This change also fixes an issue with RGBA16F bitmaps when modulated
with a color (for instance by setting an alpha on the Paint object).
The color space conversion is currently done entirely in the shader,
by doing these operations in order:
1. Sample the texture
2. Un-premultiply alpha
3. Apply the EOTF
4. Multiply by the 3x3 color space matrix
5. Apply the OETF
6. Premultiply alpha
Optimizations:
- Steps 2 & 6 are skipped for opaque (common) bitmaps
- Step 3 is skipped when the color space's EOTF is close
to sRGB (Display P3 for instance). Instead, we use
a hardware sRGB fetch (when the GPU supports it)
- When step 3 is necessary, we use one of four standard
EOTF implementations, to save cycles when possible:
+ Linear (doesn't do anything)
+ Full parametric (ICC parametric curve type 4 as defined
in ICC.1:2004-10, section 10.15)
+ Limited parametric (ICC parametric curve type 3)
+ Gamma (ICC parametric curve type 0)
Color space conversion could be done using texture samplers
instead, for instance 3D LUTs, with or without transfer
functions baked in, or 1D LUTs for transfer functions. This
would result in dependent texture fetches which may or may
not be an advantage over an ALU based implementation. The
current solution favor the use of ALUs to save precious
bandwidth.
Test: CtsUiRenderingTests, CtsGraphicsTests
Bug: 32984164
Change-Id: I10bc3db515e13973b45220f129c66b23f0f7f8fe
RGBA16F bitmaps are always encoded in linear space, which means we must
apply the opto-electronic transfer function before we can render them
in the framebuffer.
Since our linear bitmaps are assumed to be scRGB, values can be negative.
The OETF is a slightly modified sRGB OETF:
sign(x) * OETF_sRGB(abs(x))
This effectively mirrors the OETF over the negative domain.
This CL also removes the "optimized" shader generation path. With
current compilers, the optimized path doesn't do anything of value
and makes ProgramCache difficult to maintain. Shader compilers inline
everything and are really good at folding expressions and removing
unused code.
Bug: 32984164
Test: CtsUiRenderingTestCases
Change-Id: Ieb458ad53574e3a8959aa6bccbbd2d1fe203cbc5
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
Remove Clang cutout for unused parameters. Fix warnings.
Remove Clang cutout for deprecated Skia function usage. Has been
fixed in the L push.
Change-Id: I7ea073ff67127cc1e14e798b655e2c50615fe8e7
bug:16852257
Use pow(alpha, 1.5) to avoid harsh edges on shadow alpha ramps.
Also adjusts shadow constants to compensate.
Change-Id: I5869956d7d292db2a8e496bc320084b6d64c3fb7
Because the caching of projection matrix didn't account for changes in
the offset flag, the flag could be ignored. Now we use both to verify
that the cached matrix can be used.
Change-Id: I193b94eaf0b98f046a6484f0866c3d25048653fd
bug:4351353
bug:8185479
Point tessellation is similar to line special case, except that we
only tessellate one point (as a circle or rect) and duplicate it
across other instances.
Additionally:
Fixes square caps for AA=false lines
Cleanup in CanvasCompare, disabling interpolation on zoomed-in
comparison view
Change-Id: I0756fcc4b20f77878fed0d8057297c80e82ed9dc
The counter can be enabled by setting the system property called
debug.hwui.overdraw to the string "count". If the string is set
to "show", overdraw will be highlighted on screen instead of
printing out a simple counter.
Change-Id: I9a9c970d54bffab43138bbb7682f6c04bc2c40bd
When the Android runtime starts, the system preloads a series of assets
in the Zygote process. These assets are shared across all processes.
Unfortunately, each one of these assets is later uploaded in its own
OpenGL texture, once per process. This wastes memory and generates
unnecessary OpenGL state changes.
This CL introduces an asset server that provides an atlas to all processes.
Note: bitmaps used by skia shaders are *not* sampled from the atlas.
It's an uncommon use case and would require extra texture transforms
in the GL shaders.
WHAT IS THE ASSETS ATLAS
The "assets atlas" is a single, shareable graphic buffer that contains
all the system's preloaded bitmap drawables (this includes 9-patches.)
The atlas is made of two distinct objects: the graphic buffer that
contains the actual pixels and the map which indicates where each
preloaded bitmap can be found in the atlas (essentially a pair of
x and y coordinates.)
HOW IS THE ASSETS ATLAS GENERATED
Because we need to support a wide variety of devices and because it
is easy to change the list of preloaded drawables, the atlas is
generated at runtime, during the startup phase of the system process.
There are several steps that lead to the atlas generation:
1. If the device is booting for the first time, or if the device was
updated, we need to find the best atlas configuration. To do so,
the atlas service tries a number of width, height and algorithm
variations that allows us to pack as many assets as possible while
using as little memory as possible. Once a best configuration is found,
it gets written to disk in /data/system/framework_atlas
2. Given a best configuration (algorithm variant, dimensions and
number of bitmaps that can be packed in the atlas), the atlas service
packs all the preloaded bitmaps into a single graphic buffer object.
3. The packing is done using Skia in a temporary native bitmap. The
Skia bitmap is then copied into the graphic buffer using OpenGL ES
to benefit from texture swizzling.
HOW PROCESSES USE THE ATLAS
Whenever a process' hardware renderer initializes its EGL context,
it queries the atlas service for the graphic buffer and the map.
It is important to remember that both the context and the map will
be valid for the lifetime of the hardware renderer (if the system
process goes down, all apps get killed as well.)
Every time the hardware renderer needs to render a bitmap, it first
checks whether the bitmap can be found in the assets atlas. When
the bitmap is part of the atlas, texture coordinates are remapped
appropriately before rendering.
Change-Id: I8eaecf53e7f6a33d90da3d0047c5ceec89ea3af0
Add a key manipulation that makes black text/paths use the standard
simple bitmap/patch shader, since they are the same. Previously we'd
create a separate shader for each because the keys differed, even
though the shaders were functionally equivalent.
Also fixes some issues around setting DEBUG_PROGRAM
Change-Id: I0c77c684d58da03501ee9ab8239c7d4a70fd6b5c
This change adds a new property called "debug.hwui.show_stencil_clip"
that accepts the following values:
- "highlight", colorizes in green any drawing command that's tested
against a non-rectangular clip region
- "region", shows the non-rectangular clip region in blue every time
it is used
- "hide", default value, nothing is shown
Change-Id: I83c8602310edc4aaeb8b905371cdd185b17d32b5
bug:7117155
bug:8114304
Currently used for lines (with and without AA) and arcs with useCenter=false
Also removes 0.375, 0.375 offset for AA lines
Change-Id: Ic8ace418739344db1e2814edf65253fe7448b0b0
Bug #7239634
This change passes two matrices to the vertex shader instead of one.
We used to compute the final MVP matrix on the CPU to minimize the
number of operations in the vertex shaders. Shader compilers are
however smart enough to perform this optimization for us. Since we
need the MV matrix to properly compute gradients dithering, this
change splits the MVP matrix into two. This has the advantage of
removing one matrix multiplication per drawing operation on the
CPU.
The SGX 540 shader compiler produces the same number of instructions
in both cases. There is no penalty hit with having two matrices
instead of one. We also send so few vertices per frame that it
does not matter very much.
Change-Id: I17d47ac4772615418e0e1885b97493d31435a936
Modulation is normally enabled in a shader when drawing with an alpha
mask (A8 texture.) Modulation is used to do one of two things:
- Colorize the primitive (to draw text in red for instance)
- Apply extra translucency (50% translucent circle filled with a bitmap)
The current implementation has four issues:
1. Unnecessary work is performed by assigning the modulation color
to vec4 fragColor early in the shader
2. The modulation color's alpha is applied twice when the primitive
is drawn with an SkShader
3. The decision to modulate is wrong and triggers when any of the
RGB channels is < 1.0. Only the alpha channel needs to be taken
into account to make the decision
4. Gamma correction is not applied properly
This change addresses all four issues above.
Change-Id: I73fcc74efc4b094bf2d1b835f10ffaa2ea4b9eb9
Instead of calculating opacity from relative position in the shader, use a
shader varying to do this computation for us.
bug:5045101
Also adds a test to HwAccelerationTest to show incorrect antialiasing in
scaled drawAARect / boundarySize calculation.
Change-Id: Icdc41acb01dc10ce354834f8389a5aed2f439162
To enable it, the system property ro.hwui.text_gamma_shader must be
set to true. For testing, DEBUG_FONT_RENDERER_FORCE_SHADER_GAMMA
can be set to 1 in libhwui/Debug.h.
Change-Id: If345c6b71b67ecf1ef2e8847b71f30f3ef251a27
I don't know who's to blame, SGX or Tegra2 but one of those two GPUs is not
following the OpenGL ES 2.0 spec.
Change-Id: I2624e0efbc9c57d571c55c8b440a5e43f08a54f2
All accelerated lines are now rendered as quads. Hairlines used to
be rendered as GL_LINES, but these lines don't render the same as our
non-accelerated lines, so we're using quads for everything. Also, fixed
a bug in the way that we were offsetting quads (and not offseting points)
to ensure that our lines/points actuall start on the same pixels as
Skia's.
Change-Id: I51b923cc08a9858444c430ba07bc8aa0c83cbe6a
This optimization is currently disabled until Launcher is
modified to take advantage of it. The optimization can be
enabled by turning on RENDER_LAYERS_AS_REGIONS in the
OpenGLRenderer.h file.
Change-Id: I2fdf59d0f4dc690a3d7f712173ab8db3848b27b1
The generator supports features that are not yet implement in the
renderer: color matrix, lighting, porterduff color blending and
composite shaders.
This change also adds support for repeated/mirrored non-power of 2
bitmap shaders.
Change-Id: I903a11a070c0eb9cc8850a60ef305751e5b47234
With this change, all the vertex and fragment shaders used by the GL
renderer are now generated based on a program description supplied
by the caller. This allows the renderer to generate a large number
of shaders without having to write all the possible combinations by
hand. The generated shaders are stored in a program cache.
Change-Id: If54d286e77ae021c724d42090da476df12a18ebb
This change adds a new DrawLinearGradientProgram class to enable the drawing
of linear gradients. Two new vertex and fragment shaders are introduced,
based on DrawTextureProgram's shaders.
Change-Id: I885afc076bb6cef8cd3962ae21a086fa6a03bf96
Add support for the following drawing functions:
- drawBitmap(int[]...)
- drawPaint()
Optimizes shader state changes by enabling/disabling attribute arrays
only when needed.
Adds quick rejects when drawing trivial shapes to avoid unnecessary
OpenGL operations.
Change-Id: Ic2c6c2ed1523d08a63a8c95601a1ec40b6c7fbc9
