2 * Copyright (C) 2015 Pedro Arthur <bygrandao@gmail.com>
4 * This file is part of FFmpeg.
6 * FFmpeg is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU Lesser General Public
8 * License as published by the Free Software Foundation; either
9 * version 2.1 of the License, or (at your option) any later version.
11 * FFmpeg is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 * Lesser General Public License for more details.
16 * You should have received a copy of the GNU Lesser General Public
17 * License along with FFmpeg; if not, write to the Free Software
18 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
20 #include "swscale_internal.h"
22 static int lum_planar_vscale(SwsContext *c, SwsFilterDescriptor *desc, int sliceY, int sliceH)
24 VScalerContext *inst = desc->instance;
25 int dstW = desc->dst->width;
27 int first = FFMAX(1-inst->filter_size, inst->filter_pos[sliceY]);
28 int sp = first - desc->src->plane[0].sliceY;
29 int dp = sliceY - desc->dst->plane[0].sliceY;
30 uint8_t **src = desc->src->plane[0].line + sp;
31 uint8_t **dst = desc->dst->plane[0].line + dp;
32 uint16_t *filter = inst->filter[0] + (inst->isMMX ? 0 : sliceY * inst->filter_size);
34 if (inst->filter_size == 1)
35 ((yuv2planar1_fn)inst->pfn)((const int16_t*)src[0], dst[0], dstW, c->lumDither8, 0);
37 ((yuv2planarX_fn)inst->pfn)(filter, inst->filter_size, (const int16_t**)src, dst[0], dstW, c->lumDither8, 0);
40 int sp = first - desc->src->plane[3].sliceY;
41 int dp = sliceY - desc->dst->plane[3].sliceY;
42 uint8_t **src = desc->src->plane[3].line + sp;
43 uint8_t **dst = desc->dst->plane[3].line + dp;
44 uint16_t *filter = inst->filter[1] + (inst->isMMX ? 0 : sliceY * inst->filter_size);
46 if (inst->filter_size == 1)
47 ((yuv2planar1_fn)inst->pfn)((const int16_t*)src[0], dst[0], dstW, c->lumDither8, 0);
49 ((yuv2planarX_fn)inst->pfn)(filter, inst->filter_size, (const int16_t**)src, dst[0], dstW, c->lumDither8, 0);
55 static int chr_planar_vscale(SwsContext *c, SwsFilterDescriptor *desc, int sliceY, int sliceH)
57 const int chrSkipMask = (1 << desc->dst->v_chr_sub_sample) - 1;
58 if (sliceY & chrSkipMask)
61 VScalerContext *inst = desc->instance;
62 int dstW = FF_CEIL_RSHIFT(desc->dst->width, desc->dst->h_chr_sub_sample);
63 int chrSliceY = sliceY >> desc->dst->v_chr_sub_sample;
65 int first = FFMAX(1-inst->filter_size, inst->filter_pos[chrSliceY]);
66 int sp1 = first - desc->src->plane[1].sliceY;
67 int sp2 = first - desc->src->plane[2].sliceY;
68 int dp1 = chrSliceY - desc->dst->plane[1].sliceY;
69 int dp2 = chrSliceY - desc->dst->plane[2].sliceY;
70 uint8_t **src1 = desc->src->plane[1].line + sp1;
71 uint8_t **src2 = desc->src->plane[2].line + sp2;
72 uint8_t **dst1 = desc->dst->plane[1].line + dp1;
73 uint8_t **dst2 = desc->dst->plane[2].line + dp2;
74 uint16_t *filter = inst->filter[0] + (inst->isMMX ? 0 : chrSliceY * inst->filter_size);
77 ((yuv2interleavedX_fn)inst->pfn)(c, filter, inst->filter_size, (const int16_t**)src1, (const int16_t**)src2, dst1[0], dstW);
78 } else if (inst->filter_size == 1) {
79 ((yuv2planar1_fn)inst->pfn)((const int16_t*)src1[0], dst1[0], dstW, c->chrDither8, 0);
80 ((yuv2planar1_fn)inst->pfn)((const int16_t*)src2[0], dst2[0], dstW, c->chrDither8, 3);
82 ((yuv2planarX_fn)inst->pfn)(filter, inst->filter_size, (const int16_t**)src1, dst1[0], dstW, c->chrDither8, 0);
83 ((yuv2planarX_fn)inst->pfn)(filter, inst->filter_size, (const int16_t**)src2, dst2[0], dstW, c->chrDither8, inst->isMMX ? (c->uv_offx2 >> 1) : 3);
90 static int packed_vscale(SwsContext *c, SwsFilterDescriptor *desc, int sliceY, int sliceH)
92 VScalerContext *inst = desc->instance;
93 int dstW = desc->dst->width;
94 int chrSliceY = sliceY >> desc->dst->v_chr_sub_sample;
96 int lum_fsize = inst[0].filter_size;
97 int chr_fsize = inst[1].filter_size;
98 uint16_t *lum_filter = inst[0].filter[0];
99 uint16_t *chr_filter = inst[1].filter[0];
101 int firstLum = FFMAX(1-lum_fsize, inst[0].filter_pos[chrSliceY]);
102 int firstChr = FFMAX(1-chr_fsize, inst[1].filter_pos[chrSliceY]);
104 int sp0 = firstLum - desc->src->plane[0].sliceY;
105 int sp1 = firstChr - desc->src->plane[1].sliceY;
106 int sp2 = firstChr - desc->src->plane[2].sliceY;
107 int sp3 = firstLum - desc->src->plane[3].sliceY;
108 int dp = sliceY - desc->dst->plane[0].sliceY;
109 uint8_t **src0 = desc->src->plane[0].line + sp0;
110 uint8_t **src1 = desc->src->plane[1].line + sp1;
111 uint8_t **src2 = desc->src->plane[2].line + sp2;
112 uint8_t **src3 = desc->alpha ? desc->src->plane[3].line + sp3 : NULL;
113 uint8_t **dst = desc->dst->plane[0].line + dp;
116 if (c->yuv2packed1 && lum_fsize == 1 && chr_fsize <= 2) { // unscaled RGB
117 int chrAlpha = chr_fsize == 1 ? 0 : chr_filter[2 * sliceY + 1];
118 ((yuv2packed1_fn)inst->pfn)(c, (const int16_t*)*src0, (const int16_t**)src1, (const int16_t**)src2, (const int16_t*)(desc->alpha ? *src3 : NULL), *dst, dstW, chrAlpha, sliceY);
119 } else if (c->yuv2packed2 && lum_fsize == 2 && chr_fsize == 2) { // bilinear upscale RGB
120 int lumAlpha = lum_filter[2 * sliceY + 1];
121 int chrAlpha = chr_filter[2 * sliceY + 1];
123 c->lumMmxFilter[3] = lum_filter[2 * sliceY] * 0x10001;
125 c->chrMmxFilter[3] = chr_filter[2 * chrSliceY] * 0x10001;
126 ((yuv2packed2_fn)inst->pfn)(c, (const int16_t**)src0, (const int16_t**)src1, (const int16_t**)src2, (const int16_t**)src3,
127 *dst, dstW, lumAlpha, chrAlpha, sliceY);
128 } else { // general RGB
129 ((yuv2packedX_fn)inst->pfn)(c, lum_filter + sliceY * lum_fsize,
130 (const int16_t**)src0, lum_fsize, chr_filter + sliceY * chr_fsize,
131 (const int16_t**)src1, (const int16_t**)src2, chr_fsize, (const int16_t**)src3, *dst, dstW, sliceY);
136 static int any_vscale(SwsContext *c, SwsFilterDescriptor *desc, int sliceY, int sliceH)
138 VScalerContext *inst = desc->instance;
139 int dstW = desc->dst->width;
140 int chrSliceY = sliceY >> desc->dst->v_chr_sub_sample;
142 int lum_fsize = inst[0].filter_size;
143 int chr_fsize = inst[1].filter_size;
144 uint16_t *lum_filter = inst[0].filter[0];
145 uint16_t *chr_filter = inst[1].filter[0];
147 int firstLum = FFMAX(1-lum_fsize, inst[0].filter_pos[chrSliceY]);
148 int firstChr = FFMAX(1-chr_fsize, inst[1].filter_pos[chrSliceY]);
150 int sp0 = firstLum - desc->src->plane[0].sliceY;
151 int sp1 = firstChr - desc->src->plane[1].sliceY;
152 int sp2 = firstChr - desc->src->plane[2].sliceY;
153 int sp3 = firstLum - desc->src->plane[3].sliceY;
154 int dp0 = sliceY - desc->dst->plane[0].sliceY;
155 int dp1 = chrSliceY - desc->dst->plane[1].sliceY;
156 int dp2 = chrSliceY - desc->dst->plane[2].sliceY;
157 int dp3 = sliceY - desc->dst->plane[3].sliceY;
159 uint8_t **src0 = desc->src->plane[0].line + sp0;
160 uint8_t **src1 = desc->src->plane[1].line + sp1;
161 uint8_t **src2 = desc->src->plane[2].line + sp2;
162 uint8_t **src3 = desc->alpha ? desc->src->plane[3].line + sp3 : NULL;
163 uint8_t *dst[4] = { desc->dst->plane[0].line[dp0],
164 desc->dst->plane[1].line[dp1],
165 desc->dst->plane[2].line[dp2],
166 desc->alpha ? desc->dst->plane[3].line[dp3] : NULL };
168 av_assert1(!c->yuv2packed1 && !c->yuv2packed2);
169 ((yuv2anyX_fn)inst->pfn)(c, lum_filter + sliceY * lum_fsize,
170 (const int16_t**)src0, lum_fsize, chr_filter + sliceY * chr_fsize,
171 (const int16_t**)src1, (const int16_t**)src2, chr_fsize, (const int16_t**)src3, dst, dstW, sliceY);
177 int ff_init_vscale(SwsContext *c, SwsFilterDescriptor *desc, SwsSlice *src, SwsSlice *dst)
179 VScalerContext *lumCtx = NULL;
180 VScalerContext *chrCtx = NULL;
182 if (isPlanarYUV(c->dstFormat) || (isGray(c->dstFormat) && !isALPHA(c->dstFormat))) {
183 lumCtx = av_mallocz(sizeof(VScalerContext));
185 return AVERROR(ENOMEM);
188 desc[0].process = lum_planar_vscale;
189 desc[0].instance = lumCtx;
192 desc[0].alpha = c->alpPixBuf != 0;
194 if (!isGray(c->dstFormat)) {
195 chrCtx = av_mallocz(sizeof(VScalerContext));
197 return AVERROR(ENOMEM);
198 desc[1].process = chr_planar_vscale;
199 desc[1].instance = chrCtx;
204 lumCtx = av_mallocz_array(sizeof(VScalerContext), 2);
206 return AVERROR(ENOMEM);
209 desc[0].process = c->yuv2packedX ? packed_vscale : any_vscale;
210 desc[0].instance = lumCtx;
213 desc[0].alpha = c->alpPixBuf != 0;
216 ff_init_vscale_pfn(c, c->yuv2plane1, c->yuv2planeX, c->yuv2nv12cX,
217 c->yuv2packed1, c->yuv2packed2, c->yuv2packedX, c->yuv2anyX, c->use_mmx_vfilter);
221 void ff_init_vscale_pfn(SwsContext *c,
222 yuv2planar1_fn yuv2plane1,
223 yuv2planarX_fn yuv2planeX,
224 yuv2interleavedX_fn yuv2nv12cX,
225 yuv2packed1_fn yuv2packed1,
226 yuv2packed2_fn yuv2packed2,
227 yuv2packedX_fn yuv2packedX,
228 yuv2anyX_fn yuv2anyX, int use_mmx)
230 VScalerContext *lumCtx = NULL;
231 VScalerContext *chrCtx = NULL;
232 int idx = c->numDesc - (c->is_internal_gamma ? 2 : 1);
234 if (isPlanarYUV(c->dstFormat) || (isGray(c->dstFormat) && !isALPHA(c->dstFormat))) {
235 if (!isGray(c->dstFormat)) {
236 chrCtx = c->desc[idx].instance;
238 chrCtx->filter[0] = use_mmx ? (int16_t*)c->chrMmxFilter : c->vChrFilter;
239 chrCtx->filter_size = c->vChrFilterSize;
240 chrCtx->filter_pos = c->vChrFilterPos;
241 chrCtx->isMMX = use_mmx;
244 if (yuv2nv12cX) chrCtx->pfn = yuv2nv12cX;
245 else if (c->vChrFilterSize == 1) chrCtx->pfn = yuv2plane1;
246 else chrCtx->pfn = yuv2planeX;
249 lumCtx = c->desc[idx].instance;
251 lumCtx->filter[0] = use_mmx ? (int16_t*)c->lumMmxFilter : c->vLumFilter;
252 lumCtx->filter[1] = use_mmx ? (int16_t*)c->alpMmxFilter : c->vLumFilter;
253 lumCtx->filter_size = c->vLumFilterSize;
254 lumCtx->filter_pos = c->vLumFilterPos;
255 lumCtx->isMMX = use_mmx;
257 if (c->vLumFilterSize == 1) lumCtx->pfn = yuv2plane1;
258 else lumCtx->pfn = yuv2planeX;
261 lumCtx = c->desc[idx].instance;
264 lumCtx->filter[0] = c->vLumFilter;
265 lumCtx->filter_size = c->vLumFilterSize;
266 lumCtx->filter_pos = c->vLumFilterPos;
268 chrCtx->filter[0] = c->vChrFilter;
269 chrCtx->filter_size = c->vChrFilterSize;
270 chrCtx->filter_pos = c->vChrFilterPos;
272 lumCtx->isMMX = use_mmx;
273 chrCtx->isMMX = use_mmx;
276 if (c->yuv2packed1 && c->vLumFilterSize == 1 && c->vChrFilterSize <= 2)
277 lumCtx->pfn = yuv2packed1;
278 else if (c->yuv2packed2 && c->vLumFilterSize == 2 && c->vChrFilterSize == 2)
279 lumCtx->pfn = yuv2packed2;
281 lumCtx->pfn = yuv2packedX;
283 lumCtx->pfn = yuv2anyX;