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
21 #include "swscale_internal.h"
23 static void free_lines(SwsSlice *s)
26 for (i = 0; i < 2; ++i) {
27 int n = s->plane[i].available_lines;
29 for (j = 0; j < n; ++j) {
30 av_freep(&s->plane[i].line[j]);
32 s->plane[i].line[j+n] = NULL;
36 for (i = 0; i < 4; ++i)
37 memset(s->plane[i].line, 0, sizeof(uint8_t*) * s->plane[i].available_lines * (s->is_ring ? 3 : 1));
38 s->should_free_lines = 0;
42 slice lines contains extra bytes for vetorial code thus @size
43 is the allocated memory size and @width is the number of pixels
45 static int alloc_lines(SwsSlice *s, int size, int width)
50 s->should_free_lines = 1;
53 for (i = 0; i < 2; ++i) {
54 int n = s->plane[i].available_lines;
58 av_assert0(n == s->plane[ii].available_lines);
59 for (j = 0; j < n; ++j) {
60 // chroma plane line U and V are expected to be contiguous in memory
61 // by mmx vertical scaler code
62 s->plane[i].line[j] = av_malloc(size * 2 + 32);
63 if (!s->plane[i].line[j]) {
65 return AVERROR(ENOMEM);
67 s->plane[ii].line[j] = s->plane[i].line[j] + size + 16;
69 s->plane[i].line[j+n] = s->plane[i].line[j];
70 s->plane[ii].line[j+n] = s->plane[ii].line[j];
78 static int alloc_slice(SwsSlice *s, enum AVPixelFormat fmt, int lumLines, int chrLines, int h_sub_sample, int v_sub_sample, int ring)
81 int size[4] = { lumLines,
86 s->h_chr_sub_sample = h_sub_sample;
87 s->v_chr_sub_sample = v_sub_sample;
90 s->should_free_lines = 0;
92 for (i = 0; i < 4; ++i) {
93 int n = size[i] * ( ring == 0 ? 1 : 3);
94 s->plane[i].line = av_mallocz_array(sizeof(uint8_t*), n);
95 if (!s->plane[i].line)
96 return AVERROR(ENOMEM);
98 s->plane[i].tmp = ring ? s->plane[i].line + size[i] * 2 : NULL;
99 s->plane[i].available_lines = size[i];
100 s->plane[i].sliceY = 0;
101 s->plane[i].sliceH = 0;
106 static void free_slice(SwsSlice *s)
110 if (s->should_free_lines)
112 for (i = 0; i < 4; ++i) {
113 av_freep(&s->plane[i].line);
114 s->plane[i].tmp = NULL;
119 int ff_rotate_slice(SwsSlice *s, int lum, int chr)
123 for (i = 0; i < 4; i+=3) {
124 int n = s->plane[i].available_lines;
125 int l = lum - s->plane[i].sliceY;
128 s->plane[i].sliceY += n;
129 s->plane[i].sliceH -= n;
134 for (i = 1; i < 3; ++i) {
135 int n = s->plane[i].available_lines;
136 int l = chr - s->plane[i].sliceY;
139 s->plane[i].sliceY += n;
140 s->plane[i].sliceH -= n;
147 int ff_init_slice_from_src(SwsSlice * s, uint8_t *src[4], int stride[4], int srcW, int lumY, int lumH, int chrY, int chrH)
151 const int start[4] = {lumY,
156 const int end[4] = {lumY +lumH,
163 for (i = 0; i < 4; ++i) {
166 lines = s->plane[i].available_lines < lines ? s->plane[i].available_lines : lines;
168 if (end[i] > s->plane[i].sliceY+s->plane[i].sliceH) {
169 if (start[i] <= s->plane[i].sliceY+1)
170 s->plane[i].sliceY = FFMIN(start[i], s->plane[i].sliceY);
172 s->plane[i].sliceY = start[i];
173 s->plane[i].sliceH = end[i] - s->plane[i].sliceY;
175 if (end[i] >= s->plane[i].sliceY)
176 s->plane[i].sliceH = s->plane[i].sliceY + s->plane[i].sliceH - start[i];
178 s->plane[i].sliceH = end[i] - start[i];
179 s->plane[i].sliceY = start[i];
182 for (j = start[i]; j < lines; j+= 1)
183 s->plane[i].line[j] = src[i] + (start[i] + j) * stride[i];
190 static void fill_ones(SwsSlice *s, int n, int is16bit)
193 for (i = 0; i < 4; ++i) {
195 int size = s->plane[i].available_lines;
196 for (j = 0; j < size; ++j) {
198 int end = is16bit ? n>>1: n;
199 // fill also one extra element
202 for (k = 0; k < end; ++k)
203 ((int32_t*)(s->plane[i].line[j]))[k] = 1<<18;
205 for (k = 0; k < end; ++k)
206 ((int16_t*)(s->plane[i].line[j]))[k] = 1<<14;
211 int ff_init_filters(SwsContext * c)
217 int num_vdesc = isPlanarYUV(c->dstFormat) && !isGray(c->dstFormat) ? 2 : 1;
218 int need_lum_conv = c->lumToYV12 || c->readLumPlanar || c->alpToYV12 || c->readAlpPlanar;
219 int need_chr_conv = c->chrToYV12 || c->readChrPlanar;
221 int dst_stride = FFALIGN(c->dstW * sizeof(int16_t) + 66, 16);
223 uint32_t * pal = usePal(c->srcFormat) ? c->pal_yuv : (uint32_t*)c->input_rgb2yuv_table;
229 num_ydesc = need_lum_conv ? 2 : 1;
230 num_cdesc = need_chr_conv ? 2 : 1;
232 c->numSlice = FFMAX(num_ydesc, num_cdesc) + 2;
233 c->numDesc = num_ydesc + num_cdesc + num_vdesc;
234 c->descIndex[0] = num_ydesc;
235 c->descIndex[1] = num_ydesc + num_cdesc;
239 c->desc = av_mallocz_array(sizeof(SwsFilterDescriptor), c->numDesc);
241 return AVERROR(ENOMEM);
242 c->slice = av_mallocz_array(sizeof(SwsSlice), c->numSlice);
245 res = alloc_slice(&c->slice[0], c->srcFormat, c->srcH, c->chrSrcH, c->chrSrcHSubSample, c->chrSrcVSubSample, 0);
246 if (res < 0) goto cleanup;
247 for (i = 1; i < c->numSlice-2; ++i) {
248 res = alloc_slice(&c->slice[i], c->srcFormat, c->vLumFilterSize + MAX_LINES_AHEAD, c->vChrFilterSize + MAX_LINES_AHEAD, c->chrSrcHSubSample, c->chrSrcVSubSample, 0);
249 if (res < 0) goto cleanup;
250 res = alloc_lines(&c->slice[i], FFALIGN(c->srcW*2+78, 16), c->srcW);
251 if (res < 0) goto cleanup;
253 // horizontal scaler output
254 res = alloc_slice(&c->slice[i], c->srcFormat, c->vLumFilterSize + MAX_LINES_AHEAD, c->vChrFilterSize + MAX_LINES_AHEAD, c->chrDstHSubSample, c->chrDstVSubSample, 1);
255 if (res < 0) goto cleanup;
256 res = alloc_lines(&c->slice[i], dst_stride, c->dstW);
257 if (res < 0) goto cleanup;
259 fill_ones(&c->slice[i], dst_stride>>1, c->dstBpc == 16);
261 // vertical scaler output
263 res = alloc_slice(&c->slice[i], c->dstFormat, c->dstH, c->chrDstH, c->chrDstHSubSample, c->chrDstVSubSample, 0);
264 if (res < 0) goto cleanup;
271 ff_init_desc_fmt_convert(&c->desc[index], &c->slice[srcIdx], &c->slice[dstIdx], pal);
272 c->desc[index].alpha = c->alpPixBuf != 0;
278 dstIdx = FFMAX(num_ydesc, num_cdesc);
279 ff_init_desc_hscale(&c->desc[index], &c->slice[index], &c->slice[dstIdx], c->hLumFilter, c->hLumFilterPos, c->hLumFilterSize, c->lumXInc);
280 c->desc[index].alpha = c->alpPixBuf != 0;
288 ff_init_desc_cfmt_convert(&c->desc[index], &c->slice[srcIdx], &c->slice[dstIdx], pal);
293 dstIdx = FFMAX(num_ydesc, num_cdesc);
294 if (c->needs_hcscale)
295 ff_init_desc_chscale(&c->desc[index], &c->slice[srcIdx], &c->slice[dstIdx], c->hChrFilter, c->hChrFilterPos, c->hChrFilterSize, c->chrXInc);
297 ff_init_desc_no_chr(&c->desc[index], &c->slice[srcIdx], &c->slice[dstIdx]);
302 srcIdx = c->numSlice - 2;
303 dstIdx = c->numSlice - 1;
304 ff_init_vscale(c, c->desc + index, c->slice + srcIdx, c->slice + dstIdx);
314 int ff_free_filters(SwsContext *c)
318 for (i = 0; i < c->numDesc; ++i)
319 av_freep(&c->desc[i].instance);
324 for (i = 0; i < c->numSlice; ++i)
325 free_slice(&c->slice[i]);