Hello, I'm new to this community, so I ask for a bit of patience with my knowledge!. After using the search function and not finding anything specifically useful or that works, I decided to ask for help on this issue.

I'm programming a shader in SharpDX (DirectX11, C#) and I need to store the depth in a texture in a first pass, and then use it in a second one.

The thing is my texture returns only the clear value I'm using (1.0f) even though I think I've followed the necessary steps to write and use a Depth Texture in a shader.

Here's how I set the Texture2D, the ShaderResourceView, the DepthStencilView and the DepthStencilStateDescription and then bind them.

this.depthBuffer = new Texture2D(device, new Texture2DDescription()
            {
                Format = Format.R32_Typeless, 
                ArraySize = 1,
                MipLevels = 1,
                Width = (int)host.ActualWidth,
                Height = (int)host.ActualHeight,
                SampleDescription = new SampleDescription(1, 0),
                Usage = ResourceUsage.Default,
                BindFlags = BindFlags.DepthStencil | BindFlags.ShaderResource,
                CpuAccessFlags = CpuAccessFlags.None,
                OptionFlags = ResourceOptionFlags.None,
            });

            // shader resource
            this.depthBufferShaderResourceView = new ShaderResourceView(this.device, this.depthBuffer, new ShaderResourceViewDescription()
            {
                Format = Format.R32_Float,
                Dimension = ShaderResourceViewDimension.Texture2D,
                Texture2D = new ShaderResourceViewDescription.Texture2DResource()
                {
                    MipLevels = 1,
                    MostDetailedMip = 0,
                }
            });

            this.depthBufferStencilView = new DepthStencilView(device, depthBuffer, new DepthStencilViewDescription()
            {
                Format = Format.D32_Float,
                Dimension = DepthStencilViewDimension.Texture2D,
                Texture2D = new DepthStencilViewDescription.Texture2DResource()
                {
                    MipSlice = 0
                }
            });

            this.device.ImmediateContext.OutputMerger.SetTargets(depthBufferStencilView);
            this.device.ImmediateContext.ClearDepthStencilView(this.depthBufferStencilView, DepthStencilClearFlags.Depth | DepthStencilClearFlags.Stencil, 1.0f, 0);

            var depthStencilDesc = new DepthStencilStateDescription()
            {
                DepthComparison = Comparison.LessEqual,
                DepthWriteMask = global::SharpDX.Direct3D11.DepthWriteMask.All,
                IsDepthEnabled = true,
            };
            this.depthStencilState = new DepthStencilState(this.device, depthStencilDesc); 

            this.device.ImmediateContext.OutputMerger.SetDepthStencilState(this.depthStencilState);

            this.depthStencilShaderResourceVariable = effect.GetVariableByName("DepthTexture").AsShaderResource();
            this.depthStencilShaderResourceVariable.SetResource(this.depthBufferShaderResourceView);

And here's how I'm extracting the value of the Depth Texture in the pixel shader:

float depthPixel = DepthTexture.Load(posTex);
float4 color = float4(depthPixel, depthPixel , depthPixel, 1.0f ); // just a simple test, returns all pixels I in full white...

Any help would be appreciated. I would say I'm doing the "basics" correctly but I don't know if I'm either missing something or if there's something completely wrong with what I did.

Thanks, c4sh

Is it possible that the key aspect is the 2-pass structure? I'm setting all these structures up just once. And running the two passes. Could it make sense to set them up at the beginning for the first pass and then changing/unbinding some of them?

As hinted by MJP, I tried to unbind the depth stencil view before binding it as a shader resource view, so this is how I'm now setting the resources at each pass:

// running of the passes in the .cs file

this.device.ImmediateContext.InputAssembler.SetVertexBuffers(0, new VertexBufferBinding(this.vertexBuffer, LinesVertex.SizeInBytes, 0));

// PASS 0
this.device.ImmediateContext.OutputMerger.SetTargets(depthBufferStencilView);
this.device.ImmediateContext.ClearDepthStencilView(this.depthBufferStencilView, DepthStencilClearFlags.Depth | DepthStencilClearFlags.Stencil, 1.0f, 0);
this.technique.GetPassByIndex(0).Apply(this.device.ImmediateContext);
this.device.ImmediateContext.DrawIndexed(this.geometry.Indices.Length, 0, 0);

// PASS 1
this.device.ImmediateContext.OutputMerger.ResetTargets(); // unbinding the depthStencilView

this.device.ImmediateContext.InputAssembler.SetVertexBuffers(0, new VertexBufferBinding(this.vertexBuffer, LinesVertex.SizeInBytes, 0));
this.depthStencilShaderResourceVariable = effect.GetVariableByName("DepthTexture").AsShaderResource();
this.depthStencilShaderResourceVariable.SetResource(this.depthBufferShaderResourceView);
this.technique.GetPassByIndex(1).Apply(this.device.ImmediateContext);
this.device.ImmediateContext.DrawIndexed(this.geometry.Indices.Length, 0, 0);

Unfortunately I'm not getting now ANY result from the Depth Texture I'm reading. Just a white screen with nothing (not even the "clear" result I set on the Depth Stencil View).
I'm I doing something wrong on the order/way of setting and unbinding the resources here? Thanks for any hint.

EDIT:

I read this in an old topic of this forum:

If you create the backing resource as a typeless 32-bit surface (DXGI_FORMAT_R32_TYPELESS), you can create both depth stencil and shader resource views against it so you can directly access it as a texture after you've used it as an actual depth buffer.

Is this still valid? Because then maybe I woudln't need the unbinding etc and it should be correct to use it like I was doing before, and the error could be somewhere else?