NOTE: Linear blending is currently disabled in this CL as the
feature is still a work in progress
Android currently performs all blending (any kind of linear math
on colors really) on gamma-encoded colors. Since Android assumes
that the default color space is sRGB, all bitmaps and colors
are encoded with the sRGB Opto-Electronic Conversion Function
(OECF, which can be approximated with a power function). Since
the power curve is not linear, our linear math is incorrect.
The result is that we generate colors that tend to be too dark;
this affects blending but also anti-aliasing, gradients, blurs,
etc.
The solution is to convert gamma-encoded colors back to linear
space before doing any math on them, using the sRGB Electo-Optical
Conversion Function (EOCF). This is achieved in different
ways in different parts of the pipeline:
- Using hardware conversions when sampling from OpenGL textures
or writing into OpenGL frame buffers
- Using software conversion functions, to translate app-supplied
colors to and from sRGB
- Using Skia's color spaces
Any type of processing on colors must roughly ollow these steps:
[sRGB input]->EOCF->[linear data]->[processing]->OECF->[sRGB output]
For the sRGB color space, the conversion functions are defined as
follows:
OECF(linear) :=
linear <= 0.0031308 ? linear * 12.92 : (pow(linear, 1/2.4) * 1.055) - 0.055
EOCF(srgb) :=
srgb <= 0.04045 ? srgb / 12.92 : pow((srgb + 0.055) / 1.055, 2.4)
The EOCF is simply the reciprocal of the OECF.
While it is highly recommended to use the exact sRGB conversion
functions everywhere possible, it is sometimes useful or beneficial
to rely on approximations:
- pow(x,2.2) and pow(x,1/2.2)
- x^2 and sqrt(x)
The latter is particularly useful in fragment shaders (for instance
to apply dithering in sRGB space), especially if the sqrt() can be
replaced with an inversesqrt().
Here is a fairly exhaustive list of modifications implemented
in this CL:
- Set TARGET_ENABLE_LINEAR_BLENDING := false in BoardConfig.mk
to disable linear blending. This is only for GLES 2.0 GPUs
with no hardware sRGB support. This flag is currently assumed
to be false (see note above)
- sRGB writes are disabled when entering a functor (WebView).
This will need to be fixed at some point
- Skia bitmaps are created with the sRGB color space
- Bitmaps using a 565 config are expanded to 888
- Linear blending is disabled when entering a functor
- External textures are not properly sampled (see below)
- Gradients are interpolated in linear space
- Texture-based dithering was replaced with analytical dithering
- Dithering is done in the quantization color space, which is
why we must do EOCF(OECF(color)+dither)
- Text is now gamma corrected differently depending on the luminance
of the source pixel. The asumption is that a bright pixel will be
blended on a dark background and the other way around. The source
alpha is gamma corrected to thicken dark on bright and thin
bright on dark to match the intended design of fonts. This also
matches the behavior of popular design/drawing applications
- Removed the asset atlas. It did not contain anything useful and
could not be sampled in sRGB without a yet-to-be-defined GL
extension
- The last column of color matrices is converted to linear space
because its value are added to linear colors
Missing features:
- Resource qualifier?
- Regeneration of goldeng images for automated tests
- Handle alpha8/grey8 properly
- Disable sRGB write for layers with external textures
Test: Manual testing while work in progress
Bug: 29940137
Change-Id: I6a07b15ab49b554377cd33a36b6d9971a15e9a0b
Bug: 30442298
We already do this for initialize(), fix
it so that update() is parallel with the
UI thread as well.
Shaves ~7ms off of the 99th percentile on
NotificationShade open & close
Change-Id: I1791df495453fb9e1e12362c68e3d20e837e62be
Bug: 30342017
Upload recents thumbnails in the
dead gaps between frames instead of
at the start of a frame. This eliminates
jank caused by the large texture
upload.
Change-Id: I507cd286d199109c7a9a1511d68ba5ab5d28069f
There is only one caller each for the static functions here so this
CL moves the logic to the caller. Also by moving some of the code
into the pipeline it makes it easier for future changes to configure
how a pipeline handles a layer.
Change-Id: Ib735b5154325cbb658fd151f7a19dbf434ab44b7
Add a new mode, controlled by sys.use_fifo_ui property, that enables the
top app's UI and RenderThread to be SCHED_FIFO. This eliminates almost
all jank due to scheduling competition with non-UI critical
threads. This mode may not be suitable for all devices.
bug 24503801
Change-Id: I7b8a31830ad80f7efa00236928d5476998ed4e00
Add a system property debug.hwui.default_renderer, which allows
to set rendering mode to OpenGL (default), Skia OpenGL or Vulkan.
Change-Id: I8bca5bacc5108f77437e340ac61f2d8db8cc4c39
Adds googlebench output format support
Adds offscreen rendering for >60fps benchmarking
Adds 'all' alias to run all registered TestScenes
Change-Id: I2579e40f2f4c941bfbd90c75efbee384c08a116b
Bug: 27286867
WindowManager has committed to stopped state
controlling the lifecycle of the Surface, so
make that a first-class thing in HWUI as well.
This makes it more resistent to things like
a rogue updateSurface() happening while mStopped=true,
leading to bad things down the line. Instead let
the surface be changed/updated as often as desired,
and just block any attempt to draw on that surface.
Also removes some unnecessary makeCurrent()s, as
EglManager ensures that we *always* have a valid
GL context now (using a pbuffer surface if there is
no window surface set)
Change-Id: Iead78ddebc7997e8fdb0c9534836352f5e54b9bd
Bug: 27709981
This desperately needs a refactor, but to keep
the current (really needed & nice) behavior of
dispatching after sync finishes would be difficult
to handle cleanly without lots of ripping so... #yolo
Change-Id: I831a06c6ae7412a062720d68ecbe3085190f0258
Bug: 27922347
* Dump the full histogram
* Expand the histogram to have a slow-frame section with
large 50ms buckets to raise the cap to 5s to give more
insight into system-health
* Stop excluding first-frame metrics as we want to include
those in our global tracking. Automated tests already filter
these out by doing resets before running anyway.
Change-Id: Idaba8aad591f59d10a6477b11efc0767ff715083
- Rename to FrameMetrics to avoid collision with existing
android.view.FrameStats class
- Make FrameMetricsObserver implementation detail,
exposing FrameMetricsListener interface as public API
and wrapping in FrameStatsObserver to maintain state
- Remove dropped frame count call, in favor of passing as
parameter to callback method.
- Move away from raw timestamp access in favor of Metric IDs
which represent higher-level, more stable stages in a frame
lifecycle and match the categories exposed in the onscreen
bars.
- Support many-to-many Window<->FrameMetricsListener relationship
Change-Id: I00e741d664d4c868b1b6d0131a23f8316bd8c5c2
getMaximumBitmapWidth() and getMaximumBitmapHeight() of DisplayListCanvas
need HWUI cache instance. Since the initialization of the cache is
asynchronous it may crash if not yet ready. Add a staticFence() call
to guarantee the cache has been created prior issuing the call.
Change-Id: I5ed9e5cc084444c8d1872a77fef50e294ae14e93
Signed-off-by: Thomas Buhot <thomas.buhot@intel.com>
Signed-off-by: Zhiquan Liu <zhiquan.liu@intel.com>
On the critical path of the cold launch of applications
the main thread of the started application tells the RenderThread
to create a surface. This process is synchronous and blocks
the main thread of the application until the creation
of the EGLContext is complete.
As a consequence the launch time of the application is delayed
by time spent allocating the EGL Context in the RenderThread.
With this optimization the launch time of any application
is improved (for example settings by 20 to 40 ms).
Change-Id: Ibf47aaa0abb8dedf7aa00693073db3785d9d6b08
Signed-off-by: Thomas Buhot <thomas.buhot@intel.com>
Signed-off-by: Zhiquan Liu <zhiquan.liu@intel.com>
Using a multi threaded render node to render the window frame
asynchronously from the application relayout.
Bug: 22527834
Bug: 24400680
Bug: 24459827
Bug: 24409773
Bug: 24537510
Change-Id: I1010fc6a8b6e38424178140afa3ca124433ab7e4
This is adding the renderer side infrastructure to allow
rendering multiple render nodes with different threads.
This is a pre-step for decoupling a non client decor
resize reder from a content resize render.
Multiple render nodes can be added to be drawn, and to
prevent overdrawing, a content bounds area can be set
Bug: 22527834
Change-Id: Ie7271e20895bf38957e5a84aeefc883e282039ad
bug:20063841
Restores old SkShader matrix behavior from before the Glop refactor.
Many drawing operations draw without sending the canvas transform to
the GL shader. In such cases, we need to adapt the matrix sent to the
SkShader logic to invert the canvas transform that's built into
the mesh.
Change-Id: I42b6f59df36ce46436322b95bf9ad2140795ee58
