diff --git a/libswscale/output.c b/libswscale/output.c index d185dfc34e..30217c7b63 100644 --- a/libswscale/output.c +++ b/libswscale/output.c @@ -1372,13 +1372,69 @@ YUV2RGBWRAPPERX(yuv2, rgb_full, rgb4_byte_full, AV_PIX_FMT_RGB4_BYTE, 0) YUV2RGBWRAPPERX(yuv2, rgb_full, bgr8_full, AV_PIX_FMT_BGR8, 0) YUV2RGBWRAPPERX(yuv2, rgb_full, rgb8_full, AV_PIX_FMT_RGB8, 0) +static void +yuv2gbrp_full_X_c(SwsContext *c, const int16_t *lumFilter, + const int16_t **lumSrc, int lumFilterSize, + const int16_t *chrFilter, const int16_t **chrUSrc, + const int16_t **chrVSrc, int chrFilterSize, + const int16_t **alpSrc, uint8_t **dest, + int dstW, int y) +{ + int i; + int hasAlpha = 0; + + for (i = 0; i < dstW; i++) { + int j; + int Y = 1<<9; + int U = (1<<9)-(128 << 19); + int V = (1<<9)-(128 << 19); + int R, G, B, A; + + for (j = 0; j < lumFilterSize; j++) { + Y += lumSrc[j][i] * lumFilter[j]; + } + for (j = 0; j < chrFilterSize; j++) { + U += chrUSrc[j][i] * chrFilter[j]; + V += chrVSrc[j][i] * chrFilter[j]; + } + Y >>= 10; + U >>= 10; + V >>= 10; + if (hasAlpha) { + A = 1 << 18; + for (j = 0; j < lumFilterSize; j++) { + A += alpSrc[j][i] * lumFilter[j]; + } + A >>= 19; + if (A & 0x100) + A = av_clip_uint8(A); + } + Y -= c->yuv2rgb_y_offset; + Y *= c->yuv2rgb_y_coeff; + Y += 1 << 21; + R = Y + V*c->yuv2rgb_v2r_coeff; + G = Y + V*c->yuv2rgb_v2g_coeff + U*c->yuv2rgb_u2g_coeff; + B = Y + U*c->yuv2rgb_u2b_coeff; + if ((R | G | B) & 0xC0000000) { + R = av_clip_uintp2(R, 30); + G = av_clip_uintp2(G, 30); + B = av_clip_uintp2(B, 30); + } + + dest[0][i] = G >> 22; + dest[1][i] = B >> 22; + dest[2][i] = R >> 22; + } +} + av_cold void ff_sws_init_output_funcs(SwsContext *c, yuv2planar1_fn *yuv2plane1, yuv2planarX_fn *yuv2planeX, yuv2interleavedX_fn *yuv2nv12cX, yuv2packed1_fn *yuv2packed1, yuv2packed2_fn *yuv2packed2, - yuv2packedX_fn *yuv2packedX) + yuv2packedX_fn *yuv2packedX, + yuv2anyX_fn *yuv2anyX) { enum AVPixelFormat dstFormat = c->dstFormat; const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(dstFormat); @@ -1484,8 +1540,11 @@ av_cold void ff_sws_init_output_funcs(SwsContext *c, case AV_PIX_FMT_RGB8: *yuv2packedX = yuv2rgb8_full_X_c; break; + case AV_PIX_FMT_GBRP: + *yuv2anyX = yuv2gbrp_full_X_c; + break; } - if(!*yuv2packedX) + if (!*yuv2packedX && !*yuv2anyX) goto YUV_PACKED; } else { YUV_PACKED: diff --git a/libswscale/swscale.c b/libswscale/swscale.c index 8227ff50b2..4d9bd89ce5 100644 --- a/libswscale/swscale.c +++ b/libswscale/swscale.c @@ -369,6 +369,7 @@ static int swScale(SwsContext *c, const uint8_t *src[], yuv2packed1_fn yuv2packed1 = c->yuv2packed1; yuv2packed2_fn yuv2packed2 = c->yuv2packed2; yuv2packedX_fn yuv2packedX = c->yuv2packedX; + yuv2anyX_fn yuv2anyX = c->yuv2anyX; const int chrSrcSliceY = srcSliceY >> c->chrSrcVSubSample; const int chrSrcSliceH = -((-srcSliceH) >> c->chrSrcVSubSample); int should_dither = is9_OR_10BPS(c->srcFormat) || @@ -557,7 +558,7 @@ static int swScale(SwsContext *c, const uint8_t *src[], /* hmm looks like we can't use MMX here without overwriting * this array's tail */ ff_sws_init_output_funcs(c, &yuv2plane1, &yuv2planeX, &yuv2nv12cX, - &yuv2packed1, &yuv2packed2, &yuv2packedX); + &yuv2packed1, &yuv2packed2, &yuv2packedX, &yuv2anyX); use_mmx_vfilter= 0; } @@ -630,7 +631,7 @@ static int swScale(SwsContext *c, const uint8_t *src[], dstW, c->lumDither8, 0); } } - } else { + } else if (yuv2packedX) { av_assert1(lumSrcPtr + vLumFilterSize - 1 < lumPixBuf + vLumBufSize * 2); av_assert1(chrUSrcPtr + vChrFilterSize - 1 < chrUPixBuf + vChrBufSize * 2); if (c->yuv2packed1 && vLumFilterSize == 1 && @@ -657,6 +658,13 @@ static int swScale(SwsContext *c, const uint8_t *src[], chrUSrcPtr, chrVSrcPtr, vChrFilterSize, alpSrcPtr, dest[0], dstW, dstY); } + } else { + av_assert1(!yuv2packed1 && !yuv2packed2); + yuv2anyX(c, vLumFilter + dstY * vLumFilterSize, + lumSrcPtr, vLumFilterSize, + vChrFilter + dstY * vChrFilterSize, + chrUSrcPtr, chrVSrcPtr, vChrFilterSize, + alpSrcPtr, dest, dstW, dstY); } } } @@ -695,7 +703,7 @@ static av_cold void sws_init_swScale_c(SwsContext *c) ff_sws_init_output_funcs(c, &c->yuv2plane1, &c->yuv2planeX, &c->yuv2nv12cX, &c->yuv2packed1, - &c->yuv2packed2, &c->yuv2packedX); + &c->yuv2packed2, &c->yuv2packedX, &c->yuv2anyX); ff_sws_init_input_funcs(c); diff --git a/libswscale/swscale_internal.h b/libswscale/swscale_internal.h index 4f4b9531ee..9f9cc4a790 100644 --- a/libswscale/swscale_internal.h +++ b/libswscale/swscale_internal.h @@ -223,6 +223,40 @@ typedef void (*yuv2packedX_fn)(struct SwsContext *c, const int16_t *lumFilter, const int16_t **alpSrc, uint8_t *dest, int dstW, int y); +/** + * Write one line of horizontally scaled Y/U/V/A to YUV/RGB + * output by doing multi-point vertical scaling between input pixels. + * + * @param c SWS scaling context + * @param lumFilter vertical luma/alpha scaling coefficients, 12bit [0,4096] + * @param lumSrc scaled luma (Y) source data, 15bit for 8-10bit output, + * 19-bit for 16bit output (in int32_t) + * @param lumFilterSize number of vertical luma/alpha input lines to scale + * @param chrFilter vertical chroma scaling coefficients, 12bit [0,4096] + * @param chrUSrc scaled chroma (U) source data, 15bit for 8-10bit output, + * 19-bit for 16bit output (in int32_t) + * @param chrVSrc scaled chroma (V) source data, 15bit for 8-10bit output, + * 19-bit for 16bit output (in int32_t) + * @param chrFilterSize number of vertical chroma input lines to scale + * @param alpSrc scaled alpha (A) source data, 15bit for 8-10bit output, + * 19-bit for 16bit output (in int32_t) + * @param dest pointer to the output planes. For 16bit output, this is + * uint16_t + * @param dstW width of lumSrc and alpSrc in pixels, number of pixels + * to write into dest[] + * @param y vertical line number for this output. This does not need + * to be used to calculate the offset in the destination, + * but can be used to generate comfort noise using dithering + * or some output formats. + */ +typedef void (*yuv2anyX_fn)(struct SwsContext *c, const int16_t *lumFilter, + const int16_t **lumSrc, int lumFilterSize, + const int16_t *chrFilter, + const int16_t **chrUSrc, + const int16_t **chrVSrc, int chrFilterSize, + const int16_t **alpSrc, uint8_t **dest, + int dstW, int y); + /* This struct should be aligned on at least a 32-byte boundary. */ typedef struct SwsContext { /** @@ -437,6 +471,7 @@ typedef struct SwsContext { yuv2packed1_fn yuv2packed1; yuv2packed2_fn yuv2packed2; yuv2packedX_fn yuv2packedX; + yuv2anyX_fn yuv2anyX; /// Unscaled conversion of luma plane to YV12 for horizontal scaler. void (*lumToYV12)(uint8_t *dst, const uint8_t *src, const uint8_t *src2, const uint8_t *src3, @@ -768,7 +803,8 @@ void ff_sws_init_output_funcs(SwsContext *c, yuv2interleavedX_fn *yuv2nv12cX, yuv2packed1_fn *yuv2packed1, yuv2packed2_fn *yuv2packed2, - yuv2packedX_fn *yuv2packedX); + yuv2packedX_fn *yuv2packedX, + yuv2anyX_fn *yuv2anyX); void ff_sws_init_swScale_altivec(SwsContext *c); void ff_sws_init_swScale_mmx(SwsContext *c); diff --git a/libswscale/utils.c b/libswscale/utils.c index e96f255255..9e2dfc0bb0 100644 --- a/libswscale/utils.c +++ b/libswscale/utils.c @@ -185,7 +185,7 @@ static const FormatEntry format_entries[AV_PIX_FMT_NB] = { [AV_PIX_FMT_YUV444P12LE] = { 1, 1 }, [AV_PIX_FMT_YUV444P14BE] = { 1, 1 }, [AV_PIX_FMT_YUV444P14LE] = { 1, 1 }, - [AV_PIX_FMT_GBRP] = { 1, 0 }, + [AV_PIX_FMT_GBRP] = { 1, 1 }, [AV_PIX_FMT_GBRP9LE] = { 1, 0 }, [AV_PIX_FMT_GBRP9BE] = { 1, 0 }, [AV_PIX_FMT_GBRP10LE] = { 1, 0 }, @@ -1019,11 +1019,21 @@ av_cold int sws_init_context(SwsContext *c, SwsFilter *srcFilter, c->flags = flags; } } + if(dstFormat == AV_PIX_FMT_GBRP) { + if (!(flags & SWS_FULL_CHR_H_INT)) { + av_log(c, AV_LOG_DEBUG, + "%s output is not supported with half chroma resolution, switching to full\n", + av_get_pix_fmt_name(dstFormat)); + flags |= SWS_FULL_CHR_H_INT; + c->flags = flags; + } + } /* reuse chroma for 2 pixels RGB/BGR unless user wants full * chroma interpolation */ if (flags & SWS_FULL_CHR_H_INT && isAnyRGB(dstFormat) && + dstFormat != AV_PIX_FMT_GBRP && dstFormat != AV_PIX_FMT_RGBA && dstFormat != AV_PIX_FMT_ARGB && dstFormat != AV_PIX_FMT_BGRA && diff --git a/tests/ref/lavfi/field b/tests/ref/lavfi/field index 2d4a6b06ab..4ac1417acf 100644 --- a/tests/ref/lavfi/field +++ b/tests/ref/lavfi/field @@ -15,6 +15,7 @@ bgr565be bf955b9a035af0e613cf1de249f55f9d bgr565le 6dd85cd5e19266c53a54cbcf06d396a7 bgr8 9669f6974f0fc1c0afa1c7d4df093c0b bgra f7cabae31dd7465dab2203f45db646f8 +gbrp 55eab469e3e02ccc5a358565d9b1eaa7 gray 66a09b53f7d3f79dcb6096f3ec3740c5 gray16be a447af6482b922c9997ac02e5d3535f1 gray16le c1dd0db327295898ff282d07f48c105d diff --git a/tests/ref/lavfi/pixfmts_copy b/tests/ref/lavfi/pixfmts_copy index 340b3a9567..bedb352b74 100644 --- a/tests/ref/lavfi/pixfmts_copy +++ b/tests/ref/lavfi/pixfmts_copy @@ -15,6 +15,7 @@ bgr565be 13a36d6502be88fc0c2aec05b8d2d501 bgr565le ed027571692aecd522aa65a90cc7e09b bgr8 71ef789609c746c2e7e4be9dec29062c bgra 0364b074268682ea46168742a8239f7d +gbrp 89d6e4b116e3bd542fa09a19a977ad16 gray 1e5e2b8548843a6898eedd9c974c422c gray16be 389f4e5a8ab413b3af32767b59ed7f9e gray16le a1f912941247e45b394b9cf4f0e81130 diff --git a/tests/ref/lavfi/pixfmts_null b/tests/ref/lavfi/pixfmts_null index 340b3a9567..bedb352b74 100644 --- a/tests/ref/lavfi/pixfmts_null +++ b/tests/ref/lavfi/pixfmts_null @@ -15,6 +15,7 @@ bgr565be 13a36d6502be88fc0c2aec05b8d2d501 bgr565le ed027571692aecd522aa65a90cc7e09b bgr8 71ef789609c746c2e7e4be9dec29062c bgra 0364b074268682ea46168742a8239f7d +gbrp 89d6e4b116e3bd542fa09a19a977ad16 gray 1e5e2b8548843a6898eedd9c974c422c gray16be 389f4e5a8ab413b3af32767b59ed7f9e gray16le a1f912941247e45b394b9cf4f0e81130 diff --git a/tests/ref/lavfi/pixfmts_pad b/tests/ref/lavfi/pixfmts_pad index 9c319744e0..192ff6a430 100644 --- a/tests/ref/lavfi/pixfmts_pad +++ b/tests/ref/lavfi/pixfmts_pad @@ -5,6 +5,7 @@ argb e5fbb7bb282a80897b8f730627f68876 bgr0 c55368036cccbb0af471d6bd82abe02a bgr24 67f9fd70dc6d9896b7122976b33932b4 bgra c8dd017b5a3b55e8b9d0ac1cdcf327bd +gbrp 74f83deee9866bbdce3f91fa2aeddaaa gray b1abadae3718522aa57a7972da8cbe17 rgb0 b1977b45634c4db58a183a07feb2acff rgb24 e73de9dc0fdd78f4853c168603cc7aba diff --git a/tests/ref/lavfi/pixfmts_pixdesctest b/tests/ref/lavfi/pixfmts_pixdesctest index 95db3de59d..9d0ac4a1ab 100644 --- a/tests/ref/lavfi/pixfmts_pixdesctest +++ b/tests/ref/lavfi/pixfmts_pixdesctest @@ -15,6 +15,7 @@ bgr565be 13a36d6502be88fc0c2aec05b8d2d501 bgr565le ed027571692aecd522aa65a90cc7e09b bgr8 71ef789609c746c2e7e4be9dec29062c bgra 0364b074268682ea46168742a8239f7d +gbrp 89d6e4b116e3bd542fa09a19a977ad16 gray 1e5e2b8548843a6898eedd9c974c422c gray16be 389f4e5a8ab413b3af32767b59ed7f9e gray16le a1f912941247e45b394b9cf4f0e81130 diff --git a/tests/ref/lavfi/pixfmts_scale b/tests/ref/lavfi/pixfmts_scale index a85fbdd9b8..237004a148 100644 --- a/tests/ref/lavfi/pixfmts_scale +++ b/tests/ref/lavfi/pixfmts_scale @@ -15,6 +15,7 @@ bgr565be fca6f07daf23d9dd84381dd4c9afd959 bgr565le f524e9f16bdd68b247dbcb621e543fc0 bgr8 68a3a395043dc57335ad1f8e891229c5 bgra 3eaf5489b8aa13a3388aad3751b597bf +gbrp 7b83ae32c1f76bd634e50f4797a74e92 gray 045c35e6cc7d41460f2b96b3e254deab gray16be 70064f9acdc5e3935ccda67e765bf2fb gray16le 578241fb43029e5ae841a3c94d940dce diff --git a/tests/ref/lavfi/pixfmts_vflip b/tests/ref/lavfi/pixfmts_vflip index 89ce078074..dcb8ceee9d 100644 --- a/tests/ref/lavfi/pixfmts_vflip +++ b/tests/ref/lavfi/pixfmts_vflip @@ -15,6 +15,7 @@ bgr565be 7100c2ddfee42e7efafec1ccefecf7c6 bgr565le 9fab295d966386d4ef99d5b43066da47 bgr8 275ce12eeb05de67a6915f67cbb43ce5 bgra d29c35871248c476c366e678db580982 +gbrp 7852e9bbe52cfad9bab8081a5c7a5a31 gray a53528cdf80ed31c9897042a2ea16618 gray16be 9b23f3e79c54a6ccb62e0135a32e3045 gray16le 93cfa8fbb2a86ead275ce1817444e6d5