Separate simd implementation by files to maintain them easier.
Now we have avx, c, neon version implementation base,
we can add implementations to them.
Changes:
- Prepare neon verison of ALPHA_BLEND API.
- Use ALPHA_BLEND_NEON in _translucentRle
Notes:
- _translucentRle with neon support reduces execution time of this
function ~ 300 % (measured on uint32_t 400 x 400 buffer).
- API was tested on ARMv7l device with GCC 9.2 based toolchain. Results
on other devices could be different.
* Apply bilinear interpolation to images
Apply bilinear interpolation when drawing images with transformation
If the mapped coordinate value is a floating point value,
bilinear interpolation is performed using adjacent pixel values.
Interpolation is not performed in cases when the color values to beinterpolated
are the same or scale down case.
Pixel-based image picture doesn't work at size() method.
Obvisouly, we missed to implement it properly.
This patch corrects it.
@Issue: https://github.com/Samsung/thorvg/issues/656
Changes:
- Added 'neon' vector option in build system
- Introduced neon version of rasterRGBA32 API, which improves
speed of the funciton on ARM cpu's around ~35%
if many canvas instances own private memory pool,
the memory usage can be increased linearly.
To prevent memory usage, flush out memory pool from the clear()
if the canvas uses private memory pool.
We can replace the system logger method by changing single line print source in common,
This also helps to remove the THORVG_LOG_ENABLED macro from each use-cases.
TVGLOG(): To print the hint & tip messages for users.
TVGERR(): To print the error message for debugging.
@Issues: https://github.com/Samsung/thorvg/issues/36
For filled and stroked shapes with a gradient, the opacity
was applied during the composition and during the ctable update.
To prevent this, in case the composition occured, ctable
is updated with the opacity = 255.
The opacity value is interpolated as are all color channels.
It's similar to the reverted commit ac95433, but with two slight changes
(alpha_blend was missing in the ctable init and 'a' was used in an if
statement instead of 'a2').
tvg sw_engine tries to skip composition as far as its doable,
it tries composition only if the translucent shape with stroking is there.
There has been a condition bug that composition is applied unnecessarily
even if stroking is disabled. This resulted in broken blending
at gradient filling since it applied alpha values multiple times.
Issues:
https://github.com/Samsung/thorvg/issues/445
When canvas width was smaller than its stride and a composition is used,
the plots were drawn incorrectly. This is fixed by setting the image width
(image from a compositor) to be equal to the surface stride, instead of its width.
The image data are now allocated accordingly the fixed image width.
changed alpha channel data type to 32 bits from 8 bits,
since subsequent data operations requires 32 bits values.
this 8 bits (since channel range is up to 255) doesn't helpful
for saving memory size because it would generate additional data casting by compiler.
I compared the binary size and this patch saves about 600bytes.
Calculations accuracy in ALPHA_BLEND function has been
improved. Until now blending resulted in a slight hue change
(all color channels affected). The chosen method of calculation
is a compromise between the accuracy and the performance.
a corner case is detected that anti-aliasing is missing at shapes
if the shape has the dash-style stroke.
By adding the condition, it applies anti-aliasing properly.
@Issues: 394
The translucent rastering function is split into 3 other (instead of if/else statement).
An additional function is introduced to decide which one of the 3 should be called.
This refactoring is done to preserve the convention used for all other rastering functs.
The translucent rastering function is split into 3 other (instead of if/else statement).
An additional function is introduced to decide which one of the 3 should be called.
This refactoring is done to preserve the convention used for all other rastering functs.
The translucent rastering function is split into 3 other (instead of if/else statement).
An additional function is introduced to decide which one of the 3 should be called.
This refactoring is done to preserve the convention used for all other rastering functs.
The translucent rastering function is split into 3 other (instead of if/else statement).
An additional function is introduced to decide which one of the 3 should be called.
This refactoring is done to preserve the convention used for all other rastering functs.
The cases with gradient shapes with composition are handled
in the same function as gradint shapes with opacity < 255.
Parts of the code from _rasterOpaque... grad functions moved to
_rasterTranslucent... grad functions.
The cases when composition was applied were handled only for AlphaMask and
InvAlphaMask. When opacity value was to be < 255, there was no code
to handle this.
We have encountered that multi-threading usage that user creates,
multiple canvases owned by multiple user threads.
Current sw_engine memory pool has been considered only for multi-threads,
spawned by tvg task scheduler.
In this case it's safe but when user threads introduced, it can occur race-condition.
Thus, Here is a renewal policy that non-threading tvg(initialized threads with zero),
takes care of multiple user threads bu changing its policy,
each of canvases should have individual memory pool to guarantee mutual-exclusion.
@API additions
enum MempoolPolicy
{
Default = 0, ///< Default behavior that ThorVG is designed to.
Shareable, ///< Memory Pool is shared among the SwCanvases.
Individual ///< Allocate designated memory pool that is only used by current instance.
};
Result SwCanvas::mempool(MempoolPolicy policy) noexcept;
All in all, if user calls multiple threads, set memory pool policy to Individual.
