如何用C#分别控制立体声帧？(NVIDIA 3D快门眼镜)

Question

我正在尝试制作一个非常简单的应用程序,它会在每只眼睛上显示不同的图像.我有华硕VG236H显示器和NVIDIA 3D Vision套件,立体3D快门眼镜.我正在使用C#,.NET Framework 2.0,DirectX 9(Managed Direct X)和Visual Studio 2008.我一直在寻找示例和教程的高低,实际上找到了一对,并且基于我创建的程序但是由于某种原因,我不能让它工作.

在寻找如何为每只眼睛显示不同图像的示例时,许多人不断参考GDC 09(着名的GDC09-3DVision-The_In_and_Out.pdf文档)和第37-40页的NVIDIA演示文稿.我的代码主要是基于这个例子构建的:

1. 我在表面上加载了两个纹理(Red.png和Blue.png)(_imageLeft和_imageRight),在函数LoadSurfaces()中
2. Set3D()函数将这两个图像并排放在一个较大的图像上,该图像的大小为2x屏幕宽度和屏幕高度+ 1(_imageBuf).
3. 通过在最后一行附加立体声签名继续Set3D()函数.
4. OnPaint() - 函数获取后台缓冲区(_backBuf)并将组合图像(_imageBuf)的内容复制到它.

当我运行程序时,快门眼镜开始工作,但我只能在屏幕上并排看到这两个图像.有人可以帮忙告诉我我做错了什么吗？我相信解决这个问题也可能有助于其他人,因为似乎还没有一个简单的例子如何用C#做到这一点.

以下是我的代码的战术部分.完整的项目可以在这里下载:http://koti.mbnet.fi/jjantti2/NVStereoTest.rar

    public void InitializeDevice()
    {
        PresentParameters presentParams = new PresentParameters();

        presentParams.Windowed = false;
        presentParams.BackBufferFormat = Format.A8R8G8B8;
        presentParams.BackBufferWidth = _size.Width;
        presentParams.BackBufferHeight = _size.Height;
        presentParams.BackBufferCount = 1;
        presentParams.SwapEffect = SwapEffect.Discard;
        presentParams.PresentationInterval = PresentInterval.One;
        _device = new Device(0, DeviceType.Hardware, this, CreateFlags.SoftwareVertexProcessing, presentParams);
    }

    public void LoadSurfaces()
    {
        _imageBuf = _device.CreateOffscreenPlainSurface(_size.Width * 2, _size.Height + 1, Format.A8R8G8B8, Pool.Default);

        _imageLeft = Surface.FromBitmap(_device, (Bitmap)Bitmap.FromFile("Blue.png"), Pool.Default);
        _imageRight = Surface.FromBitmap(_device, (Bitmap)Bitmap.FromFile("Red.png"), Pool.Default);
    }

    private void Set3D()
    {
        Rectangle destRect = new Rectangle(0, 0, _size.Width, _size.Height);
        _device.StretchRectangle(_imageLeft, _size, _imageBuf, destRect, TextureFilter.None);
        destRect.X = _size.Width;
        _device.StretchRectangle(_imageRight, _size, _imageBuf, destRect, TextureFilter.None);

        GraphicsStream gStream = _imageBuf.LockRectangle(LockFlags.None);

        byte[] data = new byte[] {0x44, 0x33, 0x56, 0x4e,   //NVSTEREO_IMAGE_SIGNATURE         = 0x4433564e
                                  0x00, 0x00, 0x0F, 0x00,   //Screen width * 2 = 1920*2 = 3840 = 0x00000F00;
                                  0x00, 0x00, 0x04, 0x38,   //Screen height = 1080             = 0x00000438;
                                  0x00, 0x00, 0x00, 0x20,   //dwBPP = 32                       = 0x00000020;
                                  0x00, 0x00, 0x00, 0x02};  //dwFlags = SIH_SCALE_TO_FIT       = 0x00000002;

        gStream.Seek(_size.Width * 2 * _size.Height * 4, System.IO.SeekOrigin.Begin);   //last row
        gStream.Write(data, 0, data.Length);

        gStream.Close();

        _imageBuf.UnlockRectangle();
    }

    protected override void OnPaint(System.Windows.Forms.PaintEventArgs e)
    {
        _device.BeginScene();

        // Get the Backbuffer then Stretch the Surface on it.
        _backBuf = _device.GetBackBuffer(0, 0, BackBufferType.Mono);
        _device.StretchRectangle(_imageBuf, new Rectangle(0, 0, _size.Width * 2, _size.Height + 1), _backBuf, new Rectangle(0, 0, _size.Width, _size.Height), TextureFilter.None);
        _backBuf.ReleaseGraphics();

        _device.EndScene();

        _device.Present();

        this.Invalidate();
    }

Answer 1

我的一个朋友发现了这个问题。立体声签名中的字节顺序相反。这是正确的顺序：

byte[] data = new byte[] {0x4e, 0x56, 0x33, 0x44,   //NVSTEREO_IMAGE_SIGNATURE         = 0x4433564e;
0x00, 0x0F, 0x00, 0x00,   //Screen width * 2 = 1920*2 = 3840 = 0x00000F00;
0x38, 0x04, 0x00, 0x00,   //Screen height = 1080             = 0x00000438;
0x20, 0x00, 0x00, 0x00,   //dwBPP = 32                       = 0x00000020;
0x02, 0x00, 0x00, 0x00};  //dwFlags = SIH_SCALE_TO_FIT       = 0x00000002;

进行此更改后，代码可以完美运行。对于其他尝试同样事情的人来说，该代码甚至可以作为一个很好的教程。:)