions with pointers, not the whole matrices because the CPU has to copy those matrices. It's easy to modify, a few minutes work.

I hope that helps you.

##### Share on other sites
Thank you for your help. I have re-written my code as recommended, but now nothing is drawn.

Here is the updated version:
#define calculateFrustumPlane(a, b, c)     p->d[0] =  clip.m[a + 0 * 4] b clip.m[c + 0 * 4];    p->d[1] =  clip.m[a + 1 * 4] b clip.m[c + 1 * 4];    p->d[2] =  clip.m[a + 2 * 4] b clip.m[c + 2 * 4];    p->d[3] =  -(clip.m[a + 3 * 4] b clip.m[c + 3 * 4]);    normalizePlane(p);FASTCALL void updateFrustum(){    static struct Matrix4x4 proj, modl, clip;    static struct Plane * p;    glGetFloatv(GL_PROJECTION_MATRIX, &proj.m[0]);    glGetFloatv(GL_MODELVIEW_MATRIX, &modl.m[0]);    modl = getInvertedMatrix4x4(modl);    clip = multiplyMatrix4x4(proj, modl);    p = &frustumPlane;    calculateFrustumPlane(3, -, 0);    p = &frustumPlane;    calculateFrustumPlane(3, +, 0);    p = &frustumPlane[BOTTOM];    calculateFrustumPlane(3, +, 1);    p = &frustumPlane[TOP];    calculateFrustumPlane(3, -, 1);    p = &frustumPlane[BACK];    calculateFrustumPlane(3, -, 2);    p = &frustumPlane[FRONT];    calculateFrustumPlane(3, +, 2);}// make opengl do he matrix multiplicationFASTCALL struct Matrix4x4 multiplyMatrix4x4(struct Matrix4x4 a, struct Matrix4x4 b){    struct Matrix4x4 c;	glPushMatrix();	glLoadMatrixf(&b.m[0]);	glMultMatrixf(&a.m[0]);	glGetFloatv(GL_MODELVIEW_MATRIX, &c.m[0]);	glPopMatrix();	return c;}/** thanks to szecs frome GameDev.net **/FASTCALL f32 det2x2(f32 a1, f32 a2, f32 b1, f32 b2){	return a1 * b2 - b1 * a2;}/** thanks to szecs frome GameDev.net **/FASTCALL f32 det3x3(f32 a1, f32 a2, f32 a3, f32 b1, f32 b2, f32 b3, f32 c1, f32 c2, f32 c3){	return a1 * det2x2(b2, b3, c2, c3) - b1 * det2x2(a2, a3, c2, c3) + c1 * det2x2(a2, a3, b2, b3);}/** thanks to szecs frome GameDev.net **/FASTCALL struct Matrix4x4 getInvertedMatrix4x4(struct Matrix4x4 matrix){    struct Matrix4x4 ret;	f32 a11 =  det3x3(matrix.m[5], matrix.m[6], matrix.m[7], matrix.m[9], matrix.m[10], matrix.m[11],  matrix.m[13], matrix.m[14], matrix.m[15]);	f32 a21 = -det3x3(matrix.m[1], matrix.m[2], matrix.m[3], matrix.m[9], matrix.m[10], matrix.m[11],  matrix.m[13], matrix.m[14], matrix.m[15]);	f32 a31 =  det3x3(matrix.m[1], matrix.m[2], matrix.m[3], matrix.m[5], matrix.m[6],  matrix.m[7],   matrix.m[13], matrix.m[14], matrix.m[15]);	f32 a41 = -det3x3(matrix.m[1], matrix.m[2], matrix.m[3], matrix.m[5], matrix.m[6],  matrix.m[7],   matrix.m[9],  matrix.m[10], matrix.m[11]);	f32 a12 = -det3x3(matrix.m[4], matrix.m[6], matrix.m[7], matrix.m[8], matrix.m[10], matrix.m[11], matrix.m[12],  matrix.m[14], matrix.m[15]);	f32 a22 =  det3x3(matrix.m[0], matrix.m[2], matrix.m[3], matrix.m[8], matrix.m[10], matrix.m[11], matrix.m[12],  matrix.m[14], matrix.m[15]);	f32 a32 = -det3x3(matrix.m[0], matrix.m[2], matrix.m[3], matrix.m[4], matrix.m[6],  matrix.m[7],  matrix.m[12],  matrix.m[14], matrix.m[15]);	f32 a42 =  det3x3(matrix.m[0], matrix.m[2], matrix.m[3], matrix.m[4], matrix.m[6],  matrix.m[7],  matrix.m[8],   matrix.m[10], matrix.m[11]);	f32 a13 =  det3x3(matrix.m[4], matrix.m[5], matrix.m[7], matrix.m[8], matrix.m[9], matrix.m[11],  matrix.m[12],  matrix.m[13], matrix.m[15]);	f32 a23 = -det3x3(matrix.m[0], matrix.m[1], matrix.m[3], matrix.m[8], matrix.m[9], matrix.m[11],  matrix.m[12],  matrix.m[13], matrix.m[15]);	f32 a33 =  det3x3(matrix.m[0], matrix.m[1], matrix.m[3], matrix.m[4], matrix.m[5], matrix.m[7],   matrix.m[12],  matrix.m[13], matrix.m[15]);	f32 a43 = -det3x3(matrix.m[0], matrix.m[1], matrix.m[3], matrix.m[4], matrix.m[5], matrix.m[7],   matrix.m[8],   matrix.m[9],  matrix.m[11]);	f32 a14 = -det3x3(matrix.m[4], matrix.m[5], matrix.m[6], matrix.m[8], matrix.m[9], matrix.m[10],  matrix.m[12],  matrix.m[13], matrix.m[14]);	f32 a24 =  det3x3(matrix.m[0], matrix.m[1], matrix.m[2], matrix.m[8], matrix.m[9], matrix.m[10],  matrix.m[12],  matrix.m[13], matrix.m[14]);	f32 a34 = -det3x3(matrix.m[0], matrix.m[1], matrix.m[2], matrix.m[4], matrix.m[5], matrix.m[6],   matrix.m[12],  matrix.m[13], matrix.m[14]);	f32 a44 =  det3x3(matrix.m[0], matrix.m[1], matrix.m[2], matrix.m[4], matrix.m[5], matrix.m[6],   matrix.m[8],   matrix.m[9],  matrix.m[10]);	f32 det = (matrix.m[0] * a11) + (matrix.m[4] * a21) + (matrix.m[8] * a31) + (matrix.m[12] * a41);	f32 oodet = 1.0f / det;	ret.m[0]  = a11 * oodet;	ret.m[1]  = a21 * oodet;	ret.m[2]  = a31 * oodet;	ret.m[3]  = a41 * oodet;	ret.m[4]  = a12 * oodet;	ret.m[5]  = a22 * oodet;	ret.m[6]  = a32 * oodet;	ret.m[7]  = a42 * oodet;	ret.m[8]  = a13 * oodet;	ret.m[9]  = a23 * oodet;	ret.m[10] = a33 * oodet;	ret.m[11] = a43 * oodet;	ret.m[12] = a14 * oodet;	ret.m[13] = a24 * oodet;	ret.m[14] = a34 * oodet;	ret.m[15] = a44 * oodet;	return ret;}OpenGL should be multiplying the matrices correctly, and I am inverting the model view matrix as instructed.Any ideas why its not working?

##### Share on other sites
Try this, if it isn't working, I won't reply any more:

static struct Matrix4x4 proj, modl, clip;    static struct Plane * p;    glGetFloatv(GL_PROJECTION_MATRIX, &proj.m[0]);    glGetFloatv(GL_MODELVIEW_MATRIX, &modl.m[0]);    clip = multiplyMatrix4x4(proj, modl);    clip = getInvertedMatrix4x4(clip);....

And try to use own multiplication code with correct indexing (as mentioned).

##### Share on other sites
With using your matrix multiplication code and inverting the clip matrix **everything** (including off-screen objects) is now rendered.

    clip = multiplyMatrix4x4(proj, modl);    clip = getInvertedMatrix4x4(clip);

/** thanks to szecs at GameDev.net **/FASTCALL struct Matrix4x4 multiplyMatrix4x4(struct Matrix4x4 a, struct Matrix4x4 b){    struct Matrix4x4 c;    c.m[0]  = a.m[0] * b.m[0]  + a.m[4] * b.m[1]  + a.m[8]  * b.m[2]  + a.m[12] * b.m[3];    c.m[1]  = a.m[1] * b.m[0]  + a.m[5] * b.m[1]  + a.m[9]  * b.m[2]  + a.m[13] * b.m[3];    c.m[2]  = a.m[2] * b.m[0]  + a.m[6] * b.m[1]  + a.m[10] * b.m[2]  + a.m[14] * b.m[3];    c.m[3]  = a.m[3] * b.m[0]  + a.m[7] * b.m[1]  + a.m[11] * b.m[2]  + a.m[15] * b.m[3];    c.m[4]  = a.m[0] * b.m[4]  + a.m[4] * b.m[5]  + a.m[8]  * b.m[6]  + a.m[12] * b.m[7];    c.m[5]  = a.m[1] * b.m[4]  + a.m[5] * b.m[5]  + a.m[9]  * b.m[6]  + a.m[13] * b.m[7];    c.m[6]  = a.m[2] * b.m[4]  + a.m[6] * b.m[5]  + a.m[10] * b.m[6]  + a.m[14] * b.m[7];    c.m[7]  = a.m[3] * b.m[4]  + a.m[7] * b.m[5]  + a.m[11] * b.m[6]  + a.m[15] * b.m[7];    c.m[8]  = a.m[0] * b.m[8]  + a.m[4] * b.m[9]  + a.m[8]  * b.m[10] + a.m[12] * b.m[11];    c.m[9]  = a.m[1] * b.m[8]  + a.m[5] * b.m[9]  + a.m[9]  * b.m[10] + a.m[13] * b.m[11];    c.m[10] = a.m[2] * b.m[8]  + a.m[6] * b.m[9]  + a.m[10] * b.m[10] + a.m[14] * b.m[11];    c.m[11] = a.m[3] * b.m[8]  + a.m[7] * b.m[9]  + a.m[11] * b.m[10] + a.m[15] * b.m[11];    c.m[12] = a.m[0] * b.m[12] + a.m[4] * b.m[13] + a.m[8]  * b.m[14] + a.m[12] * b.m[15];    c.m[13] = a.m[1] * b.m[12] + a.m[5] * b.m[13] + a.m[9]  * b.m[14] + a.m[13] * b.m[15];    c.m[14] = a.m[2] * b.m[12] + a.m[6] * b.m[13] + a.m[10] * b.m[14] + a.m[14] * b.m[15];    c.m[15] = a.m[3] * b.m[12] + a.m[7] * b.m[13] + a.m[11] * b.m[14] + a.m[15] * b.m[15];}

*sigh*. Any ideas?

##### Share on other sites
Turns out the matrix multiplication code is the problem, still working to fix it however.

The clip matrix is being returned as the following values:
{{1.#QNAN0, 1.#QNAN0, 1.#QNAN0, 1.#QNAN0} {1.#QNAN0, 1.#QNAN0, 1.#QNAN0, 1.#QNAN0} {1.#QNAN0, 1.#QNAN0, 1.#QNAN0, 1.#QNAN0} {1.#QNAN0, 1.#QNAN0, 1.#QNAN0, 1.#QNAN0}}

I have since re-written the matrix multiplication code:
FASTCALL struct Matrix4x4 multiplyMatrix4x4(struct Matrix4x4 a, struct Matrix4x4 b){    struct Matrix4x4 c;    s32 i, j, k;    for (i = 0; i < 4; i++)        for (j = 0; j < 4; j++)            for (k = 0; k < 4; k++)                c.m[i * 4 + j] += a.m[i * 4 + k] * b.m[k * 4 + j];    return c;}

This new multiplyMatrix4x4() function does not give weird values.

Now the inverted clip matrix is being returned as the following values:
{{0.000000, -1.#IND00, 0.000000, 0.000000} {0.000000, -1.#IND00, 0.000000, 0.000000} {0.000000, 0.000000, 0.000000, 0.000000} {-1.#IND00, -1.#IND00, 0.000000, -1.#IND00}}

Any ideas?

##### Share on other sites
Just a quick comment on your matrix multiplication code: you're not zeroing out c before you perform the multiplication, which means that the result may or may not be correct. (This may not be causing you any problems currently - for example, it may be that the memory just happens to be zeroed already for whatever reason - but it's something that should be fixed nonetheless.)

##### Share on other sites
I don't know. I thought functions cannot return arrays, just single values or pointers.
So he uses arrays that's really just pointers, that had been deallocated after returning from the functions. But it worked in some parts of his code for some reason, so I don't know.
Should use output params as pointers in the param. list. (void functions of course)

• 11
• 11
• 9
• 12
• 10
• ### Similar Content

• By EddieK
Hello. I'm trying to make an android game and I have come across a problem. I want to draw different map layers at different Z depths so that some of the tiles are drawn above the player while others are drawn under him. But there's an issue where the pixels with alpha drawn above the player. This is the code i'm using:
int setup(){ GLES20.glEnable(GLES20.GL_DEPTH_TEST); GLES20.glEnable(GL10.GL_ALPHA_TEST); GLES20.glEnable(GLES20.GL_TEXTURE_2D); } int render(){ GLES20.glClearColor(0, 0, 0, 0); GLES20.glClear(GLES20.GL_ALPHA_BITS); GLES20.glClear(GLES20.GL_COLOR_BUFFER_BIT); GLES20.glClear(GLES20.GL_DEPTH_BUFFER_BIT); GLES20.glBlendFunc(GLES20.GL_ONE, GL10.GL_ONE_MINUS_SRC_ALPHA); // do the binding of textures and drawing vertices } My vertex shader:
uniform mat4 MVPMatrix; // model-view-projection matrix uniform mat4 projectionMatrix; attribute vec4 position; attribute vec2 textureCoords; attribute vec4 color; attribute vec3 normal; varying vec4 outColor; varying vec2 outTexCoords; varying vec3 outNormal; void main() { outNormal = normal; outTexCoords = textureCoords; outColor = color; gl_Position = MVPMatrix * position; } My fragment shader:
precision highp float; uniform sampler2D texture; varying vec4 outColor; varying vec2 outTexCoords; varying vec3 outNormal; void main() { vec4 color = texture2D(texture, outTexCoords) * outColor; gl_FragColor = vec4(color.r,color.g,color.b,color.a);//color.a); } I have attached a picture of how it looks. You can see the black squares near the tree. These squares should be transparent as they are in the png image:

Its strange that in this picture instead of alpha or just black color it displays the grass texture beneath the player and the tree:

Any ideas on how to fix this?

• This article uses material originally posted on Diligent Graphics web site.
Introduction
Graphics APIs have come a long way from small set of basic commands allowing limited control of configurable stages of early 3D accelerators to very low-level programming interfaces exposing almost every aspect of the underlying graphics hardware. Next-generation APIs, Direct3D12 by Microsoft and Vulkan by Khronos are relatively new and have only started getting widespread adoption and support from hardware vendors, while Direct3D11 and OpenGL are still considered industry standard. New APIs can provide substantial performance and functional improvements, but may not be supported by older hardware. An application targeting wide range of platforms needs to support Direct3D11 and OpenGL. New APIs will not give any advantage when used with old paradigms. It is totally possible to add Direct3D12 support to an existing renderer by implementing Direct3D11 interface through Direct3D12, but this will give zero benefits. Instead, new approaches and rendering architectures that leverage flexibility provided by the next-generation APIs are expected to be developed.
There are at least four APIs (Direct3D11, Direct3D12, OpenGL/GLES, Vulkan, plus Apple's Metal for iOS and osX platforms) that a cross-platform 3D application may need to support. Writing separate code paths for all APIs is clearly not an option for any real-world application and the need for a cross-platform graphics abstraction layer is evident. The following is the list of requirements that I believe such layer needs to satisfy:
Lightweight abstractions: the API should be as close to the underlying native APIs as possible to allow an application leverage all available low-level functionality. In many cases this requirement is difficult to achieve because specific features exposed by different APIs may vary considerably. Low performance overhead: the abstraction layer needs to be efficient from performance point of view. If it introduces considerable amount of overhead, there is no point in using it. Convenience: the API needs to be convenient to use. It needs to assist developers in achieving their goals not limiting their control of the graphics hardware. Multithreading: ability to efficiently parallelize work is in the core of Direct3D12 and Vulkan and one of the main selling points of the new APIs. Support for multithreading in a cross-platform layer is a must. Extensibility: no matter how well the API is designed, it still introduces some level of abstraction. In some cases the most efficient way to implement certain functionality is to directly use native API. The abstraction layer needs to provide seamless interoperability with the underlying native APIs to provide a way for the app to add features that may be missing. Diligent Engine is designed to solve these problems. Its main goal is to take advantages of the next-generation APIs such as Direct3D12 and Vulkan, but at the same time provide support for older platforms via Direct3D11, OpenGL and OpenGLES. Diligent Engine exposes common C++ front-end for all supported platforms and provides interoperability with underlying native APIs. It also supports integration with Unity and is designed to be used as graphics subsystem in a standalone game engine, Unity native plugin or any other 3D application. Full source code is available for download at GitHub and is free to use.
Overview
Diligent Engine API takes some features from Direct3D11 and Direct3D12 as well as introduces new concepts to hide certain platform-specific details and make the system easy to use. It contains the following main components:
Render device (IRenderDevice  interface) is responsible for creating all other objects (textures, buffers, shaders, pipeline states, etc.).
Device context (IDeviceContext interface) is the main interface for recording rendering commands. Similar to Direct3D11, there are immediate context and deferred contexts (which in Direct3D11 implementation map directly to the corresponding context types). Immediate context combines command queue and command list recording functionality. It records commands and submits the command list for execution when it contains sufficient number of commands. Deferred contexts are designed to only record command lists that can be submitted for execution through the immediate context.
An alternative way to design the API would be to expose command queue and command lists directly. This approach however does not map well to Direct3D11 and OpenGL. Besides, some functionality (such as dynamic descriptor allocation) can be much more efficiently implemented when it is known that a command list is recorded by a certain deferred context from some thread.
The approach taken in the engine does not limit scalability as the application is expected to create one deferred context per thread, and internally every deferred context records a command list in lock-free fashion. At the same time this approach maps well to older APIs.
In current implementation, only one immediate context that uses default graphics command queue is created. To support multiple GPUs or multiple command queue types (compute, copy, etc.), it is natural to have one immediate contexts per queue. Cross-context synchronization utilities will be necessary.
Swap Chain (ISwapChain interface). Swap chain interface represents a chain of back buffers and is responsible for showing the final rendered image on the screen.
Render device, device contexts and swap chain are created during the engine initialization.
Resources (ITexture and IBuffer interfaces). There are two types of resources - textures and buffers. There are many different texture types (2D textures, 3D textures, texture array, cubmepas, etc.) that can all be represented by ITexture interface.
Resources Views (ITextureView and IBufferView interfaces). While textures and buffers are mere data containers, texture views and buffer views describe how the data should be interpreted. For instance, a 2D texture can be used as a render target for rendering commands or as a shader resource.
Pipeline State (IPipelineState interface). GPU pipeline contains many configurable stages (depth-stencil, rasterizer and blend states, different shader stage, etc.). Direct3D11 uses coarse-grain objects to set all stage parameters at once (for instance, a rasterizer object encompasses all rasterizer attributes), while OpenGL contains myriad functions to fine-grain control every individual attribute of every stage. Both methods do not map very well to modern graphics hardware that combines all states into one monolithic state under the hood. Direct3D12 directly exposes pipeline state object in the API, and Diligent Engine uses the same approach.
Shader Resource Binding (IShaderResourceBinding interface). Shaders are programs that run on the GPU. Shaders may access various resources (textures and buffers), and setting correspondence between shader variables and actual resources is called resource binding. Resource binding implementation varies considerably between different API. Diligent Engine introduces a new object called shader resource binding that encompasses all resources needed by all shaders in a certain pipeline state.
API Basics
Creating Resources
Device resources are created by the render device. The two main resource types are buffers, which represent linear memory, and textures, which use memory layouts optimized for fast filtering. Graphics APIs usually have a native object that represents linear buffer. Diligent Engine uses IBuffer interface as an abstraction for a native buffer. To create a buffer, one needs to populate BufferDesc structure and call IRenderDevice::CreateBuffer() method as in the following example:
BufferDesc BuffDesc; BufferDesc.Name = "Uniform buffer"; BuffDesc.BindFlags = BIND_UNIFORM_BUFFER; BuffDesc.Usage = USAGE_DYNAMIC; BuffDesc.uiSizeInBytes = sizeof(ShaderConstants); BuffDesc.CPUAccessFlags = CPU_ACCESS_WRITE; m_pDevice->CreateBuffer( BuffDesc, BufferData(), &m_pConstantBuffer ); While there is usually just one buffer object, different APIs use very different approaches to represent textures. For instance, in Direct3D11, there are ID3D11Texture1D, ID3D11Texture2D, and ID3D11Texture3D objects. In OpenGL, there is individual object for every texture dimension (1D, 2D, 3D, Cube), which may be a texture array, which may also be multisampled (i.e. GL_TEXTURE_2D_MULTISAMPLE_ARRAY). As a result there are nine different GL texture types that Diligent Engine may create under the hood. In Direct3D12, there is only one resource interface. Diligent Engine hides all these details in ITexture interface. There is only one  IRenderDevice::CreateTexture() method that is capable of creating all texture types. Dimension, format, array size and all other parameters are specified by the members of the TextureDesc structure:
TextureDesc TexDesc; TexDesc.Name = "My texture 2D"; TexDesc.Type = TEXTURE_TYPE_2D; TexDesc.Width = 1024; TexDesc.Height = 1024; TexDesc.Format = TEX_FORMAT_RGBA8_UNORM; TexDesc.Usage = USAGE_DEFAULT; TexDesc.BindFlags = BIND_SHADER_RESOURCE | BIND_RENDER_TARGET | BIND_UNORDERED_ACCESS; TexDesc.Name = "Sample 2D Texture"; m_pRenderDevice->CreateTexture( TexDesc, TextureData(), &m_pTestTex ); If native API supports multithreaded resource creation, textures and buffers can be created by multiple threads simultaneously.
Interoperability with native API provides access to the native buffer/texture objects and also allows creating Diligent Engine objects from native handles. It allows applications seamlessly integrate native API-specific code with Diligent Engine.
Next-generation APIs allow fine level-control over how resources are allocated. Diligent Engine does not currently expose this functionality, but it can be added by implementing IResourceAllocator interface that encapsulates specifics of resource allocation and providing this interface to CreateBuffer() or CreateTexture() methods. If null is provided, default allocator should be used.
Initializing the Pipeline State
As it was mentioned earlier, Diligent Engine follows next-gen APIs to configure the graphics/compute pipeline. One big Pipelines State Object (PSO) encompasses all required states (all shader stages, input layout description, depth stencil, rasterizer and blend state descriptions etc.). This approach maps directly to Direct3D12/Vulkan, but is also beneficial for older APIs as it eliminates pipeline misconfiguration errors. With many individual calls tweaking various GPU pipeline settings it is very easy to forget to set one of the states or assume the stage is already properly configured when in fact it is not. Using pipeline state object helps avoid these problems as all stages are configured at once.
While in earlier APIs shaders were bound separately, in the next-generation APIs as well as in Diligent Engine shaders are part of the pipeline state object. The biggest challenge when authoring shaders is that Direct3D and OpenGL/Vulkan use different shader languages (while Apple uses yet another language in their Metal API). Maintaining two versions of every shader is not an option for real applications and Diligent Engine implements shader source code converter that allows shaders authored in HLSL to be translated to GLSL. To create a shader, one needs to populate ShaderCreationAttribs structure. SourceLanguage member of this structure tells the system which language the shader is authored in:
When sampling a texture in a shader, the texture sampler was traditionally specified as separate object that was bound to the pipeline at run time or set as part of the texture object itself. However, in most cases it is known beforehand what kind of sampler will be used in the shader. Next-generation APIs expose new type of sampler called static sampler that can be initialized directly in the pipeline state. Diligent Engine exposes this functionality: when creating a shader, textures can be assigned static samplers. If static sampler is assigned, it will always be used instead of the one initialized in the texture shader resource view. To initialize static samplers, prepare an array of StaticSamplerDesc structures and initialize StaticSamplers and NumStaticSamplers members. Static samplers are more efficient and it is highly recommended to use them whenever possible. On older APIs, static samplers are emulated via generic sampler objects.
The following is an example of shader initialization:
Creating the Pipeline State Object
After all required shaders are created, the rest of the fields of the PipelineStateDesc structure provide depth-stencil, rasterizer, and blend state descriptions, the number and format of render targets, input layout format, etc. For instance, rasterizer state can be described as follows:
PipelineStateDesc PSODesc; RasterizerStateDesc &RasterizerDesc = PSODesc.GraphicsPipeline.RasterizerDesc; RasterizerDesc.FillMode = FILL_MODE_SOLID; RasterizerDesc.CullMode = CULL_MODE_NONE; RasterizerDesc.FrontCounterClockwise = True; RasterizerDesc.ScissorEnable = True; RasterizerDesc.AntialiasedLineEnable = False; Depth-stencil and blend states are defined in a similar fashion.
Another important thing that pipeline state object encompasses is the input layout description that defines how inputs to the vertex shader, which is the very first shader stage, should be read from the memory. Input layout may define several vertex streams that contain values of different formats and sizes:
// Define input layout InputLayoutDesc &Layout = PSODesc.GraphicsPipeline.InputLayout; LayoutElement TextLayoutElems[] = {     LayoutElement( 0, 0, 3, VT_FLOAT32, False ),     LayoutElement( 1, 0, 4, VT_UINT8, True ),     LayoutElement( 2, 0, 2, VT_FLOAT32, False ), }; Layout.LayoutElements = TextLayoutElems; Layout.NumElements = _countof( TextLayoutElems ); Finally, pipeline state defines primitive topology type. When all required members are initialized, a pipeline state object can be created by IRenderDevice::CreatePipelineState() method:
// Define shader and primitive topology PSODesc.GraphicsPipeline.PrimitiveTopologyType = PRIMITIVE_TOPOLOGY_TYPE_TRIANGLE; PSODesc.GraphicsPipeline.pVS = pVertexShader; PSODesc.GraphicsPipeline.pPS = pPixelShader; PSODesc.Name = "My pipeline state"; m_pDev->CreatePipelineState(PSODesc, &m_pPSO); When PSO object is bound to the pipeline, the engine invokes all API-specific commands to set all states specified by the object. In case of Direct3D12 this maps directly to setting the D3D12 PSO object. In case of Direct3D11, this involves setting individual state objects (such as rasterizer and blend states), shaders, input layout etc. In case of OpenGL, this requires a number of fine-grain state tweaking calls. Diligent Engine keeps track of currently bound states and only calls functions to update these states that have actually changed.
Direct3D11 and OpenGL utilize fine-grain resource binding models, where an application binds individual buffers and textures to certain shader or program resource binding slots. Direct3D12 uses a very different approach, where resource descriptors are grouped into tables, and an application can bind all resources in the table at once by setting the table in the command list. Resource binding model in Diligent Engine is designed to leverage this new method. It introduces a new object called shader resource binding that encapsulates all resource bindings required for all shaders in a certain pipeline state. It also introduces the classification of shader variables based on the frequency of expected change that helps the engine group them into tables under the hood:
Static variables (SHADER_VARIABLE_TYPE_STATIC) are variables that are expected to be set only once. They may not be changed once a resource is bound to the variable. Such variables are intended to hold global constants such as camera attributes or global light attributes constant buffers. Mutable variables (SHADER_VARIABLE_TYPE_MUTABLE) define resources that are expected to change on a per-material frequency. Examples may include diffuse textures, normal maps etc. Dynamic variables (SHADER_VARIABLE_TYPE_DYNAMIC) are expected to change frequently and randomly. Shader variable type must be specified during shader creation by populating an array of ShaderVariableDesc structures and initializing ShaderCreationAttribs::Desc::VariableDesc and ShaderCreationAttribs::Desc::NumVariables members (see example of shader creation above).
Static variables cannot be changed once a resource is bound to the variable. They are bound directly to the shader object. For instance, a shadow map texture is not expected to change after it is created, so it can be bound directly to the shader:
m_pPSO->CreateShaderResourceBinding(&m_pSRB); Note that an SRB is only compatible with the pipeline state it was created from. SRB object inherits all static bindings from shaders in the pipeline, but is not allowed to change them.
Mutable resources can only be set once for every instance of a shader resource binding. Such resources are intended to define specific material properties. For instance, a diffuse texture for a specific material is not expected to change once the material is defined and can be set right after the SRB object has been created:
m_pSRB->GetVariable(SHADER_TYPE_PIXEL, "tex2DDiffuse")->Set(pDiffuseTexSRV); In some cases it is necessary to bind a new resource to a variable every time a draw command is invoked. Such variables should be labeled as dynamic, which will allow setting them multiple times through the same SRB object:
m_pSRB->GetVariable(SHADER_TYPE_VERTEX, "cbRandomAttribs")->Set(pRandomAttrsCB); Under the hood, the engine pre-allocates descriptor tables for static and mutable resources when an SRB objcet is created. Space for dynamic resources is dynamically allocated at run time. Static and mutable resources are thus more efficient and should be used whenever possible.
As you can see, Diligent Engine does not expose low-level details of how resources are bound to shader variables. One reason for this is that these details are very different for various APIs. The other reason is that using low-level binding methods is extremely error-prone: it is very easy to forget to bind some resource, or bind incorrect resource such as bind a buffer to the variable that is in fact a texture, especially during shader development when everything changes fast. Diligent Engine instead relies on shader reflection system to automatically query the list of all shader variables. Grouping variables based on three types mentioned above allows the engine to create optimized layout and take heavy lifting of matching resources to API-specific resource location, register or descriptor in the table.
This post gives more details about the resource binding model in Diligent Engine.
Setting the Pipeline State and Committing Shader Resources
Before any draw or compute command can be invoked, the pipeline state needs to be bound to the context:
m_pContext->SetPipelineState(m_pPSO); Under the hood, the engine sets the internal PSO object in the command list or calls all the required native API functions to properly configure all pipeline stages.
The next step is to bind all required shader resources to the GPU pipeline, which is accomplished by IDeviceContext::CommitShaderResources() method:
m_pContext->CommitShaderResources(m_pSRB, COMMIT_SHADER_RESOURCES_FLAG_TRANSITION_RESOURCES); The method takes a pointer to the shader resource binding object and makes all resources the object holds available for the shaders. In the case of D3D12, this only requires setting appropriate descriptor tables in the command list. For older APIs, this typically requires setting all resources individually.
Next-generation APIs require the application to track the state of every resource and explicitly inform the system about all state transitions. For instance, if a texture was used as render target before, while the next draw command is going to use it as shader resource, a transition barrier needs to be executed. Diligent Engine does the heavy lifting of state tracking.  When CommitShaderResources() method is called with COMMIT_SHADER_RESOURCES_FLAG_TRANSITION_RESOURCES flag, the engine commits and transitions resources to correct states at the same time. Note that transitioning resources does introduce some overhead. The engine tracks state of every resource and it will not issue the barrier if the state is already correct. But checking resource state is an overhead that can sometimes be avoided. The engine provides IDeviceContext::TransitionShaderResources() method that only transitions resources:
m_pContext->TransitionShaderResources(m_pPSO, m_pSRB); In some scenarios it is more efficient to transition resources once and then only commit them.
Invoking Draw Command
The final step is to set states that are not part of the PSO, such as render targets, vertex and index buffers. Diligent Engine uses Direct3D11-syle API that is translated to other native API calls under the hood:
ITextureView *pRTVs[] = {m_pRTV}; m_pContext->SetRenderTargets(_countof( pRTVs ), pRTVs, m_pDSV); // Clear render target and depth buffer const float zero[4] = {0, 0, 0, 0}; m_pContext->ClearRenderTarget(nullptr, zero); m_pContext->ClearDepthStencil(nullptr, CLEAR_DEPTH_FLAG, 1.f); // Set vertex and index buffers IBuffer *buffer[] = {m_pVertexBuffer}; Uint32 offsets[] = {0}; Uint32 strides[] = {sizeof(MyVertex)}; m_pContext->SetVertexBuffers(0, 1, buffer, strides, offsets, SET_VERTEX_BUFFERS_FLAG_RESET); m_pContext->SetIndexBuffer(m_pIndexBuffer, 0); Different native APIs use various set of function to execute draw commands depending on command details (if the command is indexed, instanced or both, what offsets in the source buffers are used etc.). For instance, there are 5 draw commands in Direct3D11 and more than 9 commands in OpenGL with something like glDrawElementsInstancedBaseVertexBaseInstance not uncommon. Diligent Engine hides all details with single IDeviceContext::Draw() method that takes takes DrawAttribs structure as an argument. The structure members define all attributes required to perform the command (primitive topology, number of vertices or indices, if draw call is indexed or not, if draw call is instanced or not, if draw call is indirect or not, etc.). For example:
DrawAttribs attrs; attrs.IsIndexed = true; attrs.IndexType = VT_UINT16; attrs.NumIndices = 36; attrs.Topology = PRIMITIVE_TOPOLOGY_TRIANGLE_LIST; pContext->Draw(attrs); For compute commands, there is IDeviceContext::DispatchCompute() method that takes DispatchComputeAttribs structure that defines compute grid dimension.
Source Code
Full engine source code is available on GitHub and is free to use. The repository contains two samples, asteroids performance benchmark and example Unity project that uses Diligent Engine in native plugin.
AntTweakBar sample is Diligent Engine’s “Hello World” example.

Atmospheric scattering sample is a more advanced example. It demonstrates how Diligent Engine can be used to implement various rendering tasks: loading textures from files, using complex shaders, rendering to multiple render targets, using compute shaders and unordered access views, etc.

Asteroids performance benchmark is based on this demo developed by Intel. It renders 50,000 unique textured asteroids and allows comparing performance of Direct3D11 and Direct3D12 implementations. Every asteroid is a combination of one of 1000 unique meshes and one of 10 unique textures.

Finally, there is an example project that shows how Diligent Engine can be integrated with Unity.

Future Work
The engine is under active development. It currently supports Windows desktop, Universal Windows and Android platforms. Direct3D11, Direct3D12, OpenGL/GLES backends are now feature complete. Vulkan backend is coming next, and support for more platforms is planned.
• By reenigne
For those that don't know me. I am the individual who's two videos are listed here under setup for https://wiki.libsdl.org/Tutorials
I also run grhmedia.com where I host the projects and code for the tutorials I have online.
Recently, I received a notice from youtube they will be implementing their new policy in protecting video content as of which I won't be monetized till I meat there required number of viewers and views each month.

Frankly, I'm pretty sick of youtube. I put up a video and someone else learns from it and puts up another video and because of the way youtube does their placement they end up with more views.
Even guys that clearly post false information such as one individual who said GLEW 2.0 was broken because he didn't know how to compile it. He in short didn't know how to modify the script he used because he didn't understand make files and how the requirements of the compiler and library changes needed some different flags.

At the end of the month when they implement this I will take down the content and host on my own server purely and it will be a paid system and or patreon.

I get my videos may be a bit dry, I generally figure people are there to learn how to do something and I rather not waste their time.
I used to also help people for free even those coming from the other videos. That won't be the case any more. I used to just take anyone emails and work with them my email is posted on the site.

I don't expect to get the required number of subscribers in that time or increased views. Even if I did well it wouldn't take care of each reoccurring month.
I figure this is simpler and I don't plan on putting some sort of exorbitant fee for a monthly subscription or the like.
I was thinking on the lines of a few dollars 1,2, and 3 and the larger subscription gets you assistance with the content in the tutorials if needed that month.
Maybe another fee if it is related but not directly in the content.
The fees would serve to cut down on the number of people who ask for help and maybe encourage some of the people to actually pay attention to what is said rather than do their own thing. That actually turns out to be 90% of the issues. I spent 6 hours helping one individual last week I must have asked him 20 times did you do exactly like I said in the video even pointed directly to the section. When he finally sent me a copy of the what he entered I knew then and there he had not. I circled it and I pointed out that wasn't what I said to do in the video. I didn't tell him what was wrong and how I knew that way he would go back and actually follow what it said to do. He then reported it worked. Yea, no kidding following directions works. But hey isn't alone and well its part of the learning process.

So the point of this isn't to be a gripe session. I'm just looking for a bit of feed back. Do you think the fees are unreasonable?
Should I keep the youtube channel and do just the fees with patreon or do you think locking the content to my site and require a subscription is an idea.

I'm just looking at the fact it is unrealistic to think youtube/google will actually get stuff right or that youtube viewers will actually bother to start looking for more accurate videos.

• i got error 1282 in my code.