Rather than aiming for extremely high accuracy,
a compromise can achieve with better performance.
This modification helps prevent unnecessary image rotation.
issue: https://github.com/thorvg/thorvg/issues/2265
[issues 1479: Fill Rule](#1479)
In this solution we dont need to find silhouette, that is not a cheep operation (decreasing performance in 2 times)
For winding, you can select separate operations for front and back faces (increment for front, decrement for back)
After rendering the fan, the value in the stencil buffer will be the winding number. You can fill the appropriate portion by rendering a screen-sized quad with stencil testing enabled and the stencil function set according to which winding rule you wish to use.
For even-odd, you don't need to distinguish front and back faces; you can just use INVERT as the operation.
The viewport function defines the rectangular area of the canvas
that will be used for drawing operations.
It is used to clip the rendering output to the boundaries of the rectangle.
Apps can use this function to set the drawing region within the canvas.
When the ThorVG canvas is partially inside the screen area such as during scrolling
it could help enhance rendering performance.
New Experimental API:
- Result Canvas::viewport(int32_t x, int32_t y, int32_t w, int32_t h) noexcept;
Issue: https://github.com/thorvg/thorvg/issues/2274
For angles close to multiples of pi/2, precision
based on FLT_EPSILON was insufficient. It either
led to the creation of an additional cubic segment
filled with erroneous values or NaNs, or it resulted
in the drawing of an angle for a sweep close to 0.
* restrict the scissor box of composite task
* do not tessellate stroke or fill geometry if there is no Fill or
Stroke color
* use actually transformed curve to calculate polyline count when doing
curve flatten
Fix some error:
* glViewport not controlled by framebuffer, and need to set manually when
target framebuffer is changed.
* change even-odd stencil operation, so no need to do third draw call to
clear stencil buffer
* fix the missing `GlCanvas::update` calls in Lottoe and LottieExtension
[issues 1479: Masking, InvMasking, LumaMasking, InvLumaMasking](#1479)
Computes composition and blending using simgle pass istead of two full screen passes
Using a mask (any type) with alpha set to less
than 255 through the fill(r, g, b, a) API resulted
in incorrect compositions of fill and stroke.
Incorrect results were also observed for luma masks,
as their alpha is calculated based on the other color
channels.
@issue: https://github.com/thorvg/thorvg/issues/1653
* Fix error when handle GradientTransform calculation. And move the inv
calculation into gradient vertex shader.
* Fix cubic tessellation not close the last point
[issues 1479: lottie](#1479)
- optimaze stroking algorithm by prevent vector normalizations
- using render meshes heaps to reduce webgpu instances re-creations
- using single instance of pilylines for standard operations such as trim and split
- "on-the-fly" strokes generations with dash pattern
- "on-the-fly" mesh objects generation in a process of path decoding
- merge strokes into single mesh by each shape
The error was visible when multiple shapes were
simultaneously trimmed. The length of a single
shape was zeroed out only in selected cases,
which caused accumulation that could lead to
incorrect extension of the variable determining
the maximum length.
@issue: https://github.com/thorvg/thorvg/issues/2173
Since blit msaa framebuffer to another msaa framebuffer may generate
GLError in some platforms. Use normal texture rendering to blit the final
color buffer onto target framebuffer.
A single animation might have a desinated markers with naming: 0 ~ 0.5 (sector A), 0.5 ~ 1.0 (sector B). Selecting one of them using a marker name(sector A) and could play only that part with animation controllers.
usage:
- `animation->segment("sectionA") // Named segment(Marker)`
- `auto cnt = animation->markerCnt()`
- `auto name = animation->markers(index)`
- `animation->segment(0, 0.5) // Segment`
- `animation->segment(&begin, &end)`
Co-authored-by: Jinny You <jinny@lottiefiles.com>
[issues 1479: lottie](#1479)
Vertex, Index and uniform buffers now updates instead of recreate.
Implemented pools form mesh objects and render shapes data
it increase performance in 30-40% in massive animations scenes
GL might need to generate a default target FBO
when the given target ID indicates the main surface.
However, users may want to draw visuals directly onto the main surface.
This policy must be reviewed thoroughly.
The Lottie loader now utilizes an individual scene instance per loader.
This guarantees that each instance retains a unique frame number,
essential for maintaining the integrity of animation states.
As a result, sharing loaders between Lottie animations is prohibited
to uphold frame context isolation.
This may be subject to optimization review in the future.
[issues 1479: lottie](#1479)
To optimize bled operations hardware pipeline blend stage are used for some blend methods:
BlendMethod::SrcOver
BlendMethod::Normal
BlendMethod::Add
BlendMethod::Multiply
BlendMethod::Darken
BlendMethod::Lighten
Other types compute shaders used
There are differences in behavior compared to the SVG spec,
especially when consecutive line drawings occur without a moveTo command
following a closePath command.
Actually, thorvg didn't care the behavior in that scenario,
this update ensures the scenario is handled correctly
to align with the SVG specification.
issue: https://github.com/thorvg/thorvg/issues/1487
