diff --git a/src/core/Sk4px.h b/src/core/Sk4px.h index e1d4dc1244..7d51fd099e 100644 --- a/src/core/Sk4px.h +++ b/src/core/Sk4px.h @@ -10,6 +10,7 @@ #include "SkNx.h" #include "SkColor.h" +#include "SkColorPriv.h" // This file may be included multiple times by .cpp files with different flags, leading // to different definitions. Usually that doesn't matter because it's all inlined, but @@ -47,6 +48,14 @@ public: void store2(SkPMColor[2]) const; void store1(SkPMColor[1]) const; + // Same as above for 565. + static Sk4px Load4(const SkPMColor16 src[4]); + static Sk4px Load2(const SkPMColor16 src[2]); + static Sk4px Load1(const SkPMColor16 src[1]); + void store4(SkPMColor16 dst[4]) const; + void store2(SkPMColor16 dst[2]) const; + void store1(SkPMColor16 dst[1]) const; + // 1, 2, or 4 SkPMColors with 16-bit components. // This is most useful as the result of a multiply, e.g. from mulWiden(). class Wide : public Sk16h { @@ -99,8 +108,8 @@ public: // A generic driver that maps fn over a src array into a dst array. // fn should take an Sk4px (4 src pixels) and return an Sk4px (4 dst pixels). - template - static void MapSrc(int n, SkPMColor* dst, const SkPMColor* src, const Fn& fn) { + template + static void MapSrc(int n, Dst* dst, const SkPMColor* src, const Fn& fn) { // This looks a bit odd, but it helps loop-invariant hoisting across different calls to fn. // Basically, we need to make sure we keep things inside a single loop. while (n > 0) { @@ -129,8 +138,8 @@ public: } // As above, but with dst4' = fn(dst4, src4). - template - static void MapDstSrc(int n, SkPMColor* dst, const SkPMColor* src, const Fn& fn) { + template + static void MapDstSrc(int n, Dst* dst, const SkPMColor* src, const Fn& fn) { while (n > 0) { if (n >= 8) { Sk4px dst0 = fn(Load4(dst+0), Load4(src+0)), @@ -157,8 +166,8 @@ public: } // As above, but with dst4' = fn(dst4, src4, alpha4). - template - static void MapDstSrcAlpha(int n, SkPMColor* dst, const SkPMColor* src, const SkAlpha* a, + template + static void MapDstSrcAlpha(int n, Dst* dst, const SkPMColor* src, const SkAlpha* a, const Fn& fn) { while (n > 0) { if (n >= 8) { @@ -189,6 +198,39 @@ private: typedef Sk16b INHERITED; }; +// TODO: specialize these per-backend + +inline Sk4px Sk4px::Load4(const SkPMColor16 src[4]) { + SkPMColor src32[4]; + for (int i = 0; i < 4; i++) { src32[i] = SkPixel16ToPixel32(src[i]); } + return Load4(src32); +} +inline Sk4px Sk4px::Load2(const SkPMColor16 src[2]) { + SkPMColor src32[2]; + for (int i = 0; i < 2; i++) { src32[i] = SkPixel16ToPixel32(src[i]); } + return Load2(src32); +} +inline Sk4px Sk4px::Load1(const SkPMColor16 src[1]) { + SkPMColor src32 = SkPixel16ToPixel32(src[0]); + return Load1(&src32); +} + +inline void Sk4px::store4(SkPMColor16 dst[4]) const { + SkPMColor dst32[4]; + this->store4(dst32); + for (int i = 0; i < 4; i++) { dst[i] = SkPixel32ToPixel16(dst32[i]); } +} +inline void Sk4px::store2(SkPMColor16 dst[2]) const { + SkPMColor dst32[2]; + this->store2(dst32); + for (int i = 0; i < 2; i++) { dst[i] = SkPixel32ToPixel16(dst32[i]); } +} +inline void Sk4px::store1(SkPMColor16 dst[1]) const { + SkPMColor dst32; + this->store1(&dst32); + dst[0] = SkPixel32ToPixel16(dst32); +} + } // namespace #ifdef SKNX_NO_SIMD diff --git a/src/core/Sk4pxXfermode.h b/src/core/Sk4pxXfermode.h index 0c8dcb5302..e912d1ecb8 100644 --- a/src/core/Sk4pxXfermode.h +++ b/src/core/Sk4pxXfermode.h @@ -222,6 +222,19 @@ public: } } + void xfer16(uint16_t dst[], const SkPMColor src[], int n, const SkAlpha aa[]) const override { + if (NULL == aa) { + Sk4px::MapDstSrc(n, dst, src, [&](const Sk4px& dst4, const Sk4px& src4) { + return ProcType::Xfer(src4, dst4); + }); + } else { + Sk4px::MapDstSrcAlpha(n, dst, src, aa, + [&](const Sk4px& dst4, const Sk4px& src4, const Sk4px& alpha) { + return xfer_aa(src4, dst4, alpha); + }); + } + } + private: SkT4pxXfermode(const ProcCoeff& rec) : INHERITED(rec, ProcType::kMode) {} @@ -237,19 +250,35 @@ public: void xfer32(SkPMColor dst[], const SkPMColor src[], int n, const SkAlpha aa[]) const override { for (int i = 0; i < n; i++) { - SkPMFloat s(src[i]), - d(dst[i]), - b(ProcType::Xfer(s,d)); - if (aa) { - // We do aa in full float precision before going back down to bytes, because we can! - SkPMFloat a = Sk4f(aa[i]) * Sk4f(1.0f/255); - b = b*a + d*(Sk4f(1)-a); - } - dst[i] = b.round(); + dst[i] = aa ? this->xfer32(dst[i], src[i], aa[i]) + : this->xfer32(dst[i], src[i]); + } + } + + void xfer16(uint16_t dst[], const SkPMColor src[], int n, const SkAlpha aa[]) const override { + for (int i = 0; i < n; i++) { + SkPMColor dst32 = SkPixel16ToPixel32(dst[i]); + dst32 = aa ? this->xfer32(dst32, src[i], aa[i]) + : this->xfer32(dst32, src[i]); + dst[i] = SkPixel32ToPixel16(dst32); } } private: + inline SkPMColor xfer32(SkPMColor dst, SkPMColor src) const { + return ProcType::Xfer(SkPMFloat(src), SkPMFloat(dst)).round(); + } + + inline SkPMColor xfer32(SkPMColor dst, SkPMColor src, SkAlpha aa) const { + SkPMFloat s(src), + d(dst), + b(ProcType::Xfer(s,d)); + // We do aa in full float precision before going back down to bytes, because we can! + SkPMFloat a = Sk4f(aa) * Sk4f(1.0f/255); + b = b*a + d*(Sk4f(1)-a); + return b.round(); + } + SkTPMFloatXfermode(const ProcCoeff& rec) : INHERITED(rec, ProcType::kMode) {} typedef SkProcCoeffXfermode INHERITED;