1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
|
/*
* Copyright 2017 Google Inc.
*
* Use of this source code is governed by a BSD-style license that can be
* found in the LICENSE file.
*/
#include "gm.h"
#include "SkSurface.h"
// This GM reproduces skia:6783, which demonstrated a bug in repeat and mirror
// image sampling tiling modes as implemented in software. We want to tile to
// [0,limit), and the old incorrect logic was:
//
// limit = ulp_before(limit)
// val = val - floor(val/limit)*limit (This is repeat; mirror is similar.)
//
// while the correct logic is more like:
//
// val = val - floor(val/limit)*limit
// val = min(val, ulp_before(limit))
//
// You would see ugly jaggies on the blue/yellow edge near the bottom left if
// the bug were still present. All stripes should now look roughly the same.
DEF_SIMPLE_GM(bug6783, canvas, 500, 500) {
sk_sp<SkSurface> surface = SkSurface::MakeRasterN32Premul(100, 100);
SkPaint p;
p.setColor(SK_ColorYELLOW);
surface->getCanvas()->drawPaint(p);
p.setColor(SK_ColorBLUE);
surface->getCanvas()->drawRect(SkRect::MakeWH(50, 100), p);
sk_sp<SkImage> img = surface->makeImageSnapshot();
SkMatrix m = SkMatrix::Concat(SkMatrix::MakeTrans(25, 214),
SkMatrix::MakeScale(2, 2));
m.preSkew(0.5f, 0.5f);
// The bug was present at all filter levels, but you might not notice it at kNone.
p.setFilterQuality(kLow_SkFilterQuality);
// It's only important to repeat or mirror in x to show off the bug.
p.setShader(img->makeShader(SkShader::kRepeat_TileMode, SkShader::kClamp_TileMode, &m));
canvas->drawPaint(p);
}
|