Add support for blending operations to CGContext #2222
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
DHowett-MSFT
requested review from
aballway,
rajsesh,
ms-jihua and
msft-Jeyaram
DHowett-MSFT
commented
Mar 14, 2017
Frameworks/CoreGraphics/CGContext.mm
Outdated
| // We cannot fulfill this request. | ||
| UNIMPLEMENTED_WITH_MSG("Unsupported operator blend mode %4.04x", mode); | ||
| } else if (mode == kCGBlendModePlusDarker) { | ||
| // No UNIMPLEMENTED here: we will proceed but with an unusual output. |
There was a problem hiding this comment.
No UNIMPLEMENTED [](start = 11, length = 16)
nit: no, not? not sure what it means.
| } | ||
|
|
||
| auto& state = context->CurrentGState(); | ||
| state.blendMode = mode; |
There was a problem hiding this comment.
nit: context->CurrentGState().blendMode = mode;
aballway
reviewed
Mar 14, 2017
| deviceContext->Clear({ 0, 0, 0, 0 }); // Clear the original target to transparent black. | ||
| RETURN_IF_FAILED(inputBitmap.As(&copiedImage)); | ||
| } | ||
| return __super::Stage(context, deviceContext); |
There was a problem hiding this comment.
nit: would prefer explicit _CRenderOpCommandListBase::Stage
Frameworks/CoreGraphics/CGContext.mm
Outdated
| // It would be ideal to store each operation on the local stack, but that is infeasible because | ||
| // of scoping concerns. | ||
| std::vector<std::unique_ptr<IRenderOperation>> operations; | ||
| operations.reserve(16); |
There was a problem hiding this comment.
nit: magic number. Why 16?
Frameworks/CoreGraphics/CGContext.mm
Outdated
| // PROPERTIES | ||
| // D2D1_BLEND_PROP_MODE: blend mode | ||
| ComPtr<ID2D1Effect> blendEffect; | ||
| FAIL_FAST_IF_FAILED(deviceContext->CreateEffect(CLSID_D2D1Blend, &blendEffect)); |
There was a problem hiding this comment.
why fail fast here when this returns HRESULT?
|
Ping! |
aballway
approved these changes
Mar 15, 2017
|
Ping 2: The Pingening! |
|
Ping 3: Ping Harder |
MSFTFox
approved these changes
Mar 15, 2017
ms-jihua
reviewed
Mar 15, 2017
| deviceContext->SetTransform(__CGAffineTransformToD2D_F(transform)); | ||
| auto revertTransform = wil::ScopeExit([this]() { this->deviceContext->SetTransform(D2D1::IdentityMatrix()); }); | ||
| RETURN_IF_FAILED(std::forward<Lambda>(drawLambda)(this, deviceContext.Get())); | ||
| // Stage and scaffold all rendering operations; this will create layers, command lists, |
There was a problem hiding this comment.
i really really really like this abstraction geez
ms-jihua
approved these changes
Mar 15, 2017
|
@ms-jihua thank you 💯 I'm going to diagram this specific implementation, as well as how the command list blending operations work. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This pull request adds support for blending to CGContext. There are a number of caveats.
kCGBlendModeClearis not supported.kCGBlendModePlusDarkeris not supported (and was not supported in the original Cairo implementation).Some blend modes (the ones without Source/Destination) in the name require a full buffer read-back and compose operation. This is unavoidable. 😦
This change unfortunately does include a rewrite of the CGD2D rendering stack. Instead of adding yet another nesting level of command lists and layers, I switched it to a scaffold/commit stack. Depending on various conditions, operations will be queued up, started in order, and committed in reverse order. A great many of those operations involve creating command lists and layers and blending them or otherwise composing them.
There is a not-insignificant performance penalty in certain cases:
While 19% seems like a lot, it's only 47μs/operation; an application running at 60fps requires its rendering operations complete in ~16000μs. It's not significant, but it is a cost taken per operation.
I believe there is opportunity for improvement; specifically, I had to compromise on allocations to achieve the desired effects of polymorphism in the new render operation stack.
Please examine this very closely.
Fixes #1389.