User Interface for Guardian High Scores and Challenges

Here is my attempt at using XNA to duplicate some of the UI functionality on WP7. Well, not really duplicate so much as “create a passing resemblance to”.

The menu is pretty boring. I’ve been trying to come up with a good idea to make it interesting, and think I finally have something that will be worth the effort. That’ll probably be the next movie.

The first WP7 phones are supposed to be released in the U.S. on November 8th – just one week away. Here’s hoping Guardian works well on the hardware without needing too many optimizations.

My name is Crappy Coding Guy, and I use Texture2D.GetData

In a previous post about texture modification, I mentioned the evils of transferring data from the GPU to the CPU, and then presented an example showing one way to avoid doing it. The post wasn’t really about deformable 2D terrain or collision detection, but was intended to help newer game programmers open up a new way of thinking when it comes to using the GPU to accomplish tasks.

Since that post, and the one showing a video of my WP7 game, I’ve received a couple of questions about how I do the collision detection in Guardian, which would seem to require the use of Texture2D.GetData.

In a previous post about texture modification, I mentioned the evils of transferring data from the GPU to the CPU, and then presented an example showing one way to avoid doing it.  The post wasn’t really about deformable 2D terrain or collision detection, but was intended to help newer game programmers open up a new way of thinking when it comes to using the GPU to accomplish tasks.

Since that post, and the one showing a video of my WP7 game, I’ve received a couple of questions about how I do the collision detection in Guardian, which would seem to require the use of Texture2D.GetData.

As it turns out, I am evil, and I do use GetData.  But, my evilness is optimized based on information from here, here, and here.

  • Crater drawing is batched, meaning that rather than draw each one as it’s created, I add them to a list and draw all of them every few frames. This reduces the number of GetData calls – one per batch of craters rather than one per crater.
  • After drawing craters to the render target, I wait a few frames before calling GetData to make sure the GPU has processed all of the drawing commands. This minimizes pipeline stalls.
  • If I have a pending GetData call to make and more craters come in, the craters will stay batched until the GetData call is complete.  In other words, the drawing and getting are synchronized so that a GetData call always happens several frames after drawing a batch of craters, and any new crater draw requests wait until after a pending GetData.

If there are a lot of craters being created the built-in delays can cause some slightly inaccurate collision detection since we may be looking at collision data that’s outdated by several frames. At least in this particular game there are never huge numbers of crater adds going on so this isn’t a problem. If there are more than several crater adds they tend to be bunched close together, so the explosion animation hides any visual oddities.

There is one other optimization that I have available but haven’t needed to use.  The collision data doesn’t need to be at the same resolution as the drawing data.  Basically have two sets of render targets – one for the visual texture, and a lower resolution set for the collision data.  Do the GetData on the collision texture and scale everything appropriately when doing the collision check. You have to draw twice – once for the visual data and once for the collision data – but you’re pulling much less data from the GPU which would possibly offset the extra drawing time (this isn’t something I’ve tested yet). You won’t be pixel perfect, but for this type of game that isn’t necessary. As I write this it seems using multiple render targets would eliminate the “draw twice” issue here, but I’ve never done that so some research would be required.

So there you have it. Is this the best or most efficient way?  I don’t know – I’m far from an expert on any of this. To be honest, I never actually tested doing this just on the CPU, so it’s entirely possible that that approach is better if there are collision detection requirements. There are also other considerations, such as whether your game is CPU or GPU bound, which would go into determining which method is better suited to your needs. Ultimately, whatever works in your situation is the right method.

Texture Modification in XNA 3.1

I’ve had a couple of questions about what changes are needed to get the texture modification tutorial to work in XNA 3.1.

So, here’s a 3.1 version of the project, and a quick overview of the major things that need to change.

  • You need to create the depth/stencil buffer yourself, set it on the GraphicsDevice when setting the render target, and restore the previous buffer when you’re done.
  • RenderTarget2D can’t be used directly as a texture, you must call RenderTarget2D.GetTexture instead and use that when drawing.
  • Render states are all set in GraphicsDevice.RenderState instead of the various classes used in 4.0.
  • Various minor syntax changes.

Texture Modification using Render Targets, with some Stencil Buffer Action

Sometimes you need to modify a texture while your game is running, and there are a number of ways to do this. One of the first things newer game programmers often try to do is use Texture2D.GetData to copy the texture data from the GPU to an array on the CPU, modify the bytes, and then send it back to the GPU with Texture2D.SetData.

This is a bad idea on many, levels. Beyond issues with pipeline stalls, GetData and SetData can be slow, especially when working with a large texture. Any time you’re tempted grab data from the GPU for use on the CPU you should very carefully consider all of your options. There are often other solutions that let you keep the data entirely on the GPU and accomplish the same thing.

This tutorial will use an example that could be solved with GetData and SetData, and show you another alternative using render targets and the stencil buffer that will let you perform the same function entirely on the GPU.

Sometimes you need to modify a texture while your game is running, and there are a number of ways to do this. One of the first things newer game programmers often try to do is use Texture2D.GetData to copy the texture data from the GPU to an array on the CPU, modify the bytes, and then send it back to the GPU with Texture2D.SetData.

This is a bad idea on many levels. Beyond issues with pipeline stalls, GetData and SetData can be slow, especially when working with a large texture. Any time you’re tempted grab data from the GPU for use on the CPU you should very carefully consider all of your options. There are often other solutions that let you keep the data entirely on the GPU and accomplish the same thing.

This tutorial will use an example that could be solved with GetData and SetData, and show you another alternative using render targets and the stencil buffer that will let you perform the same function entirely on the GPU.

CPU Craters

Let’s pretend you want to draw 2D planet, and periodically add a crater to it. You want a hole to appear somewhere on the planet, so it looks like part of it was removed.

You could do this using the GetData/SetData method by getting the data from a texture into an array, setting the color to the background (or alpha to 0) in the shape of the crater, then writing the data back to the texture. Or you could be a little cleverer and eliminate GetData by always keeping the data in the array, but you still have to do the SetData to get it into the texture on the GPU each time it’s changed.

GPU Craters

The method we’ll use to do this entirely on the GPU involves several steps. First, we need a couple of resources. We’ll use a simple textured circle for a planet, and a crater shaped texture for the crater.

It’s important to note that the black areas on these have an alpha value of 0, meaning completely transparent. For the planet this just lets us draw the round shape over the background without looking like a square image. But for the crater image the alpha value is very important since it will control what part of the crater image is removed from the planet.

Next, we need to set up two render targets (these will be referred to later as Render Target A, and Render Target B). When we need to add a crater, one of these will be used as a target for drawing to, while the other used as a texture. The next time we add a crater they will swap roles – the texture will become the target, and the target will become the texture. This is called “ping-ponging” and will be discussed more fully later.

Once we have these resources ready to go, the method for adding a crater goes like this:

  1. Activate Render Target A using GraphicsDevice.SetRenderTarget.
  2. Clear the graphics device, setting the color to solid black, and the stencil buffer to 0.
  3. Set up the stencil buffer state so whatever we draw writes a value of 1 to the stencil buffer.
  4. Set up the alpha test state so we only draw where the alpha value is zero.
  5. Draw the crater texture. Because of the way we’ve set up the graphics device, only the parts of the crater texture that have alpha = 0 will be drawn, and those parts will write a 1 to the stencil buffer. So what we have at this point is a “mask” in the stencil buffer that we can use in the next step. The white area in the following image represents the stencil mask we’ve set up – the stencil buffer contains “1” in the white area, and “0” everywhere else.
  6. Set up the stencil buffer so when we draw, anything that has a value of 1 in the stencil buffer will be masked out – meaning it won’t draw.
  7. Draw the “planet texture”. Because of the way we’ve set up the graphics device, anything with a 1 in the stencil buffer won’t be drawn – since these 1’s are in the shape of a crater, that shape will be masked out of the planet texture, leaving holes that look like craters.
  8. Set the render target to the backbuffer. We can now access Render Target A as a texture, and that texture contains the planet texture with a crater-shaped hole in it.
Step 5
Step 7

From now on, until we need to add another crater, we can treat Render Target A as a texture and draw it using SpriteBatch, and we’ll have a nice crater. Now, what if we need to add another crater? This is where the ping-ponging comes in. Since Render Target A is now the “planet texture”, we need to be able to draw somewhere else when we’re filling in the stencil buffer with our crater shape. It just so happens that we set up another place to draw to, Render Target B.

So now, in Step 1, instead of activating Render Target A we need to activate Render Target B and draw the crater shapes into that. But what happens when we get to Step 7? Well, the “planet texture” is now in Render Target A, so we draw that. And in Step 8, Render Target B now contains our new planet texture with two craters.

And if we add a third crater then we’re back to where we started – drawing to Render Target A, and using Render Target B as the source texture. In other words, we “ping-pong” between the two render targets – each time we need to modify the texture, one is used for a texture, and one is used for drawing to, and then those roles are swapped.

You may have noticed that there’s one issue here. The first time through, Render Target B has nothing in it, so we can’t use it as the planet texture. This can be handled by using the actual planet texture the first time, and the render target thereafter.

The Code

Now let’s walk through the code involved, using XNA 4.0. You can do this in 3.1, but you’ll have to make significant changes when creating the render targets and setting the render states.

The complete code is in the downloadable project linked at the end of the tutorial. We’ll just go through the highlights here, referring to the steps mentioned above as we go.

The XNA 4.0 API has been changed substantially where render states are concerned, and for the better. Render states have been grouped by functionality into several classes. You create instances of these classes to represent the state you want, then set them on the graphics device, or pass them to SpriteBatch. So first we need to create these render state objects.

Set Up Render State Objects

For Step 3, we need to use the DepthStencilState class to set up the device to always set the stencil buffer to 1. We enable the stencil buffer, set the stencil function to Always, the pass operation to Replace, and ReferenceStencil to 1. This means that as we’re drawing, each pixel will Always pass, and the value in the stencil buffer will be Replaced with 1.

stencilAlways = new DepthStencilState();
stencilAlways.StencilEnable = true;
stencilAlways.StencilFunction = CompareFunction.Always;
stencilAlways.StencilPass = StencilOperation.Replace;
stencilAlways.ReferenceStencil = 1;
stencilAlways.DepthBufferEnable = false;

And for Step 4 we need to use the standard AlphaTestEffect so we can draw the asteroid texture only where the alpha value is 0.

Matrix projection = Matrix.CreateOrthographicOffCenter(0, PlanetDataSize, PlanetDataSize, 0, 0, 1);
Matrix halfPixelOffset = Matrix.CreateTranslation(-0.5f, -0.5f, 0);
alphaTestEffect = new AlphaTestEffect(GraphicsDevice);
alphaTestEffect.VertexColorEnabled = true;
alphaTestEffect.DiffuseColor = Color.White.ToVector3();
alphaTestEffect.AlphaFunction = CompareFunction.Equal;
alphaTestEffect.ReferenceAlpha = 0;
lphaTestEffect.World = Matrix.Identity;
alphaTestEffect.View = Matrix.Identity;
alphaTestEffect.Projection = halfPixelOffset * projection;

We first set up an orthographic projection matrix that matches SpriteBatch. We set AlphaFunction to Equal, and ReferenceAlpha to 0. This means the alpha test will pass whenever the alpha value we’re drawing is equal to 0. In our crater texture, the crater area has an alpha value of 0, while the surrounding area has 1, so only the crater area will be drawn.

For Step 6 we need a stencil buffer state that allows drawing only where the stencil buffer contains a 0. We enable the stencil buffer, set the stencil function to Equal, the pass operation to Keep, and the reference stencil to 0. This means that when we’re drawing, each pixel will pass if the value in the stencil buffer is Equal to 0.

stencilKeepIfZero = new DepthStencilState();
stencilKeepIfZero.StencilEnable = true;
stencilKeepIfZero.StencilFunction = CompareFunction.Equal;
stencilKeepIfZero.StencilPass = StencilOperation.Keep;
stencilKeepIfZero.ReferenceStencil = 0;
stencilKeepIfZero.DepthBufferEnable = false;

Create Render Targets

Now that we have the render state objects created, it’s time to create the render targets. Both are the same, so just one is shown here. This creates a render target with a Color format, and a depth format that includes a stencil buffer.

renderTargetA = new RenderTarget2D(GraphicsDevice, PlanetDataSize, 
  PlanetDataSize, false, SurfaceFormat.Color, 
  DepthFormat.Depth24Stencil8, 0, 
  RenderTargetUsage.DiscardContents);

Draw the Crater Mask

Next up is drawing the crater masks (Steps 2-5). First we activate the render target, clear it to solid black, and clear the stencil buffer to 0.

GraphicsDevice.SetRenderTarget(activeRenderTarget);
GraphicsDevice.Clear(ClearOptions.Target | ClearOptions.Stencil,
                     new Color(0, 0, 0, 1), 0, 0);

Next we begin a SpriteBatch, passing in the stencilAlways and alphaTestEffect objects that we created earlier. Calculate some random rotation, size the crater texture using a Rectangle, and call SpriteBatch.Draw to draw the crater.

spriteBatch.Begin(SpriteSortMode.Immediate, BlendState.Opaque,
                  null, stencilAlways, null, alphaTestEffect);
Vector2 origin = new Vector2(craterTexture.Width * 0.5f,
                             craterTexture.Height * 0.5f);
float rotation = (float)random.NextDouble() * MathHelper.TwoPi;
Rectangle r = new Rectangle((int)position.X, (int)position.Y, 50, 50);

spriteBatch.Draw(craterTexture, r, null, Color.White, rotation,
                 origin, SpriteEffects.None, 0);

spriteBatch.End();

Draw the Planet Texture

Now we need to draw the latest planet texture, using the stencil buffer to mask out the craters (Steps 6-7). We begin a SpriteBatch, passing in the stencilKeepIfZero object we created earlier. Note that the first time we draw the actual planet texture, but subsequently we draw using the texture from the previous iteration.

spriteBatch.Begin(SpriteSortMode.Immediate, BlendState.Opaque,
                  null, stencilKeepIfZero, null, null);

if (firstTime)
{
  spriteBatch.Draw(planetTexture, Vector2.Zero, Color.White);
  firstTime = false;
}
else
  spriteBatch.Draw(textureRenderTarget, Vector2.Zero, Color.White);

spriteBatch.End();

Swap Render Targets

Finally we activate the backbuffer render target.

GraphicsDevice.SetRenderTarget(null);

And then swap the render targets as discussed previously.

RenderTarget2D t = activeRenderTarget;
activeRenderTarget = textureRenderTarget;
textureRenderTarget = t;

In the main Draw function, you draw the latest cratered planet using the textureRenderTarget. Of course, you need to deal with using the planet texture the first time through though. The downloadable code shows one simple way to do that.

GraphicsDevice.Clear(Color.CornflowerBlue);
spriteBatch.Begin();
spriteBatch.Draw(textureRenderTarget, planetPosition, Color.White);
spriteBatch.End();

Conclusion

And there you have it, a powerful technique for altering textures during your game. Doing this entirely on the GPU is quite a bit more complex than GetData/SetData, but is well worth the extra trouble.

There are some things you can do to improve this technique. If you need to add a lot of craters, rather than adding them one at a time you can batch them up for a while, then in Step 5 draw all of them at once.

I hope you found this tutorial informative. Learning about render targets and stencil buffers opens up a whole new world of possibilities beyond just making craters. What other uses can you think of?

Download the sample XNA 4.0 project

Download the sample XNA 3.1 project

Procedural Planet Video

Procedural Planet video showing the space-to-surface transition, as well as the tank driving around a bit. I’m pretty happy with how it looks, but there are a lot of things that still need to be fixed.

Here’s a video showing the space-to-surface transition, as well as the tank driving around a bit. I’m pretty happy with how it looks, but there are a lot of things that still need to be fixed:

  • Seams on neighboring nodes when they’re at a different level of detail.
  • Lighting on node edges is wrong since the normal map generation isn’t currently taking into account neighboring height values, resulting in very visible node edges.
  • Need to add some noise to the texture generation so the transitions are less obvious. This should also help minimize the repeating patterns.
  • Normals are too “strong” at high altitudes. I think I need to change the normal strength based on the quad tree level.
  • Need to add morphing between LOD changes to eliminate popping.

I think those changes will improve things quite a bit. But first I’m going to plug the atmospheric scattering shader back in. I’m currently using Sean O’Neil’s method, and that’s what I’m going to add back for now. But I’m not really happy with it and plan to give Precomputed Atmospheric Scattering a go.

So, here’s the video…

Of Tanks and Quad Trees

I needed a bit of a diversion from the planet rendering itself, into something that would give some purpose behind it. Why is the planet there? Well, what better use is there for a planet than driving a tank on it?

I needed a bit of a diversion from the planet rendering itself, into something that would give some purpose behind it. Why is the planet there? Well, what better use is there for a planet than driving a tank on it?

The tank model from the XNA Heightmap Collision with Normals sample is the perfect tank to drive on a lonely, empty, and possibly dangerous new world. I grabbed the model and texture resources, and the Tank class, added them to my project, and added some “spawn tank” functionality to get it drawing, resulting in a beautiful flying tank that seemed to defy gravity. Since nobody should be able to defy gravity, it was time to bring the tank down to earth.

The basic concept is to find the triangle underneath the tank and the exact position within that triangle. This will give us the exact height at that location, and the normal so we can orient the tank correctly. When using a single height map that’s not too bad, but when there are potentially hundreds of height maps, organized in a quad tree, there’s some more work involved.

The first thing I do is find the quad tree node that’s under the tank. There isn’t a regular grid like with a height map, so I had to come up with a different way to find it. The first thing that came to mind was to do traverse the quadtree with a simple point-in-bounding-box check, but that would only work if the tank is already very close to the ground. I want to be able to have a position 6,000 km above the surface and be able to find the proper node.

The next idea was to create a view frustum from the planet center looking outward at each node. To create the frustum I’d have to calculate the proper field of view so the frustum planes would go through the edges of the quad tree nodes. I actually tried this method but had trouble getting the field of view calculations to work so things were aligned properly. I still think this method would work if I spent enough time on it, but there’s actually quite a bit of code that has to be executed for each check, so it’s not the best method anyway.

The final idea, and the one I have working, is to do a simple ray-bounding-box intersection test. I was initially going to create the ray at the planet center, pointing out towards the position we’re looking at, but that would very likely run into floating point precision issues, so instead the ray starts 1 km below the planet surface, which gives plenty of precision to work with.

So, I traverse the quad tree, doing the ray-bounding-box intersection test. If I find a hit, I perform the check against the children, and so on, until I reach a leaf node. Once I have a leaf node I need to find the triangle within the leaf, using ray-triangle intersection tests.  Now, I understand the concept, but coding one with my current knowledge is beyond me, so I downloaded the XNA triangle picking example, which has a nice ray-triangle intersection method.

Now, the original ray-bounding-box intersection can give false hits since we’re working with a spherical surface because the bounding boxes can overlap a bit. So the ray-triangle intersection tests might fail to find a triangle. If that happens I just continue on traversing the tree with the bounding box checks until I eventually find the proper node, as well as the proper triangle. This works, 99% of the time. There is still a bug where the code fails to find a node at all. It’s very rare, but something I’ll need to figure out someday.

So, in order to get the proper height value I need to get the exact point within the triangle. When using a height map it’s common to use bi-linear interpolation to find this height, and I spent some time trying to get that to work with partial success. I finally stepped back and realized that the ray-triangle intersect was returning the distance along the ray of the intersect, so it was a simple matter of multiplying the normalized ray-direction by the intersect distance to give me the exact position I needed. Treating that position as a vector from the planet center,  the magnitude is the height value that’s needed for the tank position.

That leaves the normal so the tank can be oriented properly. When using a height map the normal is also found using bi-linear interpolation. This presented a problem since I couldn’t ever get that to work completely for the height. Instead of mucking with it too much I chose to average the normals of the triangle, which seems to fit well within my “good enough” expectations at this time.

So, my planet has gained a purpose, and in so doing I now have some code that’s going to be very useful for things like placing trees and walking and crashing and shooting. I’ll try to put up a video in the next day or so.

GPU Geometry Map Rendering – Part 2

We left off in part 1 talking about the initial failures with my GPU geometry map shader. I did fail to mention that there was a bright spot the first time I ran the new code – it was amazingly fast. So fast I was able to increase the noise octaves from the 5 that would run reasonably well on the CPU up to 30 and still run at well over 60fps. I have to admit that I spent some of that first 18 hour day just roaming around on a barren, reddish planet. That huge improvement in performance made the pain to come well worth it.
So, at the end of part 1 we set up the C# code for executing the geometry map shader. Now let’s take a look at the shader itself.

We left off in part 1 talking about the initial failures with my GPU geometry map shader. I did fail to mention that there was a bright spot the first time I ran the new code – it was amazingly fast. So fast I was able to increase the noise octaves from the 5 that would run reasonably well on the CPU up to 30 and still run at well over 60fps. I have to admit that I spent some of that first 18 hour day just roaming around on a barren, reddish planet. That huge improvement in performance made the pain to come well worth it.

So, at the end of part 1 we set up the C# code for executing the geometry map shader. Now let’s take a look at the shader itself.

float Left;
float Top;
float Width;
float Height;

struct Vertex
{
  float4 Position : Position;
  float2 UV : TexCoord0;
};

struct TransformedVertex
{
  float4 Position : Position;
  float2 UV : TexCoord0;
};

void QuadVertexShader(in Vertex input, out TransformedVertex output)
{
  output.Position = input.Position;
  output.UV = input.UV;
}

float4 QuadPixelShader(in TransformedVertex input) : COLOR0
{
  // grab the texture coordinates - you can use them directly, but doing this
  // lets you see the value in PIX more easily when you're debugging
  float2 uv = input.UV;

  // get the coordinates relative to the input dimensions
  float x = Width * uv.x;
  float y = Height * uv.y;

  // translate the final coordinates
  return float4(Left + x, y - Top, 1, 1);
}

The Left, Top, Width, and Height floats at the top are the parameters used to define the face-space area. These are the values you’re setting when using the XNA Effect class as mentioned in part 1.

quadEffect.Parameters["Left"].SetValue(-1.0f);
quadEffect.Parameters["Top"].SetValue(1.0f);
quadEffect.Parameters["Width"].SetValue(2.0f);
quadEffect.Parameters["Height"].SetValue(2.0f);

The Vertex struct defines the vertices entering the vertex shader, and the TransformedVertex defines the vertices leaving the vertex shader and entering the pixel shader. Since we’re using pre-transformed vertices the vertex shader doesn’t need to do anything but pass the input values through, which it does very nicely.

The pixel shader isn’t actually all that much more complex. The GPU takes the texture coordinates specified in each vertex and interpolates them for us as it’s rasterizing our quad. Let’s think again in just the horizontal dimension. As mentioned previously, we’ve set up the texture coordinates so they start at 0.0 on the left vertex, and 1.0 on the right vertex. The intent of the shader is to map those values to face-space. In this example the 0.0 should map to -1.0, and the right to +1.0. Also, remember that we’re working with a 5×5 geometry map, and these are the face-space values we expect to get for each of the 5 horizontal positions: -1.0, -0.5, 0.0, 0.5, 1.0.

Walking through the shader for the left pixel we expect to see this:

The uv.x value the shader receives is 0.0
x = Width * uv.x = 2.0 * 0.0 = 0.0
Left + x = -1.0 + 0.0 = -1.0

And for the right pixel we (well, at least I did at one time) expect to see this:

The uv.x value the shader receives is 1.0
x = Width * uv.x = 2.0 * 1.0 = 2.0
Left + x = -1.0 + 2.0 = 1.0

But that isn’t what happens. In fact, the 5 texture coordinates we get are 0.0, 0.2, 0.4, 0.8, resulting in face-space values of: -1.0, -0.6, -0.2, +0.2, +0.8. That did not make any sense to me at all – shouldn’t the texture coordinates be interpolated from 0.0 to 1.0? I spent hours sifting through my code to find out what I had set up wrong. I even took the step of creating a separate bare minimum test app, which I really hate to do. Everything I tried gave me the same results, and I was forced to conclude that somehow the GPU must not be interpolating the texture coordinates the way I expected. I suspected that it had something to do with how Direct X maps texels to pixels, but nothing I tried in that regard gave me the results I needed. I finally caved and started up PIX.

It was a bit daunting at first, but there are some simple tutorials out there that will walk you through the basics. I’m going to go through how I ended up using it, in order to easily run the same tests over, and over, and over again. Start by downloading the test project.

Unzip it into your project folder, open the QuadTest solution, rebuild it, and set QuadTestBad as the startup project. Run it, and verify that you get a pretty box with various shades of blue, white, and magenta.

Now start up PIX. Select File/New Experiment. Use the browse button to navigate to the QuadTestBad.exe you just built. Don’t change anything else for now and press the Start Experiment button. The app should start and you should see the pretty box, as well as some PIX overlay infromation in the upper left. If that’s the case, all is well. If not, you’ll need to figure out why on your own.

Now what we need to do is have PIX grab all the Direct3D data for a single frame. Often you can do this by selecting the “Single-frame capture of Direct3D whenever F12 is pressed” radio button before starting the experiment. That works just fine, and is the method I started out with. But there’s a better way in this case. From the experiment window (the one where you set up the program path), select the More Options button. You should see something like this:

The left tree view might have some different values in it if you changed any options on the initial screen, but the things you need to do here are the same regardless. In the tree view the green T lines are Triggers, and the purple A lines are Actions. When selecting a trigger line you’ll see the right panel which will let you select the type of trigger. Select “Frame” for the trigger type. This will reveal the options for the Frame trigger type. Enter “1” for the frame number. Now select the Action line, which will change the right panel to allow you to select an action. Select the “Set Call Capture” action. Then under Capture Type select “Single-frame capture Direct 3D”, and check the “capture d3dx calls also” box as well.

We’ve just defined a trigger based action. We’ve told PIX to capture Direct3D data on frame #1. Now let’s create a second trigger based action. Press the green T button to create the new trigger. Again select Frame for the trigger type, but this time enter 2 for the frame number. For the action select Terminate Program. So, we’ve now told PIX to exit the program on frame #2. To restate, on frame 1 PIX will capture a single frame of data, and then on frame 2 it will exit the program. You can just go the F12 route in many cases, but when you’re going to be repeating something over and over it can save a lot of time setting up the triggers. In some cases you may have to set the triggers since it can be impossible to hit F12 at just the right moment to capture the data you want.

So, go a head and start the experiment. You should see the QuadTestBad app start up, and then immediately quit. PIX should then display a screen full of data. For this discussion we’re only interested in the panels on the bottom. Events and Details. Events will show you all of the Direct3D calls, and Details will show you details about those events. There is a lot of noise in the D3D calls, so it can be difficult to find what you’re looking for by just scanning through it. There are some buttons at the top of the Events panel that let you move to the next or previous frame, and the next or previous draw call. So, press the button that has the D with a down arrow:

That will take us to the draw call we’re interested in examining. In the Detail panel select the Render tab. You should see the quad we drew. This view will let us step through the vertex and pixel shaders for each pixel displayed in the render tab. You can also look at the Mesh tab to see information at the vertex level, both pre-vertex-shader and post-vertex-shader. Go ahead and do that now, and you’ll see that the screen-space positions and texture coordinates we set up for the full-screen quad did indeed arrive at the vertex shader correctly, and they were correctly sent on to the pixel shader.

Let’s debug a pixel. Go back to the Render tab, mouse over the upper left corner of the image until the X and Y values displayed in the status area show 0, 0. You can zoom into the image if necessary using the buttons at the top of the panel. Once you have the mouse over the top left pixel, right click and select “Debug This Pixel” to open up the Debugger tab. This tab shows a history of the pixel for this frame. Scroll down a bit until you see the DrawPrimitive call. There are several links displayed for debugging the vertices, as well as the pixel. Let’s start with one of the vertices just so you can see it. Click the Debug Vertex 0 link to bring up the vertex shader debugger. Here you can step through the vertex shader code (both forward and backward) and examine all of the variables and registers and such that are involved. Press F10 to step through each line. You’ll see variable values added to the list at the bottom as the values change, or look at the Registers tab to see the individual register values. You can also switch to the Disassembly tab to see the assembly code, which contains comments to help match up registers to variable names in the HLSL code.

To get back to the initial debugger screen press the “back” toolbar button – it’s the one with the green circle and white arrow pointing left.

Now let’s do the fun part and debug the pixel shader. Click the Debug Pixel (0, 0) link,which takes you to the pixel shader debugger. You may have noticed that the pixel shader could be made much more efficient by changing things to use vectors instead of individual floats, and using the incoming texture coordinates directly rather than copying them into a local variable. If you noticed this, you would be right. But splitting things out this way makes debugging in PIX a lot easier, at least for me. You’ll notice that the pixel shader debugger doesn’t display the value for input.UV anywhere, and there is also no way to add a “watch” like you would do in Visual Studio. You could look at the Registers tab and get a good idea, but that can involve a lot of thinking and writing, and examining the assembly code to determine which register is mapped to which variable. So, I found that it helped a great deal to break things out like this because the debugger adds everything to the variable list as you make changes.

If you execute the first line, float2 uv = input.UV, you’ll see what I mean. The uv variable is added to the list and shows the current value, which is (0, 1). Now, if you debug all 5 of the top row of pixels you’ll see that the uv.x values for the pixels are 0.0, 0.2, 0.4, 0.6, 0.8. Why doesn’t it ever reach 1.0? Honestly, I still think it has something to do with texel to pixel mapping, and tried I don’t know how many combinations of shifting things around by half pixels in various coordinate systems, but I still have no idea why it doesn’t make it all the way to 1.0. I’m hoping someone will read this and let me know.

I do know how it’s coming up with those values though. It starts uv.x out at 0, and adds 1.0 / GeometryMapWidth for the next pixel. In our test case GeometryMapWidth is 5, so it’s adding 0.2 each time. If I could make the GPU add 0.25 each time I’d be in business. What I’d like to do is have the GPU add 1.0 / (GeometryMapWidth – 1) each time, but I can’t change the divisor – the GPU is always going find the step by dividing by GeometryWidth. But what if I could change is the numerator? Accessing some little used and rusty Algebra skills I came up with this:

n / GeometryMapWidth = 0.25
n = GeometryMapWidth * 0.25

Our GeometryMapWidth is 5, so n = 1.25. But how do we make the GPU use that as the numerator? Well, as it turns out the numerator in the 1.0 / GeometryMapWidth formula isn’t always 1.0. It’s really the width defined by the texture coordinates from the left and right vertices (in the horizontal case we’re considering). So far the rightmost texture value has been 1.0, and the left has been 0.0. So the formula becomes (1.0 – 0.0) / GeometryMapWidth. If the left coordinate is something besides 0, for example 0.13, the formula would look like (1.0 – 0.13) / GeometryMapWidth.

So, using that knowledge, we can change the numerator to whatever we want by manipulating the texture coordinates. Since the left value is 0.0 and needs to stay that way, we can change the right value to 1.25. The GPU will then calculate the step value as (1.25 – 0.00) / GeometryMapWidth, or 1.25 / 5, which is the 0.25 value we’re looking for! So now this is what our full screen quad definition looks like:

float pw = 1.0f / (Width - 1);
float ph = 1.0f / (Height - 1);

vertices = new VertexPositionTexture[4];
vertices[0] = new VertexPositionTexture(
  new Vector3(-1, 1, 0f), new Vector2(0, 1));
vertices[1] = new VertexPositionTexture(
  new Vector3(1, 1, 0f), new Vector2(1 + pw, 1));
vertices[2] = new VertexPositionTexture(
  new Vector3(-1, -1, 0f), new Vector2(0, 0 - ph));
vertices[3] = new VertexPositionTexture(
  new Vector3(1, -1, 0f), new Vector2(1 + pw, 0 - ph));

This generalizes the approach a bit. Instead of hard coding the 0.25 value we can calculate what it needs to be based on the size of the geometry map. We then add that value to 1 to get the final value, for the horizontal dimension. For the vertical dimension we’re actually subtracting it from the bottom coordinate due to the relationship between the different coordinate systems.

So, in the sample code, set QuadTestGood as the Startup Project and run it through PIX. Debug each pixel and you’ll see that the texture coordinates are interpolated like we want them to be.

One final piece of the puzzle though. When we generate the geometry map we need to generate an extra border of vertices around it for use in calculating the vertex normals. We can do this by simply expanding the texture coordinates in each direction by the value we calculated in the previous step.

// if we have a border then expand the texture 
// coordinates out to account for it
if (border)
{
  vertices[0].TextureCoordinate.X -= pw;
  vertices[0].TextureCoordinate.Y += ph;

  vertices[1].TextureCoordinate.X += pw;
  vertices[1].TextureCoordinate.Y += ph;

  vertices[2].TextureCoordinate.X -= pw;
  vertices[2].TextureCoordinate.Y -= ph;

  vertices[3].TextureCoordinate.X += pw;
  vertices[3].TextureCoordinate.Y -= ph;
}

If you want you can run the QuadTestGoodWithBorder project through PIX and verify that it works as well.

Now that we have the right texture coordinates, the rest of the shader just scales the coordinate by the face-space width and height to calculate the correct x and y values to pass to the noise functions. The sample shader currently just returns those values as the color. You’d want to change the render target to floating point, and replace this line:

// translate the final coordinates
return float4(Left + x, y - Top, 1, 1);

with this:

return TerrainNoise(float3(Left + x, y - Top, 1));

And create your TerrainNoise function using any of the myriad methods revealed by Google.

So, I guess that’s it. I can’t say I entirely enjoyed this ride, but the destination was worth it. And if anyone wants to explain to me why texture coordinates don’t want to interpolate all the way to 1 on a full screen quad, please feel free. 🙂

Procedural Planet Engine Status

Previously I mentioned I was going to do a mulligan on my procedural planet engine. The few hours I’ve worked on it so far have lead to a beautiful new architecture that’s doing most of the same things as before, as well as some major new things, using about 25% of the code.

Previously I mentioned I was going to do a mulligan on my procedural planet engine. The few hours I’ve worked on it so far have lead to a beautiful new architecture that’s doing most of the same things as before, as well as some major new things, using about 25% of the code.

Here is where things stand currently. I’ll go through some of these in more detail in a later post:

The planet consists of a cube, with the vertices mapped to a sphere. Each of the six cube faces is a quad tree which is used for subdividing the terrain as you move closer to the planet. Each node in the quad tree represents a patch of terrain with 33×33 vertices that are spread out evenly to cover the patch’s area.

In the previous version the quad tree nodes were subdivided synchronously, which resulted in jerkiness when moving slowly, and outright 5 second waits when moving quickly if a lot of nodes needed to be subdivided. That was good enough then since my priorities were elsewhere, but it’s not good enough for the new version. Now, when a node needs to be split the request is queued on a separate thread. The current node will continue to draw until the split is complete. The split requests can be cancelled as well if the camera has moved elsewhere before the split request reaches the head of the queue.

The nice thing about this design is that if you’re moving very fast you end up getting fewer node splits because they’re cancelled before they happen since they’re no longer necessary. Conversely, if you’re moving slowly the splits can easily keep up with your location so you get all of the required detail. On the con side, if you’re moving quickly down to a low level, then stop, it can take a bit for the queue to catch up generating the terrain patches, so the detail can take awhile to show up.

Generating a patch currently happens on the CPU using Perlin as a noise basis and various fractal algorithms such as fBm, Turbulence, and Ridged Multifractal. I will be moving this to the GPU over the coming weeks which will vastly improve the “catching up” problem mentioned previously. This will also enable creating procedural normal maps and textures on the fly.

So, the current version of the app lets me start out in space and fly to an Earth-sized planet down to ground level with ever increasing detail, and absolutely no stalling. The entire planet can be explored, but there is no texturing yet, and lighting is using vertex normals so it’s fairly ugly, but it gets the job done at this stage.

I think the next thing I will do is work on moving the patch generation to the GPU. This seemed like a daunting task 8 months ago, but it should be pretty straightforward now. This is a requirement to allow generating higher resolution procedural normal maps, which will be a big step in improving the look of the terrain.

So, that’s it for now. In future posts I’ll go through some of these features in more detail and discuss how I did things.

Sprite Sheet Creator

When developing the iPhone version of Guardian I manually created my sprite sheets. I used individual sprites up until the end so everything was pretty much set in stone by the time I created the the sprite sheet. Even then I ended up having to recreate the sprite sheet two or three times, and let me tell you, manually figuring out the texture coordinates isn’t a particularly pleasant experience. In this case I believe I made the right choice. There were few enough sprites that I would have spent more time creating the tool than I would have saved.

When developing the iPhone version of Guardian I manually created my sprite sheets. I used individual sprites up until the end so everything was pretty much set in stone by the time I created the the sprite sheet. Even then I ended up having to recreate the sprite sheet two or three times, and let me tell you, manually figuring out the texture coordinates isn’t a particularly pleasant experience. In this case I believe I made the right choice. There were few enough sprites that I would have spent more time creating the tool than I would have saved.

The XBox version has quite a few more sprites, so I decided that spending time creating a sprite sheet tool was going to be well worth the effort. It didn’t take too long to get it working well enough to use, and not too much longer than that to make it solid enough for distribution.

Sprite Sheet Creator

The application is released as open source under the MIT License.

Download SpriteSheetCreator.zip