enhanced the rendering composition target to better support features such as
alpha blending, blending, masking, post effects, and more.
This allows rasterizers to prepare the composition context with more precise control.
Additionally, resolved a crash in the software engine's post-effect process
caused by a buffer size mismatch during XY buffer flipping.
issue: https://github.com/thorvg/thorvg/issues/3009
issue: https://github.com/thorvg/thorvg/issues/2984
- Enhanced Scene management to provide users with more control.
- Scenes now support adding specific scenes at defined positions and removing them as needed.
- Ensure safe access to Canvas, Scene paints() by adding const specifiers.
- Removed virtual specifier for the canvas primitive apis.
- Introduced a nested scene in the canvas to remove logic duplication.
C++ API Modification:
- Result Scene::push(Paint* paint)
-> Result Scene::push(Paint* target, Paint* at = nullptr)
- Result Scene::clear(bool free = true)
-> Result Scene::remove(Paint* paint = nullptr)
- Result Canvas::push(Paint* paint)
-> Result Canvas::push(Paint* target, Paint* at = nullptr)
- list<Paint*>& Scene::paints()
-> const list<Paint*>& Scene::paints() const
- list<Paint*>& Canvas::paints()
-> const list<Paint*>& Canvas::paints() const
C++ API Addition:
- Result Canvas::remove(Paint* paint = nullptr);
C API Modifications:
- Tvg_Result tvg_scene_clear(Tvg_Paint* scene, bool free)
-> Tvg_Result tvg_scene_remove(Tvg_Paint* scene, Tvg_Paint* paint)
C API Addition:
- Tvg_Result tvg_scene_push_at(Tvg_Paint* scene, Tvg_Paint* target, Tvg_Paint* at)
- Tvg_Result tvg_canvas_push_at(Tvg_Canvas* canvas, Tvg_Paint* target, Tvg_Paint* at)
- Tvg_Result tvg_canvas_remove(Tvg_Canvas* canvas, Tvg_Paint* paint)
issue: https://github.com/thorvg/thorvg/issues/2957
issue: https://github.com/thorvg/thorvg/issues/1372
Since longjmp is not yet fully supported on Wasm (emscripten uses
JS to make the jumps and requires slow jumps), but also because
longjmp is hard to reason about, this patch replaces it with
return values. The logic stays exactly the same.
certain systems, may not support file I/O operations.
ThorVG should provide users with an option to configure
builds according to their requirements.
This ensures that file I/O calls are avoided,
preventing potential crashes.
Please use the meson '-Dfile=true/false' option for this.
Please note that "THORVG_FILE_IO_SUPPORT" might be expected
for your thorvg manual build.
issue: https://github.com/thorvg/thorvg/issues/3008
Support cross compile the GL backend code into WASM.
The code needs WebGL 2.0 API so, the compile flags contains `MAX_WEBGL_VERSION` and `FULL_ES3`
Also add binding code to initialize the WebGL context and GLCanvas.
- prevent dangling instances in failure scenarios by utilizing a reference counter.
- update the test suite to treat null pointer arguments as invalid argument failures.
Remove the requirement for unique_ptr in the function prototypes.
This change will simplify the API usage, making it more streamlined
and user-friendly. However, memory management will now be the
responsibility of the user.
C++ API Modification:
- Result Paint::mask(std::unique_ptr<Paint> target, MaskMethod method) -> Result Paint::mask(Paint* target, MaskMethod method)
- Result Paint::clip(std::unique_ptr<Paint> clipper) -> Result Paint::clip(Paint* clipper)
- virtual Result Canvas::push(std::unique_ptr<Paint> paint) -> virtual Result Canvas::push(Paint* paint)
- std::unique_ptr<LinearGradient> LinearGradient::gen() -> LinearGradient* LinearGradient::gen()
- std::unique_ptr<RadialGradient> RadialGradient::gen() -> RadialGradient* RadialGradient::gen()
- Result Shape::strokeFill(std::unique_ptr<Fill> f) -> Result Shape::strokeFill(Fill* f)
- Result Shape::fill(std::unique_ptr<Fill> f) -> Result Shape::fill(Fill* f)
- std::unique_ptr<Shape> Shape::gen() -> Shape* Shape::gen()
- std::unique_ptr<Picture> Picture::gen() -> Result Picture::push(Paint* paint)
- std::unique_ptr<Scene> Scene::gen() -> Scene* Scene::gen()
- Result Text::fill(std::unique_ptr<Fill> f) -> Result Text::fill(Fill* f)
- std::unique_ptr<Text> Text::gen() -> Text* Text::gen()
- std::unique_ptr<SwCanvas> SwCanvas::gen() -> SwCanvas* SwCanvas::gen()
- std::unique_ptr<GlCanvas> GlCanvas::gen() -> GlCanvas* GlCanvas::gen()
- std::unique_ptr<Animation> Animation::gen() -> Animation* Animation::gen()
- Result Saver::background(std::unique_ptr<Paint> paint) -> Result Saver::background(Paint* paint)
- Result Saver::save(std::unique_ptr<Paint> paint, const char* filename, uint32_t quality = 100) -> Result Saver::save(Paint* paint, const char* filename, uint32_t quality = 100)
- std::unique_ptr<Saver> Saver::gen() -> Saver* Saver::gen()
- std::unique_ptr<Accessor> Accessor::gen() -> Accessor* Accessor::gen()
C++ API removal:
- template<typename T = tvg::Paint> std::unique_ptr<T> cast(Paint* paint)
- template<typename T = tvg::Paint> std::unique_ptr<T> cast(Paint* paint)
issue: https://github.com/thorvg/thorvg/issues
Shape's bbox represents only fill's render region.
Render region of fill and stroke is stored in SwTask.
This was visible while rendering shapes with a stroke -
fill was to big.
@Issue: https://github.com/thorvg/thorvg/issues/2908
Co-Authored-By: Hermet Park <hermet@lottiefiles.com>
Since the translate API was used while text updating,
the subsequent range selector transformations gets overwritten
when updating the shape (scale and rotate, adding another
translation will persist). This caused unexpected results.
Fixed by using the transform API when additional transformations
are needed - also fixes applying more than one range selector.
WebGL has a strict rule that does not allow the same GLBuffer
to be bound to both ARRAY_BUFFER and ELEMENT_ARRAY_BUFFER at the same time.
(https://registry.khronos.org/webgl/specs/latest/1.0/#5.14.5)
To support WebGL in the future, a separate GLBuffer is used to store index data.
Used native hardware MSAA x4
Full multisampling support including custom blend and compositions.
Must be verified on web and 4K resolution for performance issues
