2 * Copyright (c) 2013 Vittorio Giovara
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
23 * Generate a frame packed video, by combining two views in a single surface.
28 #include "libavutil/common.h"
29 #include "libavutil/imgutils.h"
30 #include "libavutil/opt.h"
31 #include "libavutil/pixdesc.h"
32 #include "libavutil/rational.h"
33 #include "libavutil/stereo3d.h"
43 typedef struct FramepackContext {
46 const AVPixFmtDescriptor *pix_desc; ///< agreed pixel format
48 enum AVStereo3DType format; ///< frame pack type output
50 AVFrame *input_views[2]; ///< input frames
52 int64_t double_pts; ///< new pts for frameseq mode
55 static const enum AVPixelFormat formats_supported[] = {
56 AV_PIX_FMT_YUV420P, AV_PIX_FMT_YUV422P, AV_PIX_FMT_YUV444P,
57 AV_PIX_FMT_YUV410P, AV_PIX_FMT_YUVA420P, AV_PIX_FMT_YUVJ420P,
58 AV_PIX_FMT_YUVJ422P, AV_PIX_FMT_YUVJ444P, AV_PIX_FMT_YUVJ440P,
62 static int query_formats(AVFilterContext *ctx)
64 // this will ensure that formats are the same on all pads
65 AVFilterFormats *fmts_list = ff_make_format_list(formats_supported);
67 return AVERROR(ENOMEM);
68 return ff_set_common_formats(ctx, fmts_list);
71 static av_cold void framepack_uninit(AVFilterContext *ctx)
73 FramepackContext *s = ctx->priv;
75 // clean any leftover frame
76 av_frame_free(&s->input_views[LEFT]);
77 av_frame_free(&s->input_views[RIGHT]);
80 static int config_output(AVFilterLink *outlink)
82 AVFilterContext *ctx = outlink->src;
83 FramepackContext *s = outlink->src->priv;
85 int width = ctx->inputs[LEFT]->w;
86 int height = ctx->inputs[LEFT]->h;
87 AVRational time_base = ctx->inputs[LEFT]->time_base;
88 AVRational frame_rate = ctx->inputs[LEFT]->frame_rate;
90 // check size and fps match on the other input
91 if (width != ctx->inputs[RIGHT]->w ||
92 height != ctx->inputs[RIGHT]->h) {
93 av_log(ctx, AV_LOG_ERROR,
94 "Left and right sizes differ (%dx%d vs %dx%d).\n",
96 ctx->inputs[RIGHT]->w, ctx->inputs[RIGHT]->h);
97 return AVERROR_INVALIDDATA;
98 } else if (av_cmp_q(time_base, ctx->inputs[RIGHT]->time_base) != 0) {
99 av_log(ctx, AV_LOG_ERROR,
100 "Left and right time bases differ (%d/%d vs %d/%d).\n",
101 time_base.num, time_base.den,
102 ctx->inputs[RIGHT]->time_base.num,
103 ctx->inputs[RIGHT]->time_base.den);
104 return AVERROR_INVALIDDATA;
105 } else if (av_cmp_q(frame_rate, ctx->inputs[RIGHT]->frame_rate) != 0) {
106 av_log(ctx, AV_LOG_ERROR,
107 "Left and right framerates differ (%d/%d vs %d/%d).\n",
108 frame_rate.num, frame_rate.den,
109 ctx->inputs[RIGHT]->frame_rate.num,
110 ctx->inputs[RIGHT]->frame_rate.den);
111 return AVERROR_INVALIDDATA;
114 s->pix_desc = av_pix_fmt_desc_get(outlink->format);
118 // modify output properties as needed
120 case AV_STEREO3D_FRAMESEQUENCE:
124 s->double_pts = AV_NOPTS_VALUE;
126 case AV_STEREO3D_COLUMNS:
127 case AV_STEREO3D_SIDEBYSIDE:
130 case AV_STEREO3D_LINES:
131 case AV_STEREO3D_TOPBOTTOM:
135 av_log(ctx, AV_LOG_ERROR, "Unknown packing mode.");
136 return AVERROR_INVALIDDATA;
141 outlink->time_base = time_base;
142 outlink->frame_rate = frame_rate;
147 static void horizontal_frame_pack(AVFilterLink *outlink,
151 AVFilterContext *ctx = outlink->src;
152 FramepackContext *s = ctx->priv;
156 const uint8_t *leftp = s->input_views[LEFT]->data[0];
157 const uint8_t *rightp = s->input_views[RIGHT]->data[0];
158 uint8_t *dstp = out->data[0];
159 int length = out->width / 2;
160 int lines = out->height;
162 for (plane = 0; plane < s->pix_desc->nb_components; plane++) {
163 if (plane == 1 || plane == 2) {
164 length = AV_CEIL_RSHIFT(out->width / 2, s->pix_desc->log2_chroma_w);
165 lines = AV_CEIL_RSHIFT(out->height, s->pix_desc->log2_chroma_h);
167 for (i = 0; i < lines; i++) {
169 leftp = s->input_views[LEFT]->data[plane] +
170 s->input_views[LEFT]->linesize[plane] * i;
171 rightp = s->input_views[RIGHT]->data[plane] +
172 s->input_views[RIGHT]->linesize[plane] * i;
173 dstp = out->data[plane] + out->linesize[plane] * i;
174 for (j = 0; j < length; j++) {
175 // interpolate chroma as necessary
176 if ((s->pix_desc->log2_chroma_w ||
177 s->pix_desc->log2_chroma_h) &&
178 (plane == 1 || plane == 2)) {
179 *dstp++ = (*leftp + *rightp) / 2;
180 *dstp++ = (*leftp + *rightp) / 2;
191 for (i = 0; i < 2; i++) {
192 const uint8_t *src[4];
194 int sub_w = s->input_views[i]->width >> s->pix_desc->log2_chroma_w;
196 src[0] = s->input_views[i]->data[0];
197 src[1] = s->input_views[i]->data[1];
198 src[2] = s->input_views[i]->data[2];
200 dst[0] = out->data[0] + i * s->input_views[i]->width;
201 dst[1] = out->data[1] + i * sub_w;
202 dst[2] = out->data[2] + i * sub_w;
204 av_image_copy(dst, out->linesize, src, s->input_views[i]->linesize,
205 s->input_views[i]->format,
206 s->input_views[i]->width,
207 s->input_views[i]->height);
212 static void vertical_frame_pack(AVFilterLink *outlink,
216 AVFilterContext *ctx = outlink->src;
217 FramepackContext *s = ctx->priv;
220 for (i = 0; i < 2; i++) {
221 const uint8_t *src[4];
224 int sub_h = s->input_views[i]->height >> s->pix_desc->log2_chroma_h;
226 src[0] = s->input_views[i]->data[0];
227 src[1] = s->input_views[i]->data[1];
228 src[2] = s->input_views[i]->data[2];
230 dst[0] = out->data[0] + i * out->linesize[0] *
231 (interleaved + s->input_views[i]->height * (1 - interleaved));
232 dst[1] = out->data[1] + i * out->linesize[1] *
233 (interleaved + sub_h * (1 - interleaved));
234 dst[2] = out->data[2] + i * out->linesize[2] *
235 (interleaved + sub_h * (1 - interleaved));
237 linesizes[0] = out->linesize[0] +
238 interleaved * out->linesize[0];
239 linesizes[1] = out->linesize[1] +
240 interleaved * out->linesize[1];
241 linesizes[2] = out->linesize[2] +
242 interleaved * out->linesize[2];
244 av_image_copy(dst, linesizes, src, s->input_views[i]->linesize,
245 s->input_views[i]->format,
246 s->input_views[i]->width,
247 s->input_views[i]->height);
251 static av_always_inline void spatial_frame_pack(AVFilterLink *outlink,
254 AVFilterContext *ctx = outlink->src;
255 FramepackContext *s = ctx->priv;
257 case AV_STEREO3D_SIDEBYSIDE:
258 horizontal_frame_pack(outlink, dst, 0);
260 case AV_STEREO3D_COLUMNS:
261 horizontal_frame_pack(outlink, dst, 1);
263 case AV_STEREO3D_TOPBOTTOM:
264 vertical_frame_pack(outlink, dst, 0);
266 case AV_STEREO3D_LINES:
267 vertical_frame_pack(outlink, dst, 1);
272 static int try_push_frame(AVFilterContext *ctx);
274 static int filter_frame_left(AVFilterLink *inlink, AVFrame *frame)
276 FramepackContext *s = inlink->dst->priv;
277 s->input_views[LEFT] = frame;
278 return try_push_frame(inlink->dst);
281 static int filter_frame_right(AVFilterLink *inlink, AVFrame *frame)
283 FramepackContext *s = inlink->dst->priv;
284 s->input_views[RIGHT] = frame;
285 return try_push_frame(inlink->dst);
288 static int request_frame(AVFilterLink *outlink)
290 AVFilterContext *ctx = outlink->src;
291 FramepackContext *s = ctx->priv;
294 /* get a frame on the either input, stop as soon as a video ends */
295 for (i = 0; i < 2; i++) {
296 if (!s->input_views[i]) {
297 ret = ff_request_frame(ctx->inputs[i]);
305 static int try_push_frame(AVFilterContext *ctx)
307 FramepackContext *s = ctx->priv;
308 AVFilterLink *outlink = ctx->outputs[0];
312 if (!(s->input_views[0] && s->input_views[1]))
314 if (s->format == AV_STEREO3D_FRAMESEQUENCE) {
315 if (s->double_pts == AV_NOPTS_VALUE)
316 s->double_pts = s->input_views[LEFT]->pts;
318 for (i = 0; i < 2; i++) {
319 // set correct timestamps
320 s->input_views[i]->pts = s->double_pts++;
322 // set stereo3d side data
323 stereo = av_stereo3d_create_side_data(s->input_views[i]);
325 return AVERROR(ENOMEM);
326 stereo->type = s->format;
327 stereo->view = i == LEFT ? AV_STEREO3D_VIEW_LEFT
328 : AV_STEREO3D_VIEW_RIGHT;
330 // filter the frame and immediately relinquish its pointer
331 ret = ff_filter_frame(outlink, s->input_views[i]);
332 s->input_views[i] = NULL;
338 AVFrame *dst = ff_get_video_buffer(outlink, outlink->w, outlink->h);
340 return AVERROR(ENOMEM);
342 spatial_frame_pack(outlink, dst);
344 // get any property from the original frame
345 ret = av_frame_copy_props(dst, s->input_views[LEFT]);
351 for (i = 0; i < 2; i++)
352 av_frame_free(&s->input_views[i]);
354 // set stereo3d side data
355 stereo = av_stereo3d_create_side_data(dst);
358 return AVERROR(ENOMEM);
360 stereo->type = s->format;
362 return ff_filter_frame(outlink, dst);
366 #define OFFSET(x) offsetof(FramepackContext, x)
367 #define V AV_OPT_FLAG_VIDEO_PARAM
368 static const AVOption framepack_options[] = {
369 { "format", "Frame pack output format", OFFSET(format), AV_OPT_TYPE_INT,
370 { .i64 = AV_STEREO3D_SIDEBYSIDE }, 0, INT_MAX, .flags = V, .unit = "format" },
371 { "sbs", "Views are packed next to each other", 0, AV_OPT_TYPE_CONST,
372 { .i64 = AV_STEREO3D_SIDEBYSIDE }, INT_MIN, INT_MAX, .flags = V, .unit = "format" },
373 { "tab", "Views are packed on top of each other", 0, AV_OPT_TYPE_CONST,
374 { .i64 = AV_STEREO3D_TOPBOTTOM }, INT_MIN, INT_MAX, .flags = V, .unit = "format" },
375 { "frameseq", "Views are one after the other", 0, AV_OPT_TYPE_CONST,
376 { .i64 = AV_STEREO3D_FRAMESEQUENCE }, INT_MIN, INT_MAX, .flags = V, .unit = "format" },
377 { "lines", "Views are interleaved by lines", 0, AV_OPT_TYPE_CONST,
378 { .i64 = AV_STEREO3D_LINES }, INT_MIN, INT_MAX, .flags = V, .unit = "format" },
379 { "columns", "Views are interleaved by columns", 0, AV_OPT_TYPE_CONST,
380 { .i64 = AV_STEREO3D_COLUMNS }, INT_MIN, INT_MAX, .flags = V, .unit = "format" },
384 AVFILTER_DEFINE_CLASS(framepack);
386 static const AVFilterPad framepack_inputs[] = {
389 .type = AVMEDIA_TYPE_VIDEO,
390 .filter_frame = filter_frame_left,
395 .type = AVMEDIA_TYPE_VIDEO,
396 .filter_frame = filter_frame_right,
402 static const AVFilterPad framepack_outputs[] = {
405 .type = AVMEDIA_TYPE_VIDEO,
406 .config_props = config_output,
407 .request_frame = request_frame,
412 AVFilter ff_vf_framepack = {
414 .description = NULL_IF_CONFIG_SMALL("Generate a frame packed stereoscopic video."),
415 .priv_size = sizeof(FramepackContext),
416 .priv_class = &framepack_class,
417 .query_formats = query_formats,
418 .inputs = framepack_inputs,
419 .outputs = framepack_outputs,
420 .uninit = framepack_uninit,