| 1 | /* |
| 2 | * Copyright (c) 2013 Vittorio Giovara |
| 3 | * |
| 4 | * This file is part of FFmpeg. |
| 5 | * |
| 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. |
| 10 | * |
| 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. |
| 15 | * |
| 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 |
| 19 | */ |
| 20 | |
| 21 | /** |
| 22 | * @file |
| 23 | * Generate a frame packed video, by combining two views in a single surface. |
| 24 | */ |
| 25 | |
| 26 | #include <string.h> |
| 27 | |
| 28 | #include "libavutil/imgutils.h" |
| 29 | #include "libavutil/opt.h" |
| 30 | #include "libavutil/pixdesc.h" |
| 31 | #include "libavutil/rational.h" |
| 32 | #include "libavutil/stereo3d.h" |
| 33 | |
| 34 | #include "avfilter.h" |
| 35 | #include "formats.h" |
| 36 | #include "internal.h" |
| 37 | #include "video.h" |
| 38 | |
| 39 | #define LEFT 0 |
| 40 | #define RIGHT 1 |
| 41 | |
| 42 | typedef struct FramepackContext { |
| 43 | const AVClass *class; |
| 44 | |
| 45 | const AVPixFmtDescriptor *pix_desc; ///< agreed pixel format |
| 46 | |
| 47 | enum AVStereo3DType format; ///< frame pack type output |
| 48 | |
| 49 | AVFrame *input_views[2]; ///< input frames |
| 50 | |
| 51 | int64_t double_pts; ///< new pts for frameseq mode |
| 52 | } FramepackContext; |
| 53 | |
| 54 | static const enum AVPixelFormat formats_supported[] = { |
| 55 | AV_PIX_FMT_YUV420P, AV_PIX_FMT_YUV422P, AV_PIX_FMT_YUV444P, |
| 56 | AV_PIX_FMT_YUV410P, AV_PIX_FMT_YUVA420P, AV_PIX_FMT_YUVJ420P, |
| 57 | AV_PIX_FMT_YUVJ422P, AV_PIX_FMT_YUVJ444P, AV_PIX_FMT_YUVJ440P, |
| 58 | AV_PIX_FMT_NONE |
| 59 | }; |
| 60 | |
| 61 | static int query_formats(AVFilterContext *ctx) |
| 62 | { |
| 63 | // this will ensure that formats are the same on all pads |
| 64 | ff_set_common_formats(ctx, ff_make_format_list(formats_supported)); |
| 65 | return 0; |
| 66 | } |
| 67 | |
| 68 | static av_cold void framepack_uninit(AVFilterContext *ctx) |
| 69 | { |
| 70 | FramepackContext *s = ctx->priv; |
| 71 | |
| 72 | // clean any leftover frame |
| 73 | av_frame_free(&s->input_views[LEFT]); |
| 74 | av_frame_free(&s->input_views[RIGHT]); |
| 75 | } |
| 76 | |
| 77 | static int config_output(AVFilterLink *outlink) |
| 78 | { |
| 79 | AVFilterContext *ctx = outlink->src; |
| 80 | FramepackContext *s = outlink->src->priv; |
| 81 | |
| 82 | int width = ctx->inputs[LEFT]->w; |
| 83 | int height = ctx->inputs[LEFT]->h; |
| 84 | AVRational time_base = ctx->inputs[LEFT]->time_base; |
| 85 | |
| 86 | // check size and fps match on the other input |
| 87 | if (width != ctx->inputs[RIGHT]->w || |
| 88 | height != ctx->inputs[RIGHT]->h) { |
| 89 | av_log(ctx, AV_LOG_ERROR, |
| 90 | "Left and right sizes differ (%dx%d vs %dx%d).\n", |
| 91 | width, height, |
| 92 | ctx->inputs[RIGHT]->w, ctx->inputs[RIGHT]->h); |
| 93 | return AVERROR_INVALIDDATA; |
| 94 | } else if (av_cmp_q(time_base, ctx->inputs[RIGHT]->time_base) != 0) { |
| 95 | av_log(ctx, AV_LOG_ERROR, |
| 96 | "Left and right framerates differ (%d/%d vs %d/%d).\n", |
| 97 | time_base.num, time_base.den, |
| 98 | ctx->inputs[RIGHT]->time_base.num, |
| 99 | ctx->inputs[RIGHT]->time_base.den); |
| 100 | return AVERROR_INVALIDDATA; |
| 101 | } |
| 102 | |
| 103 | s->pix_desc = av_pix_fmt_desc_get(outlink->format); |
| 104 | if (!s->pix_desc) |
| 105 | return AVERROR_BUG; |
| 106 | |
| 107 | // modify output properties as needed |
| 108 | switch (s->format) { |
| 109 | case AV_STEREO3D_FRAMESEQUENCE: |
| 110 | time_base.den *= 2; |
| 111 | s->double_pts = AV_NOPTS_VALUE; |
| 112 | break; |
| 113 | case AV_STEREO3D_COLUMNS: |
| 114 | case AV_STEREO3D_SIDEBYSIDE: |
| 115 | width *= 2; |
| 116 | break; |
| 117 | case AV_STEREO3D_LINES: |
| 118 | case AV_STEREO3D_TOPBOTTOM: |
| 119 | height *= 2; |
| 120 | break; |
| 121 | default: |
| 122 | av_log(ctx, AV_LOG_ERROR, "Unknown packing mode."); |
| 123 | return AVERROR_INVALIDDATA; |
| 124 | } |
| 125 | |
| 126 | outlink->w = width; |
| 127 | outlink->h = height; |
| 128 | outlink->time_base = time_base; |
| 129 | |
| 130 | return 0; |
| 131 | } |
| 132 | |
| 133 | static void horizontal_frame_pack(FramepackContext *s, |
| 134 | AVFrame *dst, |
| 135 | int interleaved) |
| 136 | { |
| 137 | int plane, i; |
| 138 | int length = dst->width / 2; |
| 139 | int lines = dst->height; |
| 140 | |
| 141 | for (plane = 0; plane < s->pix_desc->nb_components; plane++) { |
| 142 | const uint8_t *leftp = s->input_views[LEFT]->data[plane]; |
| 143 | const uint8_t *rightp = s->input_views[RIGHT]->data[plane]; |
| 144 | uint8_t *dstp = dst->data[plane]; |
| 145 | |
| 146 | if (plane == 1 || plane == 2) { |
| 147 | length = -(-(dst->width / 2) >> s->pix_desc->log2_chroma_w); |
| 148 | lines = -(-(dst->height) >> s->pix_desc->log2_chroma_h); |
| 149 | } |
| 150 | |
| 151 | if (interleaved) { |
| 152 | for (i = 0; i < lines; i++) { |
| 153 | int j; |
| 154 | int k = 0; |
| 155 | |
| 156 | for (j = 0; j < length; j++) { |
| 157 | dstp[k++] = leftp[j]; |
| 158 | dstp[k++] = rightp[j]; |
| 159 | } |
| 160 | |
| 161 | dstp += dst->linesize[plane]; |
| 162 | leftp += s->input_views[LEFT]->linesize[plane]; |
| 163 | rightp += s->input_views[RIGHT]->linesize[plane]; |
| 164 | } |
| 165 | } else { |
| 166 | av_image_copy_plane(dst->data[plane], dst->linesize[plane], |
| 167 | leftp, s->input_views[LEFT]->linesize[plane], |
| 168 | length, lines); |
| 169 | av_image_copy_plane(dst->data[plane] + length, dst->linesize[plane], |
| 170 | rightp, s->input_views[RIGHT]->linesize[plane], |
| 171 | length, lines); |
| 172 | } |
| 173 | } |
| 174 | } |
| 175 | |
| 176 | static void vertical_frame_pack(FramepackContext *s, |
| 177 | AVFrame *dst, |
| 178 | int interleaved) |
| 179 | { |
| 180 | int plane, offset; |
| 181 | int length = dst->width; |
| 182 | int lines = dst->height / 2; |
| 183 | |
| 184 | for (plane = 0; plane < s->pix_desc->nb_components; plane++) { |
| 185 | if (plane == 1 || plane == 2) { |
| 186 | length = -(-(dst->width) >> s->pix_desc->log2_chroma_w); |
| 187 | lines = -(-(dst->height / 2) >> s->pix_desc->log2_chroma_h); |
| 188 | } |
| 189 | |
| 190 | offset = interleaved ? dst->linesize[plane] : dst->linesize[plane] * lines; |
| 191 | |
| 192 | av_image_copy_plane(dst->data[plane], |
| 193 | dst->linesize[plane] << interleaved, |
| 194 | s->input_views[LEFT]->data[plane], |
| 195 | s->input_views[LEFT]->linesize[plane], |
| 196 | length, lines); |
| 197 | av_image_copy_plane(dst->data[plane] + offset, |
| 198 | dst->linesize[plane] << interleaved, |
| 199 | s->input_views[RIGHT]->data[plane], |
| 200 | s->input_views[RIGHT]->linesize[plane], |
| 201 | length, lines); |
| 202 | } |
| 203 | } |
| 204 | |
| 205 | static av_always_inline void spatial_frame_pack(FramepackContext *s, AVFrame *dst) |
| 206 | { |
| 207 | switch (s->format) { |
| 208 | case AV_STEREO3D_SIDEBYSIDE: |
| 209 | horizontal_frame_pack(s, dst, 0); |
| 210 | break; |
| 211 | case AV_STEREO3D_COLUMNS: |
| 212 | horizontal_frame_pack(s, dst, 1); |
| 213 | break; |
| 214 | case AV_STEREO3D_TOPBOTTOM: |
| 215 | vertical_frame_pack(s, dst, 0); |
| 216 | break; |
| 217 | case AV_STEREO3D_LINES: |
| 218 | vertical_frame_pack(s, dst, 1); |
| 219 | break; |
| 220 | } |
| 221 | } |
| 222 | |
| 223 | static int filter_frame_left(AVFilterLink *inlink, AVFrame *frame) |
| 224 | { |
| 225 | FramepackContext *s = inlink->dst->priv; |
| 226 | s->input_views[LEFT] = frame; |
| 227 | return 0; |
| 228 | } |
| 229 | |
| 230 | static int filter_frame_right(AVFilterLink *inlink, AVFrame *frame) |
| 231 | { |
| 232 | FramepackContext *s = inlink->dst->priv; |
| 233 | s->input_views[RIGHT] = frame; |
| 234 | return 0; |
| 235 | } |
| 236 | |
| 237 | static int request_frame(AVFilterLink *outlink) |
| 238 | { |
| 239 | AVFilterContext *ctx = outlink->src; |
| 240 | FramepackContext *s = ctx->priv; |
| 241 | AVStereo3D *stereo; |
| 242 | int ret, i; |
| 243 | |
| 244 | /* get a frame on the either input, stop as soon as a video ends */ |
| 245 | for (i = 0; i < 2; i++) { |
| 246 | if (!s->input_views[i]) { |
| 247 | ret = ff_request_frame(ctx->inputs[i]); |
| 248 | if (ret < 0) |
| 249 | return ret; |
| 250 | } |
| 251 | } |
| 252 | |
| 253 | if (s->format == AV_STEREO3D_FRAMESEQUENCE) { |
| 254 | if (s->double_pts == AV_NOPTS_VALUE) |
| 255 | s->double_pts = s->input_views[LEFT]->pts; |
| 256 | |
| 257 | for (i = 0; i < 2; i++) { |
| 258 | // set correct timestamps |
| 259 | s->input_views[i]->pts = s->double_pts++; |
| 260 | |
| 261 | // set stereo3d side data |
| 262 | stereo = av_stereo3d_create_side_data(s->input_views[i]); |
| 263 | if (!stereo) |
| 264 | return AVERROR(ENOMEM); |
| 265 | stereo->type = s->format; |
| 266 | |
| 267 | // filter the frame and immediately relinquish its pointer |
| 268 | ret = ff_filter_frame(outlink, s->input_views[i]); |
| 269 | s->input_views[i] = NULL; |
| 270 | if (ret < 0) |
| 271 | return ret; |
| 272 | } |
| 273 | return ret; |
| 274 | } else { |
| 275 | AVFrame *dst = ff_get_video_buffer(outlink, outlink->w, outlink->h); |
| 276 | if (!dst) |
| 277 | return AVERROR(ENOMEM); |
| 278 | |
| 279 | spatial_frame_pack(s, dst); |
| 280 | |
| 281 | // get any property from the original frame |
| 282 | ret = av_frame_copy_props(dst, s->input_views[LEFT]); |
| 283 | if (ret < 0) { |
| 284 | av_frame_free(&dst); |
| 285 | return ret; |
| 286 | } |
| 287 | |
| 288 | for (i = 0; i < 2; i++) |
| 289 | av_frame_free(&s->input_views[i]); |
| 290 | |
| 291 | // set stereo3d side data |
| 292 | stereo = av_stereo3d_create_side_data(dst); |
| 293 | if (!stereo) { |
| 294 | av_frame_free(&dst); |
| 295 | return AVERROR(ENOMEM); |
| 296 | } |
| 297 | stereo->type = s->format; |
| 298 | |
| 299 | return ff_filter_frame(outlink, dst); |
| 300 | } |
| 301 | } |
| 302 | |
| 303 | #define OFFSET(x) offsetof(FramepackContext, x) |
| 304 | #define V AV_OPT_FLAG_VIDEO_PARAM |
| 305 | static const AVOption options[] = { |
| 306 | { "format", "Frame pack output format", OFFSET(format), AV_OPT_TYPE_INT, |
| 307 | { .i64 = AV_STEREO3D_SIDEBYSIDE }, 0, INT_MAX, .flags = V, .unit = "format" }, |
| 308 | { "sbs", "Views are packed next to each other", 0, AV_OPT_TYPE_CONST, |
| 309 | { .i64 = AV_STEREO3D_SIDEBYSIDE }, INT_MIN, INT_MAX, .flags = V, .unit = "format" }, |
| 310 | { "tab", "Views are packed on top of each other", 0, AV_OPT_TYPE_CONST, |
| 311 | { .i64 = AV_STEREO3D_TOPBOTTOM }, INT_MIN, INT_MAX, .flags = V, .unit = "format" }, |
| 312 | { "frameseq", "Views are one after the other", 0, AV_OPT_TYPE_CONST, |
| 313 | { .i64 = AV_STEREO3D_FRAMESEQUENCE }, INT_MIN, INT_MAX, .flags = V, .unit = "format" }, |
| 314 | { "lines", "Views are interleaved by lines", 0, AV_OPT_TYPE_CONST, |
| 315 | { .i64 = AV_STEREO3D_LINES }, INT_MIN, INT_MAX, .flags = V, .unit = "format" }, |
| 316 | { "columns", "Views are interleaved by columns", 0, AV_OPT_TYPE_CONST, |
| 317 | { .i64 = AV_STEREO3D_COLUMNS }, INT_MIN, INT_MAX, .flags = V, .unit = "format" }, |
| 318 | { NULL }, |
| 319 | }; |
| 320 | |
| 321 | static const AVClass framepack_class = { |
| 322 | .class_name = "framepack", |
| 323 | .item_name = av_default_item_name, |
| 324 | .option = options, |
| 325 | .version = LIBAVUTIL_VERSION_INT, |
| 326 | }; |
| 327 | |
| 328 | static const AVFilterPad framepack_inputs[] = { |
| 329 | { |
| 330 | .name = "left", |
| 331 | .type = AVMEDIA_TYPE_VIDEO, |
| 332 | .filter_frame = filter_frame_left, |
| 333 | .needs_fifo = 1, |
| 334 | }, |
| 335 | { |
| 336 | .name = "right", |
| 337 | .type = AVMEDIA_TYPE_VIDEO, |
| 338 | .filter_frame = filter_frame_right, |
| 339 | .needs_fifo = 1, |
| 340 | }, |
| 341 | { NULL } |
| 342 | }; |
| 343 | |
| 344 | static const AVFilterPad framepack_outputs[] = { |
| 345 | { |
| 346 | .name = "packed", |
| 347 | .type = AVMEDIA_TYPE_VIDEO, |
| 348 | .config_props = config_output, |
| 349 | .request_frame = request_frame, |
| 350 | }, |
| 351 | { NULL } |
| 352 | }; |
| 353 | |
| 354 | AVFilter ff_vf_framepack = { |
| 355 | .name = "framepack", |
| 356 | .description = NULL_IF_CONFIG_SMALL("Generate a frame packed stereoscopic video."), |
| 357 | .priv_size = sizeof(FramepackContext), |
| 358 | .priv_class = &framepack_class, |
| 359 | .query_formats = query_formats, |
| 360 | .inputs = framepack_inputs, |
| 361 | .outputs = framepack_outputs, |
| 362 | .uninit = framepack_uninit, |
| 363 | }; |