This PR implements an image comparison algorithm in the NTSC YIQ color
space, as described in:
Measuring perceived color difference using YIQ NTSC
transmission color space in mobile applications.
Yuriy Kotsarenko, Fernando Ramos.
An electronic version is available at:
- http://www.progmat.uaem.mx:8080/artVol2Num2/Articulo3Vol2Num2.pdf
This should allow the image comparison to be a tad more robust than
comparing plain uint8 pixel values.
Signed-off-by: Sebastien Binet <s@sbinet.org>
The old renderer is still the default, so the new compute renderer will only be
used in the rare case the old renderer is not supported but the new is. That
happens on the Samsung J2 Prime and Moto C Android phones. Or set the
GIORENDERER environment variable to "forcecompute" to disable the old renderer:
$ GIORENDERER=forcecompute go run ...
Missing features:
- Gradients are not supported yet, and render as a solid color.
- Draw timers are not added, and profile.Events are not emitted.
- Stroked paths may in some cases appear corrupted because their clip
outlines are not continuous when generated by Gio. Sebastien is
working on a fix.
- The new renderer shares most CPU-side logic with the old renderer,
resulting in several inefficient conversion steps between the old
operations representation and the new. This is slower, but minimizes
divergence in features and bugs between the two renderers.
Roadmap:
- The compute renderer supports features that Gio does not yet
exploit: stroked paths with round caps, transformations, lines,
cubic beziér curves.
- More stroke styles and maybe dashed strokes natively in shaders.
- Metal and Direct3D ports.
The most important feature is porting the renderer to run on the CPU. A
CPU renderer will both support Gio on devices with insufficient GPU
support, and allow us to remove the old renderer. Two renderers is twice
the maintenance but the feature set of the weakest implementation.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
Modern graphics APIs have immutable objects, with mutable data. For example,
a texture's dimensions are immutable, while the texture contents is not.
Change the GPU API abstraction to match.
Clearing a Texture is convenient to do with a plain []byte. Generalize
Texture.Upload to take a plain byte slice and introduce a helper function for
uploading *image.RGBA data.
Add TextureFormatRGBA8, a format for the linear RGB colorspace.
Add OpenGL ES 3.1 functions for compute programs.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
Modern graphics APIs have immutable objects, with mutable data. For example,
a texture's dimensions are immutable, while the texture contents is not.
Change the GPU API abstraction to match.
Clearing a Texture is convenient to do with a plain []byte. Generalize
Texture.Upload to take a plain byte slice and introduce a helper function for
uploading *image.RGBA data.
Add TextureFormatRGBA8 a format for the linear RGB colorspace.
Add OpenGL ES 3.1 functions for compute programs.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
Before this change, the entire byte buffer would be passed to WebGL, even
though its size may be (much) larger than the source data.
Remakably, this hasn't resulted in any problems until the use of
glBufferSubData, which require the passed data fits the buffer size.
As a bonus, this fix should speed up the wasm port by virtue of passing
much less data across the wasm<->js barrier.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
The Mesa software OpenGL implementation strays enough from the
reference values that tests fail. Relax the tests to make them
pass again.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
This CL introduces 2 new path builders:
- Outline which takes a PathSpec to be outlined
- Stroke which takes a PathSpec and a stroke style, to stroke a path.
typically, code like this:
var p clip.Path
...
p.Outline().Add(o)
should be replaced with:
var p clip.Path
...
clip.Outline{Path: p.End()}.Op().Add(o)
similarly, stroking should be modified from:
var p clip.Path
...
p.Stroke(width, clip.StrokeStyle{...}).Add(o)
to:
var p clip.Path
...
clip.Stroke{Path: p.End(), Style: clip.StrokeStyle{Width:...}}.Op().Add(o)
here are tentative 'rf' scripts (see rsc.io/rf for more details):
```
ex {
import "gioui.org/op";
import "gioui.org/op/clip";
var p clip.Path;
var o *op.Ops;
p.Outline().Add(o) -> clip.Outline{Path:p.End()}.Op().Add(o);
}
ex {
import "gioui.org/op";
import "gioui.org/op/clip";
var o *op.Ops;
var p clip.Path;
var sty clip.StrokeStyle;
var width float32;
p.Stroke(width, sty).Add(o) -> \
clip.Stroke{ \
Path:p.End(), \
Style: clip.StrokeStyle{ \
Width: width, \
}}.Op().Add(o);
}
```
Signed-off-by: Sebastien Binet <s@sbinet.org>
Previously, the only way to manipulate the clipboard (read or write) is
using the `app.Window`.
The new `clipboard.ReadOp` and `clipboard.WriteOp`makes possible to
read/write from the widget.
Signed-off-by: Inkeliz <inkeliz@inkeliz.com>
As a consequence, most API is gone from gpu/gl, and embedding Gio in
foreign frameworks don't need to provide an OpenGL implementation.
The next change simplifies the GLFW embedding example accordingly.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
The existing implementation cannot remove the focus of some widget,
doesn't have an option to focus without display the on-screen keyboard
and it automatically focuses the first InputOp, aggressively.
That change aims to make possible: remove focus from any widget. Add
focus without displaying the on-screen-keyboard/soft keyboard. Don't
automatically focus any widget. Don't recover focus when the widget is
visible again.
Fixes gio#180.
Signed-off-by: Inkeliz <inkeliz@inkeliz.com>
Currently BCE is unable to understand that the accesses in the code are
safe. Added an explicit slice to make the length bounds obvious.
Signed-off-by: Egon Elbre <egonelbre@gmail.com>
Using delta position with Line and Quad can drift over successive calls.
Also, in some cases it's much more convenient to use absolute
coordinates rather than relative.
Signed-off-by: Egon Elbre <egonelbre@gmail.com>
color.RGBA has two problems with regards to using it.
First the color values need to be premultiplied, whereas most APIs
have non-premultiplied values. This is mainly to preserve color components
with low alpha values.
Second there are two ways to premultiply with sRGB. One is to premultiply
after sRGB conversion, the other is before. This makes using the API more
confusing.
Using color.NRGBA in sRGB makes it align with CSS.e
Signed-off-by: Egon Elbre <egonelbre@gmail.com>
I don't know why the 1/2 factor is there, but it leads to images being
rendered with a 0.5 pixel offset.
Remove the other useless checks while here: clipping 1px images shouldn't
be a problem and the destination rectangle is always non-zero (otherwise
it wouldn't be rendered).
Update the reference images that are subtly changed because of this fix.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
Flat and Square caps are implemented.
Bevel joins are implemented.
Round caps, Round joins and Miter joins are left for another PR.
Signed-off-by: Sebastien Binet <s@sbinet.org>
PaintOp.Rect is the wrong abstraction; it implies a clip operation
better handled by package clip, and not all paints need it (colors).
Furthermore, it's awkward to specify a PaintOp that fills up the
current clip area, regardless of its size.
Redefine PathOp to mean "fill current clip area".
API change. Replace uses of PaintOp.Rect with a TransformOp applied
before the PaintOp.
Leave a TODO for the PathOp infinity area.
Fixes gio#167
Signed-off-by: Elias Naur <mail@eliasnaur.com>
We're about to remove PaintOp.Rect. Replacing PaintOps with Fill or
FillShape where possible will ease the transition.
Using Fill in tests exposed a problem with the infinity in paint.Fill.
Adjust it for now; it will be removed later.
Updates gio#167
Signed-off-by: Elias Naur <mail@eliasnaur.com>
This is effectively a revert of commit gioui.org/commit/69dfd2e3a5541.
ImageOp.Rect is the wrong abstraction; it implies a clipping operation that is
better handled by package clip.
API change. Uses of ImageOp.Rect should apply a clip.Rect before the PaintOp,
or use image.RGBA.SubImage (or similar).
Signed-off-by: Elias Naur <mail@eliasnaur.com>
Egon Elbre pointed out that a difference of 20 means a 10% difference.
Lowering the tolerance to 5 didn't work on my setup; leave it at 10.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
Package material's ad-hoc mulAlpha didn't take the sRGB color-space
into account, which meant that alpha-scaled colors were subtly wrong.
Introduce f32color.MulAlpha and convert all uses to it.
Thanks to René Post for finding and debugging the issue.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
Fixes a bug due to that f32.Rect.Intersect will not return the
empty rectangle for non intersecting rectangles - but instead
a swapped rectangle. By removing the .Canon() call in gpu.go we
ensure that non overlapping clipping rects and paint rects will
lead to no painting.
The Canon() call is not needed since boundsForTransformedRect()
was previously updated to always return a canonical rectangle.
Test case added.
Signed-off-by: Viktor <viktor.ogeman@gmail.com>
Use op.Offset instead, or create and manipulate a f32.Affine2D.
API change. Update your code with a gofmt rule:
gofmt -r 'op.TransformOp{}.Offset -> op.Offset'
Signed-off-by: Elias Naur <mail@eliasnaur.com>
This commit fixes a bug where a shape first drawn off-screen
and later moved into screen would not display properly. Since we
cache CPU operations (vertex transform / construction) we need to
upload the constructed data to the GPU after it was build, or a later
frame will use non-initialized memory for it's draw call.
Note that this fix removes the optimization of not processing clip
paths outside the screen - but this is assumed to be uncommon except
when it is first drawn off screen to later be moved in (e.g. in a scrolling list)
in which case we do want to upload the data and prepare for that later
call.
This commit also does a few minor clean ups and adds a test case.
Signed-off-by: Viktor <viktor.ogeman@gmail.com>
To avoid duplicate work when using macros and non-offset transforms,
cache also the new bounding boxes set up for them. The ops.Reader
already generates Keys for all operations, so use them in the cache.
Signed-off-by: Viktor <viktor.ogeman@gmail.com>
Uses app/headless to create a set of test cases for drawing operations, including clipping
textures and transforms. This commit tests for approximate pixel matches, if future changes affect
local drawing operations it will be easy to change the reference images, it thus becomes and
should be an intentional operation if changes lead to local changes in drawn results.
Ideally we should be able to make the tests check for exact pixel matches down the line to ensure
consistent results between platforms.
Signed-off-by: Viktor <viktor.ogeman@gmail.com>
Add support for affine transformations. The key changes are outlined
below.
- Painting/clipping with rectangles is handled by, for complex
transforms, creating clipping paths representing the transformed
rectangle and using a larger bounding box. Cover/Blit shaders updated
correspondingly to correctly map texture cordinates from the new
bounding boxes.
- Since path splitting must happen on CPU the transforms must happen CPU
side as well - offsets removed from shaders.
- Complex transforms will lead to different path splitting which means
that GPU arrays can no longer be cached if the transform has changed.
Thus the current transform is added as a key to the cache.
- Add a public API to op for setting Affine transformations.
There are a number of optimizations that could be explored further but
which are left out now:
- Caching also of CPU operations (e.g path splitting & transforms) and
not only caching the GPU arrays.
- Allow for re-use of cached GPU vertices if the transformation change
is a pure offset / scaling since the splitting is then the same.
Signed-off-by: Viktor <viktor.ogeman@gmail.com>
Sebastien Binet