Extend D3D12 Depth Test Overlay #3126
Conversation
| ID3D12RootSignature *DepthResolveRootSig = NULL; | ||
| std::unordered_map<DXGI_FORMAT, ID3D12PipelineState *[8]> DepthResolvePipe; | ||
| ID3D12RootSignature *DepthCopyRootSig = NULL; | ||
| std::unordered_map<DXGI_FORMAT, ID3D12PipelineState *[8]> DepthCopyPipe; |
There was a problem hiding this comment.
A map here is overkill when we know this will only ever be used with two keys - I'd prefer just making this a simple array with [0] being D24S8 and [1] being D32S8, as that's what is done in other situations. If you want you can have a helper function to get the index for a given format, or you can do it with an if inline where needed.
| RDCASSERT(root); | ||
| hr = device->CreateRootSignature(0, root->GetBufferPointer(), root->GetBufferSize(), | ||
| __uuidof(ID3D12RootSignature), (void **)&DepthCopyRootSig); | ||
| SAFE_RELEASE(root); |
There was a problem hiding this comment.
There's no need for a root signature to be strictly limited to only the used elements, so you could use this same root signature for both passes. The only thing I'm not sure about is if D3D12 allows you to leave unused root signature parameters unset - it might be legal, but for safety you could bind the same parameter in both cases even if it's only used in one.
| for(DXGI_FORMAT fmt : {DXGI_FORMAT_D24_UNORM_S8_UINT, DXGI_FORMAT_D32_FLOAT_S8X24_UINT}) | ||
| { | ||
| ID3D12PipelineState **psos = DepthResolvePipe[fmt]; | ||
| for(size_t i = 0; i < 8; i++) |
There was a problem hiding this comment.
You can reduce this loop down to 5 instead of 8 - 16x MSAA is as high as any hardware supports currently or is likely to support.
| if(overlayTexDesc.SampleDesc.Count == 1) | ||
| { | ||
| srvDesc.ViewDimension = D3D12_SRV_DIMENSION_TEXTURE2D; | ||
| srvDesc.Texture2D.MipLevels = (UINT)-1; |
There was a problem hiding this comment.
This kind of cast looks 'wrong' to me, I'd rather use ~0U or a #define if there is one.
Support shader exported depth by replaying using the capture pixel shader to determine passing pixels
Zorro666
force-pushed
the
support-shader-exported-depth-in-depth-test-overlay-d3d12
branch
from
fb06951 to
cc52b2e
Compare
Zorro666
deleted the
support-shader-exported-depth-in-depth-test-overlay-d3d12
branch
Description
D3D12 Replay support for Issue #2858
This is a PR in a sequence of PRs which all together complete the implementation of the feature.
Future PRs will cover replay support for Vulkan and extending the automated overlay tests
GL_Overlay_Test,D3D11_Overlay_Test,D3D12_Overlay_Test,VK_Overlay_Test.The feature enables supporting shader exported depth for the Depth Test overlay by replaying the draw using the pixel shader from the capture instead of a replacement shader. Depth test passing pixels are identified by using the stencil buffer to generate a stencil mask.
If the depth buffer used for the draw does not have a stencil buffer then a new depth buffer is created with a depth+stencil format and a fullscreen pass is used to copy depth from the original depth buffer to the newly created depth+stencil buffer.
Tested on Windows using changes to
D3D12_Overlay_TestThe change makes the pixel shader write out depth 0.0 for a small rectangle just bottom-left of the centre
Before (using v1.29) : the rectangle of output depth 0.0 is shown as failing the depth test
Depth Buffer format
Depth
D24_S8Depth Buffer format
D32F_S8Depth Buffer format
D16MSAADepth Buffer format
D32FMSAAAfter : the rectangle of output depth 0.0 is shown as passing the depth test
Depth Buffer format
D24_S8Depth Buffer format
D32F_S8Depth Buffer format
D16MSAABuffer format
D32FMSAA