slice.c 11.9 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
/*
 * Copyright (C) 2015 Pedro Arthur <bygrandao@gmail.com>
 *
 * This file is part of FFmpeg.
 *
 * FFmpeg is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * FFmpeg is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
 * License along with FFmpeg; if not, write to the Free Software
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
 */

21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41
#include "swscale_internal.h"

static void free_lines(SwsSlice *s)
{
    int i;
    for (i = 0; i < 2; ++i) {
        int n = s->plane[i].available_lines;
        int j;
        for (j = 0; j < n; ++j) {
            av_freep(&s->plane[i].line[j]);
            if (s->is_ring)
               s->plane[i].line[j+n] = NULL;
        }
    }

    for (i = 0; i < 4; ++i)
        memset(s->plane[i].line, 0, sizeof(uint8_t*) * s->plane[i].available_lines * (s->is_ring ? 3 : 1));
    s->should_free_lines = 0;
}

/*
42
 slice lines contains extra bytes for vectorial code thus @size
43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146
 is the allocated memory size and @width is the number of pixels
*/
static int alloc_lines(SwsSlice *s, int size, int width)
{
    int i;
    int idx[2] = {3, 2};

    s->should_free_lines = 1;
    s->width = width;

    for (i = 0; i < 2; ++i) {
        int n = s->plane[i].available_lines;
        int j;
        int ii = idx[i];

        av_assert0(n == s->plane[ii].available_lines);
        for (j = 0; j < n; ++j) {
            // chroma plane line U and V are expected to be contiguous in memory
            // by mmx vertical scaler code
            s->plane[i].line[j] = av_malloc(size * 2 + 32);
            if (!s->plane[i].line[j]) {
                free_lines(s);
                return AVERROR(ENOMEM);
            }
            s->plane[ii].line[j] = s->plane[i].line[j] + size + 16;
            if (s->is_ring) {
               s->plane[i].line[j+n] = s->plane[i].line[j];
               s->plane[ii].line[j+n] = s->plane[ii].line[j];
            }
        }
    }

    return 0;
}

static int alloc_slice(SwsSlice *s, enum AVPixelFormat fmt, int lumLines, int chrLines, int h_sub_sample, int v_sub_sample, int ring)
{
    int i;
    int size[4] = { lumLines,
                    chrLines,
                    chrLines,
                    lumLines };

    s->h_chr_sub_sample = h_sub_sample;
    s->v_chr_sub_sample = v_sub_sample;
    s->fmt = fmt;
    s->is_ring = ring;
    s->should_free_lines = 0;

    for (i = 0; i < 4; ++i) {
        int n = size[i] * ( ring == 0 ? 1 : 3);
        s->plane[i].line = av_mallocz_array(sizeof(uint8_t*), n);
        if (!s->plane[i].line)
            return AVERROR(ENOMEM);

        s->plane[i].tmp = ring ? s->plane[i].line + size[i] * 2 : NULL;
        s->plane[i].available_lines = size[i];
        s->plane[i].sliceY = 0;
        s->plane[i].sliceH = 0;
    }
    return 0;
}

static void free_slice(SwsSlice *s)
{
    int i;
    if (s) {
        if (s->should_free_lines)
            free_lines(s);
        for (i = 0; i < 4; ++i) {
            av_freep(&s->plane[i].line);
            s->plane[i].tmp = NULL;
        }
    }
}

int ff_rotate_slice(SwsSlice *s, int lum, int chr)
{
    int i;
    if (lum) {
        for (i = 0; i < 4; i+=3) {
            int n = s->plane[i].available_lines;
            int l = lum - s->plane[i].sliceY;

            if (l >= n * 2) {
                s->plane[i].sliceY += n;
                s->plane[i].sliceH -= n;
            }
        }
    }
    if (chr) {
        for (i = 1; i < 3; ++i) {
            int n = s->plane[i].available_lines;
            int l = chr - s->plane[i].sliceY;

            if (l >= n * 2) {
                s->plane[i].sliceY += n;
                s->plane[i].sliceH -= n;
            }
        }
    }
    return 0;
}

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, int relative)
148 149 150 151
{
    int i = 0;

    const int start[4] = {lumY,
152 153 154
                          chrY,
                          chrY,
                          lumY};
155 156 157 158 159 160

    const int end[4] = {lumY +lumH,
                        chrY + chrH,
                        chrY + chrH,
                        lumY + lumH};

161
    const uint8_t *src_[4] = {src[0] + (relative ? 0 : start[0]) * stride[0],
162 163 164
                              src[1] + (relative ? 0 : start[1]) * stride[1],
                              src[2] + (relative ? 0 : start[2]) * stride[2],
                              src[3] + (relative ? 0 : start[3]) * stride[3]};
165

166 167 168 169
    s->width = srcW;

    for (i = 0; i < 4; ++i) {
        int j;
170 171 172 173
        int first = s->plane[i].sliceY;
        int n = s->plane[i].available_lines;
        int lines = end[i] - start[i];
        int tot_lines = end[i] - first;
174

175 176 177 178
        if (start[i] >= first && n >= tot_lines) {
            s->plane[i].sliceH = FFMAX(tot_lines, s->plane[i].sliceH);
            for (j = 0; j < lines; j+= 1)
                s->plane[i].line[start[i] - first + j] = src_[i] +  j * stride[i];
179 180
        } else {
            s->plane[i].sliceY = start[i];
181 182 183 184
            lines = lines > n ? n : lines;
            s->plane[i].sliceH = lines;
            for (j = 0; j < lines; j+= 1)
                s->plane[i].line[j] = src_[i] +  j * stride[i];
185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212
        }

    }

    return 0;
}

static void fill_ones(SwsSlice *s, int n, int is16bit)
{
    int i;
    for (i = 0; i < 4; ++i) {
        int j;
        int size = s->plane[i].available_lines;
        for (j = 0; j < size; ++j) {
            int k;
            int end = is16bit ? n>>1: n;
            // fill also one extra element
            end += 1;
            if (is16bit)
                for (k = 0; k < end; ++k)
                    ((int32_t*)(s->plane[i].line[j]))[k] = 1<<18;
            else
                for (k = 0; k < end; ++k)
                    ((int16_t*)(s->plane[i].line[j]))[k] = 1<<14;
        }
    }
}

213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248
/*
 Calculates the minimum ring buffer size, it should be able to store vFilterSize
 more n lines where n is the max difference between each adjacent slice which
 outputs a line.
 The n lines are needed only when there is not enough src lines to output a single
 dst line, then we should buffer these lines to process them on the next call to scale.
*/
static void get_min_buffer_size(SwsContext *c, int *out_lum_size, int *out_chr_size)
{
    int lumY;
    int dstH = c->dstH;
    int chrDstH = c->chrDstH;
    int *lumFilterPos = c->vLumFilterPos;
    int *chrFilterPos = c->vChrFilterPos;
    int lumFilterSize = c->vLumFilterSize;
    int chrFilterSize = c->vChrFilterSize;
    int chrSubSample = c->chrSrcVSubSample;

    *out_lum_size = lumFilterSize;
    *out_chr_size = chrFilterSize;

    for (lumY = 0; lumY < dstH; lumY++) {
        int chrY      = (int64_t)lumY * chrDstH / dstH;
        int nextSlice = FFMAX(lumFilterPos[lumY] + lumFilterSize - 1,
                              ((chrFilterPos[chrY] + chrFilterSize - 1)
                               << chrSubSample));

        nextSlice >>= chrSubSample;
        nextSlice <<= chrSubSample;
        (*out_lum_size) = FFMAX((*out_lum_size), nextSlice - lumFilterPos[lumY]);
        (*out_chr_size) = FFMAX((*out_chr_size), (nextSlice >> chrSubSample) - chrFilterPos[chrY]);
    }
}



249 250 251 252 253 254
int ff_init_filters(SwsContext * c)
{
    int i;
    int index;
    int num_ydesc;
    int num_cdesc;
255
    int num_vdesc = isPlanarYUV(c->dstFormat) && !isGray(c->dstFormat) ? 2 : 1;
256 257
    int need_lum_conv = c->lumToYV12 || c->readLumPlanar || c->alpToYV12 || c->readAlpPlanar;
    int need_chr_conv = c->chrToYV12 || c->readChrPlanar;
258
    int need_gamma = c->is_internal_gamma;
259 260 261 262 263 264
    int srcIdx, dstIdx;
    int dst_stride = FFALIGN(c->dstW * sizeof(int16_t) + 66, 16);

    uint32_t * pal = usePal(c->srcFormat) ? c->pal_yuv : (uint32_t*)c->input_rgb2yuv_table;
    int res = 0;

265 266 267 268 269 270 271
    int lumBufSize;
    int chrBufSize;

    get_min_buffer_size(c, &lumBufSize, &chrBufSize);
    lumBufSize = FFMAX(lumBufSize, c->vLumFilterSize + MAX_LINES_AHEAD);
    chrBufSize = FFMAX(chrBufSize, c->vChrFilterSize + MAX_LINES_AHEAD);

272 273 274 275 276 277
    if (c->dstBpc == 16)
        dst_stride <<= 1;

    num_ydesc = need_lum_conv ? 2 : 1;
    num_cdesc = need_chr_conv ? 2 : 1;

278
    c->numSlice = FFMAX(num_ydesc, num_cdesc) + 2;
279 280 281
    c->numDesc = num_ydesc + num_cdesc + num_vdesc + (need_gamma ? 2 : 0);
    c->descIndex[0] = num_ydesc + (need_gamma ? 1 : 0);
    c->descIndex[1] = num_ydesc + num_cdesc + (need_gamma ? 1 : 0);
282 283 284 285 286 287 288 289 290 291 292



    c->desc = av_mallocz_array(sizeof(SwsFilterDescriptor), c->numDesc);
    if (!c->desc)
        return AVERROR(ENOMEM);
    c->slice = av_mallocz_array(sizeof(SwsSlice), c->numSlice);


    res = alloc_slice(&c->slice[0], c->srcFormat, c->srcH, c->chrSrcH, c->chrSrcHSubSample, c->chrSrcVSubSample, 0);
    if (res < 0) goto cleanup;
293
    for (i = 1; i < c->numSlice-2; ++i) {
294
        res = alloc_slice(&c->slice[i], c->srcFormat, lumBufSize, chrBufSize, c->chrSrcHSubSample, c->chrSrcVSubSample, 0);
295 296 297 298
        if (res < 0) goto cleanup;
        res = alloc_lines(&c->slice[i], FFALIGN(c->srcW*2+78, 16), c->srcW);
        if (res < 0) goto cleanup;
    }
299
    // horizontal scaler output
300
    res = alloc_slice(&c->slice[i], c->srcFormat, lumBufSize, chrBufSize, c->chrDstHSubSample, c->chrDstVSubSample, 1);
301 302 303 304 305 306
    if (res < 0) goto cleanup;
    res = alloc_lines(&c->slice[i], dst_stride, c->dstW);
    if (res < 0) goto cleanup;

    fill_ones(&c->slice[i], dst_stride>>1, c->dstBpc == 16);

307 308 309 310 311
    // vertical scaler output
    ++i;
    res = alloc_slice(&c->slice[i], c->dstFormat, c->dstH, c->chrDstH, c->chrDstHSubSample, c->chrDstVSubSample, 0);
    if (res < 0) goto cleanup;

312 313 314 315
    index = 0;
    srcIdx = 0;
    dstIdx = 1;

316 317 318 319 320 321
    if (need_gamma) {
        res = ff_init_gamma_convert(c->desc + index, c->slice + srcIdx, c->inv_gamma);
        if (res < 0) goto cleanup;
        ++index;
    }

322
    if (need_lum_conv) {
323 324
        res = ff_init_desc_fmt_convert(&c->desc[index], &c->slice[srcIdx], &c->slice[dstIdx], pal);
        if (res < 0) goto cleanup;
325
        c->desc[index].alpha = c->needAlpha;
326 327 328 329 330 331
        ++index;
        srcIdx = dstIdx;
    }


    dstIdx = FFMAX(num_ydesc, num_cdesc);
332 333
    res = ff_init_desc_hscale(&c->desc[index], &c->slice[srcIdx], &c->slice[dstIdx], c->hLumFilter, c->hLumFilterPos, c->hLumFilterSize, c->lumXInc);
    if (res < 0) goto cleanup;
334
    c->desc[index].alpha = c->needAlpha;
335 336 337 338 339 340 341


    ++index;
    {
        srcIdx = 0;
        dstIdx = 1;
        if (need_chr_conv) {
342 343
            res = ff_init_desc_cfmt_convert(&c->desc[index], &c->slice[srcIdx], &c->slice[dstIdx], pal);
            if (res < 0) goto cleanup;
344 345 346 347 348 349
            ++index;
            srcIdx = dstIdx;
        }

        dstIdx = FFMAX(num_ydesc, num_cdesc);
        if (c->needs_hcscale)
350
            res = ff_init_desc_chscale(&c->desc[index], &c->slice[srcIdx], &c->slice[dstIdx], c->hChrFilter, c->hChrFilterPos, c->hChrFilterSize, c->chrXInc);
351
        else
352 353
            res = ff_init_desc_no_chr(&c->desc[index], &c->slice[srcIdx], &c->slice[dstIdx]);
        if (res < 0) goto cleanup;
354 355
    }

356 357 358 359
    ++index;
    {
        srcIdx = c->numSlice - 2;
        dstIdx = c->numSlice - 1;
360 361
        res = ff_init_vscale(c, c->desc + index, c->slice + srcIdx, c->slice + dstIdx);
        if (res < 0) goto cleanup;
362 363
    }

364 365 366 367 368 369
    ++index;
    if (need_gamma) {
        res = ff_init_gamma_convert(c->desc + index, c->slice + dstIdx, c->gamma);
        if (res < 0) goto cleanup;
    }

370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392
    return 0;

cleanup:
    ff_free_filters(c);
    return res;
}

int ff_free_filters(SwsContext *c)
{
    int i;
    if (c->desc) {
        for (i = 0; i < c->numDesc; ++i)
            av_freep(&c->desc[i].instance);
        av_freep(&c->desc);
    }

    if (c->slice) {
        for (i = 0; i < c->numSlice; ++i)
            free_slice(&c->slice[i]);
        av_freep(&c->slice);
    }
    return 0;
}