D3D11 CreateTexture2D in 32 bit format

Question

Surprisingly I am not able to find any help on the subject, even though I am pretty certain this is quite a basic question. So maybe somebody here could help me to progress on that.

I am trying to create a d3d11 texture, mono-channel, 32 bit per pixel. The official documentation is interesting ( http://msdn.microsoft.com/en-us/library/windows/desktop/ff476521(v=vs.85).aspx) and I found this example for a four-channel 8 bits per pixel:

ID3D11Texture2D *MakeCheckerboard(ID3D11Device *myDevice)
{
    ID3D11Texture2D *tex;
    D3D11_TEXTURE2D_DESC tdesc;
    D3D11_SUBRESOURCE_DATA tbsd;

    int w = 512;
    int h = 512;
    int bpp = 4;
    int *buf = new int[w*h];

    // filling the image
    for(int i=0;iCreateTexture2D(&tdesc, &tbsd, NULL) == S_FALSE )
        std::cout << "Inputs correct" << std::endl;
    else
        std::cout << "wrong inputs" << std::endl;

    // create the texture
    if(FAILED(myDevice->CreateTexture2D(&tdesc,&tbsd,&tex)))
    {
        std::cout << "Failed" << std::endl;
        return(0);
    }
    else
        std::cout << "Success" << std::endl;

    delete[] buf;

    return(tex);
}

This exemple works fine, so the output is:

Inputs correct
Success

Then, if I modify it a little for a 32 bits monochannel:

ID3D11Texture2D *MakeCheckerboard(ID3D11Device *myDevice)
{
    ID3D11Texture2D *tex;
    D3D11_TEXTURE2D_DESC tdesc;
    D3D11_SUBRESOURCE_DATA tbsd;

    int w = 512;
    int h = 512;
    int bpp = 4;
    int *buf = new int[w*h];

    for(int i=0;iCreateTexture2D(&tdesc, &tbsd, NULL) == S_FALSE )
        std::cout << "Inputs correct" << std::endl;
    else
        std::cout << "wrong inputs" << std::endl;

    if(FAILED(myDevice->CreateTexture2D(&tdesc,&tbsd,&tex)))
    {
        std::cout << "Failed" << std::endl;
        return(0);
    }
    else
        std::cout << "Sucess" << std::endl;

    delete[] buf;

    return(tex);
}

The corresponding output becomes

wrong inputs
Failed

So I guess inputs are wrong, but they seems good to me and the documentation is not very verbose on that. Does somebody has a clue ?

Edit - after Chuck Walbourn answer

I see three points here:

1. The debug layer:

Thanks for the advice, I read the link you gave, and also your answser on this question here, and this doc (4), and the last paragraph from here (5):

msdn.microsoft.com/en-us/library/windows/desktop/jj200584(v=vs.85).aspx

msdn.microsoft.com/en-us/library/windows/desktop/jj863687(v=vs.85).aspx

(cannot post links because of low reput)

For developers currently working on applications in Microsoft Visual Studio 2010 or earlier using the D3D11_CREATE_DEVICE_DEBUG flag, be aware that calls to D3D11CreateDevice will fail. This is because the D3D11.1 runtime now requires D3D11_1SDKLayers.dll instead of D3D11SDKLayers.dll.

I have to develop using VS 2010, and if I activate the flag D3D11_CREATE_DEVICE_DEBUG, the call for D3D11CreateDevice() is not creating any error. But there is no debug msg during runtime. So I did a research on my PC, and the dll D3D11_1SDKLayers.dll is missing. Thus I assume I will have to install the Windows SDK 8.1.

This raises two questions I will try to answer:

Can I install Windows 8.1 SDK on a Windows 7 PC?

Where does the debug msg shall appear? In my console during runtime? (silly question I guess, but I have to confess I am not a real programmer and things apparently trivia for you guys are not for me I guess...)

EDIT: According to

The legacy DirectX SDK (June 2010) has the correct DirectX Debug Runtime for Windows Vista SP2 and Windows 7 aka DirectX 11.0

I should be able to use the debug layer. But activating the flag is not making the exe more verbose. I am digging further. (I am using mvs 2010 with DirectX SDK June 2010, on windows 7, with DirectX 11 acording to dxdiag)

EDIT: solution OKAY, so indeed the output msg are visible in the "output" panel of msv, NOT in the console! THIS is a big step forward ^^. Thus this point is solved and closed.

2. The DXGI_FORMAT_R32_FLOAT format:

I did not really understand what you mean by "initData". I guess you mean that I am creating the texture and filling it in the same time, whereas I could use an empty subresource data and use UpdateSubresource(). But anyway I tried using the DXGI_FORMAT_R32_FLOAT. It seemed to work great as I did not had any errors, but the texture is black, or there is no texture. So I guess it's not working properly and I will need to work on that. I hope using the debug layer will help. Even though I am quite surprised because again, I have no error at runtime.

EDIT: As the debug layer is now OK, I am now certain that the texture is correctly created using Chuck Walbourn advice. GOOD! But another problem remains : the texture is either completely black, or completely red. Digging deeper.

EDIT: solution The problem is simple: each pixel value has to a float between 0.0 and 1.0. That's all. Problem solved. But then the image is red...

3. The Direct3D feature level above 9.2:

I don't know. This is starting to be complete alien language to me. The PC is using a NVIDIA Quadro K4000 and I will just secretly assume this is enough. I hope...

EDIT: solution Apparently it is, so it's okay.

4. Thought of a hopeless mind:

I am not a core programmer, I just want to use a mono-channel 32 bits texture instead of a four-channel 32 bits, seriously, why is that so complex ?

NoonanRosenblum · Accepted Answer

I simply found that the DXGI_FORMAT_R32G32B32_FLOAT exists. So I am creating a r32g32b32 texture and, as I need a b&w image, I use the same level on each colour. Pretty simple.

Here is the declination of the example in my question for this solution:

ID3D11Texture2D *MakeCheckerboard(ID3D11Device *myDevice)
{
    ID3D11Texture2D*       tex;
    D3D11_TEXTURE2D_DESC   tdesc;
    D3D11_SUBRESOURCE_DATA tbsd;

    int width    = 3;   // three pixel wide
    int height   = 1;   // one pixel height
    int bpp      = 12;
    int nb_color = 3;

    // CREATING THE IMAGE
    float* buf[ width * height * nb_color ];

    // pixel white
    buf[0] = 1.0f; // red
    buf[1] = 1.0f; // green
    buf[2] = 1.0f; // blue

    // pixel black
    buf[3] = 0.0f;
    buf[4] = 0.0f;
    buf[5] = 0.0f;

    // pixel white
    buf[6] = 1.0f;
    buf[7] = 1.0f;
    buf[8] = 1.0f;

    tbsd.pSysMem          = (void *) buf;
    tbsd.SysMemPitch      = width * bpp ;
    tbsd.SysMemSlicePitch = width * height * bpp ; // Not needed since this is a 2d texture

    tdesc.Width              = width;
    tdesc.Height             = height;
    tdesc.MipLevels          = 1;
    tdesc.ArraySize          = 1;
    tdesc.SampleDesc.Count   = 1;
    tdesc.SampleDesc.Quality = 0;
    tdesc.Usage              = D3D11_USAGE_DEFAULT;
    tdesc.Format             = DXGI_FORMAT_R32G32B32_FLOAT;
    tdesc.BindFlags          = D3D11_BIND_SHADER_RESOURCE;
    tdesc.CPUAccessFlags     = 0;
    tdesc.MiscFlags          = 0;


    if(FAILED(myDevice->CreateTexture2D(&tdesc,&tbsd,&tex)))
        return(0);

    delete[] buf;

    return(tex);
}

D3D11 CreateTexture2D in 32 bit format

Answers (2)

Related Questions