//--------------------------------------------------------------------------------------
// class ShadowMapRenderTarget: public RenderTarget;
// 用途: シャドウマップのレンダリングターゲット
//--------------------------------------------------------------------------------------
//構築と破棄
ShadowMapRenderTarget::ShadowMapRenderTarget(float ShadowMapDimension) :
pImpl(new Impl(ShadowMapDimension))
{
try{
//デバイスとコンテキストインターフェイスの取得
auto Dev = App::GetApp()->GetDeviceResources();
auto pD3D11Device = Dev->GetD3DDevice();
auto pSwapChain = Dev->GetSwapChain();
auto pD3D11DeviceContext = Dev->GetD3DDeviceContext();
//シャドウマップテクスチャの作成
D3D11_TEXTURE2D_DESC shadowMapDesc;
ZeroMemory(&shadowMapDesc, sizeof(D3D11_TEXTURE2D_DESC));
shadowMapDesc.Width = static_cast<UINT>(pImpl->m_ShadowMapDimension);
shadowMapDesc.Height = static_cast<UINT>(pImpl->m_ShadowMapDimension);
shadowMapDesc.MipLevels = 1;
shadowMapDesc.ArraySize = 1;
shadowMapDesc.SampleDesc.Count = 1;
shadowMapDesc.Usage = D3D11_USAGE_DEFAULT;
shadowMapDesc.BindFlags = D3D11_BIND_DEPTH_STENCIL | D3D11_BIND_SHADER_RESOURCE;
shadowMapDesc.Format = DXGI_FORMAT_R24G8_TYPELESS;
ThrowIfFailed(
pD3D11Device->CreateTexture2D(&shadowMapDesc, nullptr, &pImpl->m_DepthStencil),
L"シャドウマップテクスチャの作成に失敗しました。",
L"pD3D11Device->CreateTexture2D(&shadowMapDesc, nullptr, &m_DepthStencil)",
L"ShadowMapRenderTarget::ShadowMapRenderTarget()"
);
//深度ステンシルビュー作成
D3D11_DEPTH_STENCIL_VIEW_DESC depthStencilViewDesc;
ZeroMemory(&depthStencilViewDesc, sizeof(D3D11_DEPTH_STENCIL_VIEW_DESC));
depthStencilViewDesc.Format = DXGI_FORMAT_D24_UNORM_S8_UINT;
depthStencilViewDesc.ViewDimension = D3D11_DSV_DIMENSION_TEXTURE2D;
depthStencilViewDesc.Texture2D.MipSlice = 0;
ThrowIfFailed(
pD3D11Device->CreateDepthStencilView(pImpl->m_DepthStencil.Get(), &depthStencilViewDesc, &pImpl->m_DepthStencilView),
L"深度ステンシルビューの作成に失敗しました。",
L"pD3D11Device->CreateDepthStencilView(&m_DepthStencil, &depthStencilViewDesc, &m_DepthStencilView)",
L"ShadowMapRenderTarget::ShadowMapRenderTarget()"
);
//シェーダーリソースビュー作成
D3D11_SHADER_RESOURCE_VIEW_DESC shaderResourceViewDesc;
ZeroMemory(&shaderResourceViewDesc, sizeof(D3D11_SHADER_RESOURCE_VIEW_DESC));
shaderResourceViewDesc.Format = DXGI_FORMAT_R24_UNORM_X8_TYPELESS;
shaderResourceViewDesc.ViewDimension = D3D11_SRV_DIMENSION_TEXTURE2D;
shaderResourceViewDesc.Texture2D.MipLevels = 1;
ThrowIfFailed(
pD3D11Device->CreateShaderResourceView(pImpl->m_DepthStencil.Get(), &shaderResourceViewDesc, &pImpl->m_ShaderResourceView),
L"シェーダーリソースビューの作成に失敗しました。",
L"pD3D11Device->CreateShaderResourceView(m_DepthStencil.Get(), &shaderResourceViewDesc, &m_ShaderResourceView)",
L"ShadowMapRenderTarget::ShadowMapRenderTarget()"
);
D3D11_RASTERIZER_DESC shadowRenderStateDesc;
ZeroMemory(&shadowRenderStateDesc, sizeof(D3D11_RASTERIZER_DESC));
shadowRenderStateDesc.CullMode = D3D11_CULL_FRONT;
shadowRenderStateDesc.FillMode = D3D11_FILL_SOLID;
shadowRenderStateDesc.DepthClipEnable = true;
ThrowIfFailed(pD3D11Device->CreateRasterizerState(&shadowRenderStateDesc,&pImpl->m_ShadowRenderState),
L"シャドウマップのラスタライザステートの作成に失敗しました。",
L"pD3D11Device->CreateRasterizerState(&shadowRenderStateDesc,&pImpl->m_ShadowRenderState)",
L"ShadowMapRenderTarget::ShadowMapRenderTarget()"
);
//ビューポートの作成
//シャドウレンダリングビューポート
ZeroMemory(&pImpl->m_ViewPort, sizeof(D3D11_VIEWPORT));
pImpl->m_ViewPort.Height = pImpl->m_ShadowMapDimension;
pImpl->m_ViewPort.Width = pImpl->m_ShadowMapDimension;
pImpl->m_ViewPort.MinDepth = 0.f;
pImpl->m_ViewPort.MaxDepth = 1.f;
}
catch (...){
throw;
}
}
//-------------------------------------------------------------------------------------
_Use_decl_annotations_
HRESULT GPUCompressBC::Compress( const Image& srcImage, const Image& destImage )
{
if ( !srcImage.pixels || !destImage.pixels )
return E_INVALIDARG;
if ( srcImage.width != destImage.width
|| srcImage.height != destImage.height
|| srcImage.width != m_width
|| srcImage.height != m_height
|| srcImage.format != m_srcformat
|| destImage.format != m_bcformat )
{
return E_UNEXPECTED;
}
//--- Create input texture --------------------------------------------------------
auto pDevice = m_device.Get();
if ( !pDevice )
return E_POINTER;
// We need to avoid the hardware doing additional colorspace conversion
DXGI_FORMAT inputFormat = ( m_srcformat == DXGI_FORMAT_R8G8B8A8_UNORM_SRGB ) ? DXGI_FORMAT_R8G8B8A8_UNORM : m_srcformat;
ScopedObject<ID3D11Texture2D> sourceTex;
{
D3D11_TEXTURE2D_DESC desc;
memset( &desc, 0, sizeof(desc) );
desc.Width = static_cast<UINT>( srcImage.width );
desc.Height = static_cast<UINT>( srcImage.height );
desc.MipLevels = 1;
desc.ArraySize = 1;
desc.Format = inputFormat;
desc.SampleDesc.Count = 1;
desc.Usage = D3D11_USAGE_DEFAULT;
desc.BindFlags = D3D11_BIND_SHADER_RESOURCE;
D3D11_SUBRESOURCE_DATA initData;
initData.pSysMem = srcImage.pixels;
initData.SysMemPitch = static_cast<DWORD>( srcImage.rowPitch );
initData.SysMemSlicePitch = static_cast<DWORD>( srcImage.slicePitch );
HRESULT hr = pDevice->CreateTexture2D( &desc, &initData, sourceTex.GetAddressOf() );
if ( FAILED(hr) )
{
return hr;
}
}
ScopedObject<ID3D11ShaderResourceView> sourceSRV;
{
D3D11_SHADER_RESOURCE_VIEW_DESC desc;
memset( &desc, 0, sizeof(desc) );
desc.Texture2D.MipLevels = 1;
desc.Format = inputFormat;
desc.ViewDimension = D3D11_SRV_DIMENSION_TEXTURE2D;
HRESULT hr = pDevice->CreateShaderResourceView( sourceTex.Get(), &desc, sourceSRV.ReleaseAndGetAddressOf() );
if ( FAILED(hr) )
{
return hr;
}
}
//--- Compress using DirectCompute ------------------------------------------------
bool isbc7 = false;
switch( m_bcformat )
{
case DXGI_FORMAT_BC6H_TYPELESS:
case DXGI_FORMAT_BC6H_UF16:
case DXGI_FORMAT_BC6H_SF16:
break;
case DXGI_FORMAT_BC7_TYPELESS:
case DXGI_FORMAT_BC7_UNORM:
case DXGI_FORMAT_BC7_UNORM_SRGB:
isbc7 = true;
break;
default:
return E_UNEXPECTED;
}
const UINT MAX_BLOCK_BATCH = 64;
auto pContext = m_context.Get();
if ( !pContext )
return E_UNEXPECTED;
size_t xblocks = std::max<size_t>( 1, (m_width + 3) >> 2 );
size_t yblocks = std::max<size_t>( 1, (m_height + 3) >> 2 );
UINT num_total_blocks = static_cast<UINT>( xblocks * yblocks );
UINT num_blocks = num_total_blocks;
int start_block_id = 0;
while (num_blocks > 0)
{
UINT n = std::min<UINT>( num_blocks, MAX_BLOCK_BATCH );
UINT uThreadGroupCount = n;
{
D3D11_MAPPED_SUBRESOURCE mapped;
HRESULT hr = pContext->Map( m_constBuffer.Get(), 0, D3D11_MAP_WRITE_DISCARD, 0, &mapped );
if ( FAILED(hr) )
return hr;
ConstantsBC6HBC7 param;
param.tex_width = static_cast<UINT>( srcImage.width );
param.num_block_x = static_cast<UINT>( xblocks );
param.format = m_bcformat;
param.mode_id = 0;
param.start_block_id = start_block_id;
param.num_total_blocks = num_total_blocks;
param.alpha_weight = m_alphaWeight;
memcpy( mapped.pData, ¶m, sizeof( param ) );
pContext->Unmap( m_constBuffer.Get(), 0 );
}
if ( isbc7 )
{
//--- BC7 -----------------------------------------------------------------
ID3D11ShaderResourceView* pSRVs[] = { sourceSRV.Get(), nullptr };
RunComputeShader( pContext, m_BC7_tryMode456CS.Get(), pSRVs, 2, m_constBuffer.Get(),
m_err1UAV.Get(), std::max<UINT>(uThreadGroupCount / 4, 1) );
for ( UINT i = 0; i < 3; ++i )
{
static const UINT modes[] = { 1, 3, 7 };
{
D3D11_MAPPED_SUBRESOURCE mapped;
HRESULT hr = pContext->Map( m_constBuffer.Get(), 0, D3D11_MAP_WRITE_DISCARD, 0, &mapped );
if ( FAILED(hr) )
{
ResetContext( pContext );
return hr;
}
ConstantsBC6HBC7 param;
param.tex_width = static_cast<UINT>( srcImage.width );
param.num_block_x = static_cast<UINT>( xblocks );
param.format = m_bcformat;
param.mode_id = modes[i];
param.start_block_id = start_block_id;
param.num_total_blocks = num_total_blocks;
param.alpha_weight = m_alphaWeight;
memcpy( mapped.pData, ¶m, sizeof( param ) );
pContext->Unmap( m_constBuffer.Get(), 0 );
}
pSRVs[1] = (i & 1) ? m_err2SRV.Get() : m_err1SRV.Get();
RunComputeShader( pContext, m_BC7_tryMode137CS.Get(), pSRVs, 2, m_constBuffer.Get(),
(i & 1) ? m_err1UAV.Get() : m_err2UAV.Get(), uThreadGroupCount );
}
for ( UINT i = 0; i < 2; ++i )
{
static const UINT modes[] = { 0, 2 };
{
D3D11_MAPPED_SUBRESOURCE mapped;
HRESULT hr = pContext->Map( m_constBuffer.Get(), 0, D3D11_MAP_WRITE_DISCARD, 0, &mapped );
if ( FAILED(hr) )
{
ResetContext( pContext );
return hr;
}
ConstantsBC6HBC7 param;
param.tex_width = static_cast<UINT>( srcImage.width );
param.num_block_x = static_cast<UINT>( xblocks );
param.format = m_bcformat;
param.mode_id = modes[i];
param.start_block_id = start_block_id;
param.num_total_blocks = num_total_blocks;
param.alpha_weight = m_alphaWeight;
memcpy( mapped.pData, ¶m, sizeof( param ) );
pContext->Unmap( m_constBuffer.Get(), 0 );
}
pSRVs[1] = (i & 1) ? m_err1SRV.Get() : m_err2SRV.Get();
RunComputeShader( pContext, m_BC7_tryMode02CS.Get(), pSRVs, 2, m_constBuffer.Get(),
(i & 1) ? m_err2UAV.Get() : m_err1UAV.Get(), uThreadGroupCount );
}
pSRVs[1] = m_err2SRV.Get();
RunComputeShader( pContext, m_BC7_encodeBlockCS.Get(), pSRVs, 2, m_constBuffer.Get(),
m_outputUAV.Get(), std::max<UINT>(uThreadGroupCount / 4, 1) );
}
else
{
//--- BC6H ----------------------------------------------------------------
ID3D11ShaderResourceView* pSRVs[] = { sourceSRV.Get(), nullptr };
RunComputeShader( pContext, m_BC6H_tryModeG10CS.Get(), pSRVs, 2, m_constBuffer.Get(),
m_err1UAV.Get(), std::max<UINT>(uThreadGroupCount / 4, 1) );
for ( UINT i = 0; i < 10; ++i )
{
{
D3D11_MAPPED_SUBRESOURCE mapped;
HRESULT hr = pContext->Map( m_constBuffer.Get(), 0, D3D11_MAP_WRITE_DISCARD, 0, &mapped );
if ( FAILED(hr) )
{
ResetContext( pContext );
return hr;
}
ConstantsBC6HBC7 param;
param.tex_width = static_cast<UINT>( srcImage.width );
param.num_block_x = static_cast<UINT>( xblocks );
param.format = m_bcformat;
param.mode_id = i;
param.start_block_id = start_block_id;
param.num_total_blocks = num_total_blocks;
memcpy( mapped.pData, ¶m, sizeof( param ) );
pContext->Unmap( m_constBuffer.Get(), 0 );
}
pSRVs[1] = (i & 1) ? m_err2SRV.Get() : m_err1SRV.Get();
RunComputeShader( pContext, m_BC6H_tryModeLE10CS.Get(), pSRVs, 2, m_constBuffer.Get(),
(i & 1) ? m_err1UAV.Get() : m_err2UAV.Get(), std::max<UINT>(uThreadGroupCount / 2, 1) );
}
pSRVs[1] = m_err1SRV.Get();
RunComputeShader( pContext, m_BC6H_encodeBlockCS.Get(), pSRVs, 2, m_constBuffer.Get(),
m_outputUAV.Get(), std::max<UINT>(uThreadGroupCount / 2, 1) );
}
start_block_id += n;
num_blocks -= n;
}
ResetContext( pContext );
//--- Copy output texture back to CPU ---------------------------------------------
pContext->CopyResource( m_outputCPU.Get(), m_output.Get() );
D3D11_MAPPED_SUBRESOURCE mapped;
HRESULT hr = pContext->Map( m_outputCPU.Get(), 0, D3D11_MAP_READ, 0, &mapped );
if ( SUCCEEDED(hr) )
{
const uint8_t *pSrc = reinterpret_cast<const uint8_t *>( mapped.pData );
uint8_t *pDest = destImage.pixels;
size_t pitch = xblocks * sizeof( BufferBC6HBC7 );
size_t rows = std::max<size_t>( 1, ( destImage.height + 3 ) >> 2 );
for( size_t h = 0; h < rows; ++h )
{
memcpy( pDest, pSrc, destImage.rowPitch );
pSrc += pitch;
pDest += destImage.rowPitch;
}
pContext->Unmap( m_outputCPU.Get(), 0 );
}
return hr;
}
/**
* Callback to draw things in world space.
***/
void OSVR_DirectMode::DrawWorld(void* userData, osvr::renderkit::GraphicsLibrary cLibrary, osvr::renderkit::RenderBuffer cBuffers,
osvr::renderkit::OSVR_ViewportDescription sViewport, OSVR_PoseState pose, osvr::renderkit::OSVR_ProjectionMatrix sProjection, OSVR_TimeValue deadline)
{
static int nEye = 0;
static float fAspect = 1.0f;
// Make sure our pointers are filled in correctly. The config file selects
// the graphics library to use, and may not match our needs.
if (cLibrary.D3D11 == nullptr)
{
std::cerr << "SetupDisplay: No D3D11 GraphicsLibrary" << std::endl;
return;
}
if (cBuffers.D3D11 == nullptr)
{
std::cerr << "SetupDisplay: No D3D11 RenderBuffer" << std::endl;
return;
}
// auto pcContext = cLibrary.D3D11->context;
auto pcDevice = cLibrary.D3D11->device;
auto pcContext = cLibrary.D3D11->context;
// create all bool
bool bAllCreated = true;
// create vertex shader
if (!m_pcVertexShader11)
{
if (FAILED(Create2DVertexShader(pcDevice, &m_pcVertexShader11, &m_pcVertexLayout11)))
{
OutputDebugString(L"FAILED");
bAllCreated = false;
}
}
// create pixel shader...
if (!m_pcPixelShader11)
{
if (FAILED(CreatePixelShaderEffect(pcDevice, &m_pcPixelShader11, PixelShaderTechnique::FullscreenGammaCorrection)))
bAllCreated = false;
}
// Create vertex buffer
if (!m_pcVertexBuffer11)
{
if (FAILED(CreateFullScreenVertexBuffer(pcDevice, &m_pcVertexBuffer11)))
bAllCreated = false;
}
// create constant buffer
if (!m_pcConstantBufferDirect11)
{
if (FAILED(CreateMatrixConstantBuffer(pcDevice, &m_pcConstantBufferDirect11)))
bAllCreated = false;
}
// sampler ?
if (!m_pcSamplerState)
{
bAllCreated = false;
}
if ((bAllCreated) && (m_sStereoTextureViews.m_ppcTexView11[nEye]))
{
// Set the input layout
pcContext->IASetInputLayout(m_pcVertexLayout11);
// Set vertex buffer
UINT stride = sizeof(TexturedVertex);
UINT offset = 0;
pcContext->IASetVertexBuffers(0, 1, &m_pcVertexBuffer11, &stride, &offset);
// get orthographic matrix from projection and normalize it by its width (since we use a fullscreen shader here)
float afProjectionD3D[16];
osvr::renderkit::OSVR_Projection_to_D3D(afProjectionD3D, sProjection);
D3DXMATRIX sProj(afProjectionD3D);
// due to the aspect ratio (90° horizontal, 90° vertical) of the HDK we adjust the screen by
// the height, not by the width... in this case we need to set a higher FOV by following formular:
// V = 2 * arctan( tan(H / 2) * aspectratio ) - so we get V 90° and H 121°
sProj.m[0][0] = sProj.m[0][0] * fAspect; // < incorporate game screen aspect ratio;
sProj.m[0][1] = 0.0f;
sProj.m[0][3] = sProj.m[0][2];
sProj.m[0][2] = 0.0f;
sProj.m[1][0] = 0.0f;
sProj.m[1][1] = sProj.m[1][1];
sProj.m[1][3] = sProj.m[1][2];
sProj.m[1][2] = 0.0f;
sProj.m[2][0] = 0.0f;
sProj.m[2][1] = 0.0f;
sProj.m[2][2] = 1.0f; // 1.0f here... fullscreen shader !
sProj.m[2][3] = 0.0f;
sProj.m[3][0] = 0.0f;
sProj.m[3][1] = 0.0f;
sProj.m[3][2] = 0.0f;
sProj.m[3][3] = 1.0f;
// zoom out ?
if (m_pbZoomOut)
{
if (*m_pbZoomOut)
{
sProj.m[0][0] /= 2.0f;
sProj.m[1][1] /= 2.0f;
}
}
// Set constant buffer, first update it... scale and translate the left and right image
pcContext->UpdateSubresource((ID3D11Resource*)m_pcConstantBufferDirect11, 0, NULL, &sProj, 0, 0);
pcContext->VSSetConstantBuffers(0, 1, &m_pcConstantBufferDirect11);
// Set primitive topology
pcContext->IASetPrimitiveTopology(D3D11_PRIMITIVE_TOPOLOGY_TRIANGLELIST);
// texture connected ?
if ((m_sStereoTextureViews.m_ppcTexView11[nEye]) && (*m_sStereoTextureViews.m_ppcTexView11[nEye]))
{
if (m_eMethod == OSVR_DirectModeMethods::OSVR_D3D11_use_Game_Device)
{
// set texture, sampler state
pcContext->PSSetShaderResources(0, 1, m_sStereoTextureViews.m_ppcTexView11[nEye]);
pcContext->PSSetSamplers(0, 1, &m_pcSamplerState);
}
else
{
ID3D11Resource* pcResource = nullptr;
(*m_sStereoTextureViews.m_ppcTexView11[nEye])->GetResource(&pcResource);
if (!m_sStereoTextureCopies.m_pcTex11Copy[nEye])
{
// get the description and create the copy texture
D3D11_TEXTURE2D_DESC sDesc;
((ID3D11Texture2D*)pcResource)->GetDesc(&sDesc);
sDesc.MiscFlags |= D3D11_RESOURCE_MISC_SHARED;
sDesc.BindFlags |= D3D11_BIND_SHADER_RESOURCE;
if (FAILED(((ID3D11Device*)m_pcGameDevice)->CreateTexture2D(&sDesc, NULL, (ID3D11Texture2D**)&m_sStereoTextureCopies.m_pcTex11Copy[nEye])))
{
OutputDebugString(L"StereoSplitterDX10 : Failed to create twin texture !");
return;
}
// aspect ratio
fAspect = (float)sDesc.Width / (float)sDesc.Height;
// TODO !! DX9 // DX10 !!
// get shared handle
IDXGIResource* pcDXGIResource(NULL);
m_sStereoTextureCopies.m_pcTex11Copy[nEye]->QueryInterface(__uuidof(IDXGIResource), (void**)&pcDXGIResource);
HANDLE sharedHandle;
if (pcDXGIResource)
{
pcDXGIResource->GetSharedHandle(&sharedHandle);
pcDXGIResource->Release();
}
else OutputDebugString(L"Failed to query IDXGIResource.");
// open the shared handle with the temporary device
ID3D11Resource* pcResourceShared;
pcDevice->OpenSharedResource(sharedHandle, __uuidof(ID3D11Resource), (void**)(&pcResourceShared));
if (pcResourceShared)
{
pcResourceShared->QueryInterface(__uuidof(ID3D11Texture2D), (void**)(&m_sStereoFrameTextures.m_pcFrameTexture[nEye]));
pcResourceShared->Release();
}
else OutputDebugString(L"Could not open shared resource.");
// create shader resource view
if (m_sStereoFrameTextures.m_pcFrameTexture[nEye])
{
D3D11_SHADER_RESOURCE_VIEW_DESC sDescSRV;
ZeroMemory(&sDescSRV, sizeof(D3D11_SHADER_RESOURCE_VIEW_DESC));
sDescSRV.Format = sDesc.Format;
sDescSRV.ViewDimension = D3D11_SRV_DIMENSION_TEXTURE2D;
sDescSRV.Texture2D.MostDetailedMip = 0;
sDescSRV.Texture2D.MipLevels = 1;
if (FAILED(pcDevice->CreateShaderResourceView(m_sStereoFrameTextures.m_pcFrameTexture[nEye], &sDescSRV, &m_sSteroFrameTextureSRViews.m_pcFrameTextureSRView[nEye])))
OutputDebugString(L"Failed to create shader resource view.");
}
else OutputDebugString(L"No Texture available.");
}
else
{
// copy the frame tex to shared texture
m_pcGameDeviceContext->CopyResource(m_sStereoTextureCopies.m_pcTex11Copy[nEye], pcResource);
if (pcResource) pcResource->Release();
// set texture, sampler state
pcContext->PSSetShaderResources(0, 1, &m_sSteroFrameTextureSRViews.m_pcFrameTextureSRView[nEye]);
pcContext->PSSetSamplers(0, 1, &m_pcSamplerState);
}
}
}
// set shaders
pcContext->VSSetShader(m_pcVertexShader11, 0, 0);
pcContext->PSSetShader(m_pcPixelShader11, 0, 0);
// Render a triangle
pcContext->Draw(6, 0);
// switch eye for next call
nEye = !nEye;
}
}
//-------------------------------------------------------------------------------------
_Use_decl_annotations_
HRESULT GPUCompressBC::Prepare( size_t width, size_t height, DXGI_FORMAT format, float alphaWeight )
{
if ( !width || !height || alphaWeight < 0.f )
return E_INVALIDARG;
#ifdef _M_X64
if ( (width > 0xFFFFFFFF) || (height > 0xFFFFFFFF) )
return E_INVALIDARG;
#endif
m_width = width;
m_height = height;
size_t xblocks = std::max<size_t>( 1, (width + 3) >> 2 );
size_t yblocks = std::max<size_t>( 1, (height + 3) >> 2 );
size_t num_blocks = xblocks * yblocks;
switch( format )
{
// BC6H GPU compressor takes RGBAF32 as input
case DXGI_FORMAT_BC6H_TYPELESS:
case DXGI_FORMAT_BC6H_UF16:
case DXGI_FORMAT_BC6H_SF16:
m_srcformat = DXGI_FORMAT_R32G32B32A32_FLOAT;
break;
// BC7 GPU compressor takes RGBA32 as input
case DXGI_FORMAT_BC7_TYPELESS:
case DXGI_FORMAT_BC7_UNORM:
m_srcformat = DXGI_FORMAT_R8G8B8A8_UNORM;
break;
case DXGI_FORMAT_BC7_UNORM_SRGB:
m_srcformat = DXGI_FORMAT_R8G8B8A8_UNORM_SRGB;
break;
default:
m_bcformat = m_srcformat = DXGI_FORMAT_UNKNOWN;
return HRESULT_FROM_WIN32( ERROR_NOT_SUPPORTED );
}
m_bcformat = format;
auto pDevice = m_device.Get();
if ( !pDevice )
return E_POINTER;
// Create structured buffers
size_t bufferSize = num_blocks * sizeof( BufferBC6HBC7 );
{
D3D11_BUFFER_DESC desc;
memset( &desc, 0, sizeof(desc) );
desc.BindFlags = D3D11_BIND_UNORDERED_ACCESS | D3D11_BIND_SHADER_RESOURCE;
desc.Usage = D3D11_USAGE_DEFAULT;
desc.MiscFlags = D3D11_RESOURCE_MISC_BUFFER_STRUCTURED;
desc.StructureByteStride = sizeof( BufferBC6HBC7 );
desc.ByteWidth = static_cast<UINT>( bufferSize );
HRESULT hr = pDevice->CreateBuffer( &desc, nullptr, m_output.ReleaseAndGetAddressOf() );
if ( FAILED(hr) )
{
return hr;
}
hr = pDevice->CreateBuffer( &desc, nullptr, m_err1.ReleaseAndGetAddressOf() );
if ( FAILED(hr) )
{
return hr;
}
hr = pDevice->CreateBuffer( &desc, nullptr, m_err2.ReleaseAndGetAddressOf() );
if ( FAILED(hr) )
{
return hr;
}
}
// Create staging output buffer
{
D3D11_BUFFER_DESC desc;
memset( &desc, 0, sizeof(desc) );
desc.Usage = D3D11_USAGE_STAGING;
desc.CPUAccessFlags = D3D11_CPU_ACCESS_READ;
desc.ByteWidth = static_cast<UINT>( bufferSize );
HRESULT hr = pDevice->CreateBuffer( &desc, nullptr, m_outputCPU.ReleaseAndGetAddressOf() );
if ( FAILED(hr) )
{
return hr;
}
}
// Create constant buffer
{
D3D11_BUFFER_DESC desc;
memset( &desc, 0, sizeof(desc) );
desc.BindFlags = D3D11_BIND_CONSTANT_BUFFER;
desc.Usage = D3D11_USAGE_DYNAMIC;
desc.CPUAccessFlags = D3D11_CPU_ACCESS_WRITE;
desc.ByteWidth = sizeof( ConstantsBC6HBC7 );
HRESULT hr = pDevice->CreateBuffer( &desc, nullptr, m_constBuffer.ReleaseAndGetAddressOf() );
if ( FAILED(hr) )
{
return hr;
}
}
// Create shader resource views
{
D3D11_SHADER_RESOURCE_VIEW_DESC desc;
memset( &desc, 0, sizeof(desc) );
desc.Buffer.NumElements = static_cast<UINT>( num_blocks );
desc.ViewDimension = D3D11_SRV_DIMENSION_BUFFER;
HRESULT hr = pDevice->CreateShaderResourceView( m_err1.Get(), &desc, m_err1SRV.ReleaseAndGetAddressOf() );
if ( FAILED(hr) )
{
return hr;
}
hr = pDevice->CreateShaderResourceView( m_err2.Get(), &desc, m_err2SRV.ReleaseAndGetAddressOf() );
if ( FAILED(hr) )
{
return hr;
}
}
// Create unordered access views
{
D3D11_UNORDERED_ACCESS_VIEW_DESC desc;
memset( &desc, 0, sizeof(desc) );
desc.Buffer.NumElements = static_cast<UINT>( num_blocks );
desc.ViewDimension = D3D11_UAV_DIMENSION_BUFFER;
HRESULT hr = pDevice->CreateUnorderedAccessView( m_output.Get(), &desc, m_outputUAV.ReleaseAndGetAddressOf() );
if ( FAILED(hr) )
{
return hr;
}
hr = pDevice->CreateUnorderedAccessView( m_err1.Get(), &desc, m_err1UAV.ReleaseAndGetAddressOf() );
if ( FAILED(hr) )
{
return hr;
}
hr = pDevice->CreateUnorderedAccessView( m_err2.Get(), &desc, m_err2UAV.ReleaseAndGetAddressOf() );
if ( FAILED(hr) )
{
return hr;
}
}
return S_OK;
}
Texture* Texture::createFromData(const sSize& size, const unsigned char* in_pData, bool in_generateMipmaps)
{
ID3D11Texture2D* pTexture = NULL;
ID3D11ShaderResourceView* pTextureView = NULL;
auto pRet = new Texture();
// Manually generate mip levels
bool allowMipMaps = true;
UINT w2 = 1;
UINT h2 = 1;
while (w2 < (UINT)size.x) w2 *= 2;
if (size.x != w2) allowMipMaps = false;
while (h2 < (UINT)size.y) h2 *= 2;
if (size.y != h2) allowMipMaps = false;
unsigned char* pMipMaps = NULL;
int mipLevels = 1;
D3D11_SUBRESOURCE_DATA* mipsData = NULL;
allowMipMaps = allowMipMaps && in_generateMipmaps;
if (allowMipMaps)
{
UINT biggest = std::max<>(w2, h2);
UINT w2t = w2;
UINT h2t = h2;
UINT totalSize = w2t * h2t * 4;
while (!(w2t == 1 && h2t == 1))
{
++mipLevels;
w2t /= 2;
if (w2t < 1) w2t = 1;
h2t /= 2;
if (h2t < 1) h2t = 1;
totalSize += w2t * h2t * 4;
}
pMipMaps = new byte[totalSize];
memcpy(pMipMaps, in_pData, size.x * size.y * 4);
mipsData = new D3D11_SUBRESOURCE_DATA[mipLevels];
w2t = w2;
h2t = h2;
totalSize = 0;
int mipTarget = mipLevels;
mipLevels = 0;
byte* prev;
byte* cur;
while (mipLevels != mipTarget)
{
prev = pMipMaps + totalSize;
mipsData[mipLevels].pSysMem = prev;
mipsData[mipLevels].SysMemPitch = w2t * 4;
mipsData[mipLevels].SysMemSlicePitch = 0;
totalSize += w2t * h2t * 4;
cur = pMipMaps + totalSize;
w2t /= 2;
if (w2t < 1) w2t = 1;
h2t /= 2;
if (h2t < 1) h2t = 1;
++mipLevels;
if (mipLevels == mipTarget) break;
int accum;
// Generate the mips
int multX = w2 / w2t;
int multY = h2 / h2t;
for (UINT y = 0; y < h2t; ++y)
{
for (UINT x = 0; x < w2t; ++x)
{
for (UINT k = 0; k < 4; ++k)
{
accum = 0;
accum += prev[(y * multY * w2 + x * multX) * 4 + k];
accum += prev[(y * multY * w2 + (x + multX / 2) * multX) * 4 + k];
accum += prev[((y + multY / 2) * multY * w2 + x * multX) * 4 + k];
accum += prev[((y + multY / 2) * multY * w2 + (x + multX / 2) * multX) * 4 + k];
cur[(y * w2t + x) * 4 + k] = accum / 4;
}
}
}
w2 = w2t;
h2 = h2t;
}
}
D3D11_TEXTURE2D_DESC desc;
desc.Width = size.x;
desc.Height = size.y;
desc.MipLevels = mipLevels;
desc.ArraySize = 1;
desc.Format = DXGI_FORMAT_R8G8B8A8_UNORM;
desc.SampleDesc.Count = 1;
desc.SampleDesc.Quality = 0;
desc.Usage = D3D11_USAGE_IMMUTABLE;
desc.BindFlags = D3D11_BIND_SHADER_RESOURCE;
desc.CPUAccessFlags = 0;
desc.MiscFlags = 0;
D3D11_SUBRESOURCE_DATA data;
data.pSysMem = (pMipMaps) ? pMipMaps : in_pData;
data.SysMemPitch = size.x * 4;
data.SysMemSlicePitch = 0;
auto pDevice = ORenderer->getDevice();
auto ret = pDevice->CreateTexture2D(&desc, (mipsData) ? mipsData : &data, &pTexture);
assert(ret == S_OK);
ret = pDevice->CreateShaderResourceView(pTexture, NULL, &pTextureView);
assert(ret == S_OK);
pTexture->Release();
if (pMipMaps) delete[] pMipMaps;
if (mipsData) delete[] mipsData;
pRet->m_size = size;
pRet->m_pTextureView = pTextureView;
return pRet;
}
