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/imgutils.h"
29 #include "libavutil/opt.h"
30 #include "libavutil/pixdesc.h"
31 #include "libavutil/rational.h"
32 #include "libavutil/stereo3d.h"
42 typedef struct FramepackContext {
45 const AVPixFmtDescriptor *pix_desc; ///< agreed pixel format
47 enum AVStereo3DType format; ///< frame pack type output
49 AVFrame *input_views[2]; ///< input frames
51 int64_t double_pts; ///< new pts for frameseq mode
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,
61 static int query_formats(AVFilterContext *ctx)
63 // this will ensure that formats are the same on all pads
64 AVFilterFormats *fmts_list = ff_make_format_list(formats_supported);
66 return AVERROR(ENOMEM);
67 return ff_set_common_formats(ctx, fmts_list);
70 static av_cold void framepack_uninit(AVFilterContext *ctx)
72 FramepackContext *s = ctx->priv;
74 // clean any leftover frame
75 av_frame_free(&s->input_views[LEFT]);
76 av_frame_free(&s->input_views[RIGHT]);
79 static int config_output(AVFilterLink *outlink)
81 AVFilterContext *ctx = outlink->src;
82 FramepackContext *s = outlink->src->priv;
84 int width = ctx->inputs[LEFT]->w;
85 int height = ctx->inputs[LEFT]->h;
86 AVRational time_base = ctx->inputs[LEFT]->time_base;
87 AVRational frame_rate = ctx->inputs[LEFT]->frame_rate;
89 // check size and fps match on the other input
90 if (width != ctx->inputs[RIGHT]->w ||
91 height != ctx->inputs[RIGHT]->h) {
92 av_log(ctx, AV_LOG_ERROR,
93 "Left and right sizes differ (%dx%d vs %dx%d).\n",
95 ctx->inputs[RIGHT]->w, ctx->inputs[RIGHT]->h);
96 return AVERROR_INVALIDDATA;
97 } else if (av_cmp_q(time_base, ctx->inputs[RIGHT]->time_base) != 0) {
98 av_log(ctx, AV_LOG_ERROR,
99 "Left and right time bases differ (%d/%d vs %d/%d).\n",
100 time_base.num, time_base.den,
101 ctx->inputs[RIGHT]->time_base.num,
102 ctx->inputs[RIGHT]->time_base.den);
103 return AVERROR_INVALIDDATA;
104 } else if (av_cmp_q(frame_rate, ctx->inputs[RIGHT]->frame_rate) != 0) {
105 av_log(ctx, AV_LOG_ERROR,
106 "Left and right framerates differ (%d/%d vs %d/%d).\n",
107 frame_rate.num, frame_rate.den,
108 ctx->inputs[RIGHT]->frame_rate.num,
109 ctx->inputs[RIGHT]->frame_rate.den);
110 return AVERROR_INVALIDDATA;
113 s->pix_desc = av_pix_fmt_desc_get(outlink->format);
117 // modify output properties as needed
119 case AV_STEREO3D_FRAMESEQUENCE:
123 s->double_pts = AV_NOPTS_VALUE;
125 case AV_STEREO3D_COLUMNS:
126 case AV_STEREO3D_SIDEBYSIDE:
129 case AV_STEREO3D_LINES:
130 case AV_STEREO3D_TOPBOTTOM:
134 av_log(ctx, AV_LOG_ERROR, "Unknown packing mode.");
135 return AVERROR_INVALIDDATA;
140 outlink->time_base = time_base;
141 outlink->frame_rate = frame_rate;
146 static void horizontal_frame_pack(AVFilterLink *outlink,
150 AVFilterContext *ctx = outlink->src;
151 FramepackContext *s = ctx->priv;
155 const uint8_t *leftp = s->input_views[LEFT]->data[0];
156 const uint8_t *rightp = s->input_views[RIGHT]->data[0];
157 uint8_t *dstp = out->data[0];
158 int length = out->width / 2;
159 int lines = out->height;
161 for (plane = 0; plane < s->pix_desc->nb_components; plane++) {
162 if (plane == 1 || plane == 2) {
163 length = FF_CEIL_RSHIFT(out->width / 2, s->pix_desc->log2_chroma_w);
164 lines = FF_CEIL_RSHIFT(out->height, s->pix_desc->log2_chroma_h);
166 for (i = 0; i < lines; i++) {
168 leftp = s->input_views[LEFT]->data[plane] +
169 s->input_views[LEFT]->linesize[plane] * i;
170 rightp = s->input_views[RIGHT]->data[plane] +
171 s->input_views[RIGHT]->linesize[plane] * i;
172 dstp = out->data[plane] + out->linesize[plane] * i;
173 for (j = 0; j < length; j++) {
174 // interpolate chroma as necessary
175 if ((s->pix_desc->log2_chroma_w ||
176 s->pix_desc->log2_chroma_h) &&
177 (plane == 1 || plane == 2)) {
178 *dstp++ = (*leftp + *rightp) / 2;
179 *dstp++ = (*leftp + *rightp) / 2;
190 for (i = 0; i < 2; i++) {
191 const uint8_t *src[4];
193 int sub_w = s->input_views[i]->width >> s->pix_desc->log2_chroma_w;
195 src[0] = s->input_views[i]->data[0];
196 src[1] = s->input_views[i]->data[1];
197 src[2] = s->input_views[i]->data[2];
199 dst[0] = out->data[0] + i * s->input_views[i]->width;
200 dst[1] = out->data[1] + i * sub_w;
201 dst[2] = out->data[2] + i * sub_w;
203 av_image_copy(dst, out->linesize, src, s->input_views[i]->linesize,
204 s->input_views[i]->format,
205 s->input_views[i]->width,
206 s->input_views[i]->height);
211 static void vertical_frame_pack(AVFilterLink *outlink,
215 AVFilterContext *ctx = outlink->src;
216 FramepackContext *s = ctx->priv;
219 for (i = 0; i < 2; i++) {
220 const uint8_t *src[4];
223 int sub_h = s->input_views[i]->height >> s->pix_desc->log2_chroma_h;
225 src[0] = s->input_views[i]->data[0];
226 src[1] = s->input_views[i]->data[1];
227 src[2] = s->input_views[i]->data[2];
229 dst[0] = out->data[0] + i * out->linesize[0] *
230 (interleaved + s->input_views[i]->height * (1 - interleaved));
231 dst[1] = out->data[1] + i * out->linesize[1] *
232 (interleaved + sub_h * (1 - interleaved));
233 dst[2] = out->data[2] + i * out->linesize[2] *
234 (interleaved + sub_h * (1 - interleaved));
236 linesizes[0] = out->linesize[0] +
237 interleaved * out->linesize[0];
238 linesizes[1] = out->linesize[1] +
239 interleaved * out->linesize[1];
240 linesizes[2] = out->linesize[2] +
241 interleaved * out->linesize[2];
243 av_image_copy(dst, linesizes, src, s->input_views[i]->linesize,
244 s->input_views[i]->format,
245 s->input_views[i]->width,
246 s->input_views[i]->height);
250 static av_always_inline void spatial_frame_pack(AVFilterLink *outlink,
253 AVFilterContext *ctx = outlink->src;
254 FramepackContext *s = ctx->priv;
256 case AV_STEREO3D_SIDEBYSIDE:
257 horizontal_frame_pack(outlink, dst, 0);
259 case AV_STEREO3D_COLUMNS:
260 horizontal_frame_pack(outlink, dst, 1);
262 case AV_STEREO3D_TOPBOTTOM:
263 vertical_frame_pack(outlink, dst, 0);
265 case AV_STEREO3D_LINES:
266 vertical_frame_pack(outlink, dst, 1);
271 static int try_push_frame(AVFilterContext *ctx);
273 static int filter_frame_left(AVFilterLink *inlink, AVFrame *frame)
275 FramepackContext *s = inlink->dst->priv;
276 s->input_views[LEFT] = frame;
277 return try_push_frame(inlink->dst);
280 static int filter_frame_right(AVFilterLink *inlink, AVFrame *frame)
282 FramepackContext *s = inlink->dst->priv;
283 s->input_views[RIGHT] = frame;
284 return try_push_frame(inlink->dst);
287 static int request_frame(AVFilterLink *outlink)
289 AVFilterContext *ctx = outlink->src;
290 FramepackContext *s = ctx->priv;
293 /* get a frame on the either input, stop as soon as a video ends */
294 for (i = 0; i < 2; i++) {
295 if (!s->input_views[i]) {
296 ret = ff_request_frame(ctx->inputs[i]);
304 static int try_push_frame(AVFilterContext *ctx)
306 FramepackContext *s = ctx->priv;
307 AVFilterLink *outlink = ctx->outputs[0];
311 if (!(s->input_views[0] && s->input_views[1]))
313 if (s->format == AV_STEREO3D_FRAMESEQUENCE) {
314 if (s->double_pts == AV_NOPTS_VALUE)
315 s->double_pts = s->input_views[LEFT]->pts;
317 for (i = 0; i < 2; i++) {
318 // set correct timestamps
319 s->input_views[i]->pts = s->double_pts++;
321 // set stereo3d side data
322 stereo = av_stereo3d_create_side_data(s->input_views[i]);
324 return AVERROR(ENOMEM);
325 stereo->type = s->format;
327 // filter the frame and immediately relinquish its pointer
328 ret = ff_filter_frame(outlink, s->input_views[i]);
329 s->input_views[i] = NULL;
335 AVFrame *dst = ff_get_video_buffer(outlink, outlink->w, outlink->h);
337 return AVERROR(ENOMEM);
339 spatial_frame_pack(outlink, dst);
341 // get any property from the original frame
342 ret = av_frame_copy_props(dst, s->input_views[LEFT]);
348 for (i = 0; i < 2; i++)
349 av_frame_free(&s->input_views[i]);
351 // set stereo3d side data
352 stereo = av_stereo3d_create_side_data(dst);
355 return AVERROR(ENOMEM);
357 stereo->type = s->format;
359 return ff_filter_frame(outlink, dst);
363 #define OFFSET(x) offsetof(FramepackContext, x)
364 #define V AV_OPT_FLAG_VIDEO_PARAM
365 static const AVOption framepack_options[] = {
366 { "format", "Frame pack output format", OFFSET(format), AV_OPT_TYPE_INT,
367 { .i64 = AV_STEREO3D_SIDEBYSIDE }, 0, INT_MAX, .flags = V, .unit = "format" },
368 { "sbs", "Views are packed next to each other", 0, AV_OPT_TYPE_CONST,
369 { .i64 = AV_STEREO3D_SIDEBYSIDE }, INT_MIN, INT_MAX, .flags = V, .unit = "format" },
370 { "tab", "Views are packed on top of each other", 0, AV_OPT_TYPE_CONST,
371 { .i64 = AV_STEREO3D_TOPBOTTOM }, INT_MIN, INT_MAX, .flags = V, .unit = "format" },
372 { "frameseq", "Views are one after the other", 0, AV_OPT_TYPE_CONST,
373 { .i64 = AV_STEREO3D_FRAMESEQUENCE }, INT_MIN, INT_MAX, .flags = V, .unit = "format" },
374 { "lines", "Views are interleaved by lines", 0, AV_OPT_TYPE_CONST,
375 { .i64 = AV_STEREO3D_LINES }, INT_MIN, INT_MAX, .flags = V, .unit = "format" },
376 { "columns", "Views are interleaved by columns", 0, AV_OPT_TYPE_CONST,
377 { .i64 = AV_STEREO3D_COLUMNS }, INT_MIN, INT_MAX, .flags = V, .unit = "format" },
381 AVFILTER_DEFINE_CLASS(framepack);
383 static const AVFilterPad framepack_inputs[] = {
386 .type = AVMEDIA_TYPE_VIDEO,
387 .filter_frame = filter_frame_left,
392 .type = AVMEDIA_TYPE_VIDEO,
393 .filter_frame = filter_frame_right,
399 static const AVFilterPad framepack_outputs[] = {
402 .type = AVMEDIA_TYPE_VIDEO,
403 .config_props = config_output,
404 .request_frame = request_frame,
409 AVFilter ff_vf_framepack = {
411 .description = NULL_IF_CONFIG_SMALL("Generate a frame packed stereoscopic video."),
412 .priv_size = sizeof(FramepackContext),
413 .priv_class = &framepack_class,
414 .query_formats = query_formats,
415 .inputs = framepack_inputs,
416 .outputs = framepack_outputs,
417 .uninit = framepack_uninit,