8 replies to this topic

[Juan Camilo Acosta Arango]

Hi...
I am trying to load a texture with the DXGI_FORMAT_R8_UNORM to obtain its luminance version(grayScale), in D3D9 i was used D3DFMT_L8 and the result is okay but in D3D11 i only obtain the red channel, what i am doing wrong?
 
Rendered Color image: 

 
Render with DXGI_FORMAT_R8_UNORM(bad!!): 

 
 
here is the code:
 

	D3DX11_IMAGE_INFO* pImgInfo = 0;
	HRESULT imghr = D3DX11GetImageInfoFromFile( L"Tiles.png", NULL, pImgInfo, NULL);
	
	D3DX11_IMAGE_LOAD_INFO imgLoadInfo;
		imgLoadInfo.Width = D3DX11_DEFAULT;
		imgLoadInfo.Height = D3DX11_DEFAULT;
		imgLoadInfo.Depth = D3DX11_DEFAULT;
		imgLoadInfo.FirstMipLevel = D3DX11_DEFAULT;
		imgLoadInfo.MipLevels = D3DX11_DEFAULT;
		imgLoadInfo.Usage = D3D11_USAGE_DEFAULT;
		imgLoadInfo.BindFlags = D3D11_BIND_SHADER_RESOURCE;
		imgLoadInfo.CpuAccessFlags = 0;
		imgLoadInfo.MiscFlags = 0;
		imgLoadInfo.Format = DXGI_FORMAT_R8_UNORM;
		//imgLoadInfo.Format = DXGI_FORMAT_R16G16B16A16_UNORM;
		imgLoadInfo.Filter = D3DX11_FILTER_NONE;
		imgLoadInfo.MipFilter = D3DX11_DEFAULT;
		imgLoadInfo.pSrcInfo = pImgInfo;


	m_Texture = m_pRenderer11->LoadTexture( L"Tiles.png", &imgLoadInfo  );

[MJP]

R8_UNORM means you have a texture with just a red channel. If you load an RGB texture with that format, you will only get the red channel from that texture. If you want a texture containing the luminance of the RGB values, you'll need to do that yourself.

[Juan Camilo Acosta Arango]

R8_UNORM means you have a texture with just a red channel. If you load an RGB texture with that format, you will only get the red channel from that texture. If you want a texture containing the luminance of the RGB values, you'll need to do that yourself.

 

Is there no a way like in D3D9 with the format  D3DFMT_L8? when i used these forma to load a texture it visually return a gray scale version

[L. Spiro]

Use the R value from your DXGI_FORMAT_R8_UNORM texture for all channels of the returned color except for A, which should be set to 1.0f.

 

 

L. Spiro

[Juan Camilo Acosta Arango]

Use the R value from your DXGI_FORMAT_R8_UNORM texture for all channels of the returned color except for A, which should be set to 1.0f.

 

 

L. Spiro

 

i'm so sorry L.Spiro, i'm a beginner in d3d11, may you be more specific? When i call LoadTexture how can i set the channels as you said?

[Mike.Popoloski]

As noted in the documentation, you can replicate the D3D9 behavior by using .r to swizzle the data in the shader.

 

So something like this:

 

float4 color = float4(textureColor.rrr, 1.0f);

[Juan Camilo Acosta Arango]

Ok thanks for your help...

 

Now i have errors with the texture sampler, it is only happening with the R8_unorm format.

 

Here is the result: 

 

And here is the source image:

 

this is the pixel shader:

// Pixel Shader
float4 main_ps( v2p input ) : SV_TARGET
{
	float4 l_color0;
	l_color0 = g_frostTexture.Sample(SampleType, input.texCoord);
	l_color1 = float4( l_color1.rrr, 1.0f);
	returnl_color1;
}

 

thanks in advance....

[L. Spiro]

I don’t use Ogre but the error seems apparent: The image is now 4 times as wide and the RGB channels are all the same (so every 3 pixels are grey) and the A channel is white (so every 4th pixel is white).

 

In other words, something is expecting 1 result but is being fed 4.  For whatever reason it is actually using all 4 results and treating them as separate results.  Instead of combining the RGBA values into one pixel, it is spreading them out over 4 pixels.

 

That is basically the problem, but I have never seen this and I am not familiar with Ogre, so I can’t tell you where to look.  The first things I would check would be the texture-creation values and the render-target creation values.

 

 

L. Spiro

[kubera]

The bottom of blog would be interesting (comments):

http://blogs.msdn.com/b/chuckw/archive/2010/07/15/ddswithoutd3dx-sample-update.aspx