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
53 static const enum AVPixelFormat formats_supported[] = {
54 AV_PIX_FMT_YUV420P, AV_PIX_FMT_YUV422P, AV_PIX_FMT_YUV444P,
55 AV_PIX_FMT_YUV410P, AV_PIX_FMT_YUVA420P, AV_PIX_FMT_YUVJ420P,
56 AV_PIX_FMT_YUVJ422P, AV_PIX_FMT_YUVJ444P, AV_PIX_FMT_YUVJ440P,
60 static int query_formats(AVFilterContext *ctx)
62 // this will ensure that formats are the same on all pads
63 AVFilterFormats *fmts_list = ff_make_format_list(formats_supported);
65 return AVERROR(ENOMEM);
66 return ff_set_common_formats(ctx, fmts_list);
69 static av_cold void framepack_uninit(AVFilterContext *ctx)
71 FramepackContext *s = ctx->priv;
73 // clean any leftover frame
74 av_frame_free(&s->input_views[LEFT]);
75 av_frame_free(&s->input_views[RIGHT]);
78 static int config_output(AVFilterLink *outlink)
80 AVFilterContext *ctx = outlink->src;
81 FramepackContext *s = outlink->src->priv;
83 int width = ctx->inputs[LEFT]->w;
84 int height = ctx->inputs[LEFT]->h;
85 AVRational time_base = ctx->inputs[LEFT]->time_base;
86 AVRational frame_rate = ctx->inputs[LEFT]->frame_rate;
88 // check size and fps match on the other input
89 if (width != ctx->inputs[RIGHT]->w ||
90 height != ctx->inputs[RIGHT]->h) {
91 av_log(ctx, AV_LOG_ERROR,
92 "Left and right sizes differ (%dx%d vs %dx%d).\n",
94 ctx->inputs[RIGHT]->w, ctx->inputs[RIGHT]->h);
95 return AVERROR_INVALIDDATA;
96 } else if (av_cmp_q(time_base, ctx->inputs[RIGHT]->time_base) != 0) {
97 av_log(ctx, AV_LOG_ERROR,
98 "Left and right time bases differ (%d/%d vs %d/%d).\n",
99 time_base.num, time_base.den,
100 ctx->inputs[RIGHT]->time_base.num,
101 ctx->inputs[RIGHT]->time_base.den);
102 return AVERROR_INVALIDDATA;
103 } else if (av_cmp_q(frame_rate, ctx->inputs[RIGHT]->frame_rate) != 0) {
104 av_log(ctx, AV_LOG_ERROR,
105 "Left and right framerates differ (%d/%d vs %d/%d).\n",
106 frame_rate.num, frame_rate.den,
107 ctx->inputs[RIGHT]->frame_rate.num,
108 ctx->inputs[RIGHT]->frame_rate.den);
109 return AVERROR_INVALIDDATA;
112 s->pix_desc = av_pix_fmt_desc_get(outlink->format);
116 // modify output properties as needed
118 case AV_STEREO3D_FRAMESEQUENCE:
122 case AV_STEREO3D_COLUMNS:
123 case AV_STEREO3D_SIDEBYSIDE:
126 case AV_STEREO3D_LINES:
127 case AV_STEREO3D_TOPBOTTOM:
131 av_log(ctx, AV_LOG_ERROR, "Unknown packing mode.");
132 return AVERROR_INVALIDDATA;
137 outlink->time_base = time_base;
138 outlink->frame_rate = frame_rate;
143 static void horizontal_frame_pack(AVFilterLink *outlink,
147 AVFilterContext *ctx = outlink->src;
148 FramepackContext *s = ctx->priv;
152 const uint8_t *leftp = s->input_views[LEFT]->data[0];
153 const uint8_t *rightp = s->input_views[RIGHT]->data[0];
154 uint8_t *dstp = out->data[0];
155 int length = out->width / 2;
156 int lines = out->height;
158 for (plane = 0; plane < s->pix_desc->nb_components; plane++) {
159 if (plane == 1 || plane == 2) {
160 length = AV_CEIL_RSHIFT(out->width / 2, s->pix_desc->log2_chroma_w);
161 lines = AV_CEIL_RSHIFT(out->height, s->pix_desc->log2_chroma_h);
163 for (i = 0; i < lines; i++) {
165 leftp = s->input_views[LEFT]->data[plane] +
166 s->input_views[LEFT]->linesize[plane] * i;
167 rightp = s->input_views[RIGHT]->data[plane] +
168 s->input_views[RIGHT]->linesize[plane] * i;
169 dstp = out->data[plane] + out->linesize[plane] * i;
170 for (j = 0; j < length; j++) {
171 // interpolate chroma as necessary
172 if ((s->pix_desc->log2_chroma_w ||
173 s->pix_desc->log2_chroma_h) &&
174 (plane == 1 || plane == 2)) {
175 *dstp++ = (*leftp + *rightp) / 2;
176 *dstp++ = (*leftp + *rightp) / 2;
187 for (i = 0; i < 2; i++) {
188 const uint8_t *src[4];
190 int sub_w = s->input_views[i]->width >> s->pix_desc->log2_chroma_w;
192 src[0] = s->input_views[i]->data[0];
193 src[1] = s->input_views[i]->data[1];
194 src[2] = s->input_views[i]->data[2];
196 dst[0] = out->data[0] + i * s->input_views[i]->width;
197 dst[1] = out->data[1] + i * sub_w;
198 dst[2] = out->data[2] + i * sub_w;
200 av_image_copy(dst, out->linesize, src, s->input_views[i]->linesize,
201 s->input_views[i]->format,
202 s->input_views[i]->width,
203 s->input_views[i]->height);
208 static void vertical_frame_pack(AVFilterLink *outlink,
212 AVFilterContext *ctx = outlink->src;
213 FramepackContext *s = ctx->priv;
216 for (i = 0; i < 2; i++) {
217 const uint8_t *src[4];
220 int sub_h = s->input_views[i]->height >> s->pix_desc->log2_chroma_h;
222 src[0] = s->input_views[i]->data[0];
223 src[1] = s->input_views[i]->data[1];
224 src[2] = s->input_views[i]->data[2];
226 dst[0] = out->data[0] + i * out->linesize[0] *
227 (interleaved + s->input_views[i]->height * (1 - interleaved));
228 dst[1] = out->data[1] + i * out->linesize[1] *
229 (interleaved + sub_h * (1 - interleaved));
230 dst[2] = out->data[2] + i * out->linesize[2] *
231 (interleaved + sub_h * (1 - interleaved));
233 linesizes[0] = out->linesize[0] +
234 interleaved * out->linesize[0];
235 linesizes[1] = out->linesize[1] +
236 interleaved * out->linesize[1];
237 linesizes[2] = out->linesize[2] +
238 interleaved * out->linesize[2];
240 av_image_copy(dst, linesizes, src, s->input_views[i]->linesize,
241 s->input_views[i]->format,
242 s->input_views[i]->width,
243 s->input_views[i]->height);
247 static av_always_inline void spatial_frame_pack(AVFilterLink *outlink,
250 AVFilterContext *ctx = outlink->src;
251 FramepackContext *s = ctx->priv;
253 case AV_STEREO3D_SIDEBYSIDE:
254 horizontal_frame_pack(outlink, dst, 0);
256 case AV_STEREO3D_COLUMNS:
257 horizontal_frame_pack(outlink, dst, 1);
259 case AV_STEREO3D_TOPBOTTOM:
260 vertical_frame_pack(outlink, dst, 0);
262 case AV_STEREO3D_LINES:
263 vertical_frame_pack(outlink, dst, 1);
268 static int try_push_frame(AVFilterContext *ctx);
270 static int filter_frame_left(AVFilterLink *inlink, AVFrame *frame)
272 FramepackContext *s = inlink->dst->priv;
273 s->input_views[LEFT] = frame;
274 return try_push_frame(inlink->dst);
277 static int filter_frame_right(AVFilterLink *inlink, AVFrame *frame)
279 FramepackContext *s = inlink->dst->priv;
280 s->input_views[RIGHT] = frame;
281 return try_push_frame(inlink->dst);
284 static int request_frame(AVFilterLink *outlink)
286 AVFilterContext *ctx = outlink->src;
287 FramepackContext *s = ctx->priv;
290 /* get a frame on the either input, stop as soon as a video ends */
291 for (i = 0; i < 2; i++) {
292 if (!s->input_views[i]) {
293 ret = ff_request_frame(ctx->inputs[i]);
301 static int try_push_frame(AVFilterContext *ctx)
303 FramepackContext *s = ctx->priv;
304 AVFilterLink *outlink = ctx->outputs[0];
308 if (!(s->input_views[0] && s->input_views[1]))
310 if (s->format == AV_STEREO3D_FRAMESEQUENCE) {
311 int64_t pts = s->input_views[0]->pts;
313 for (i = 0; i < 2; i++) {
314 // set correct timestamps
315 if (pts != AV_NOPTS_VALUE)
316 s->input_views[i]->pts = i == 0 ? pts * 2 : s->input_views[1]->pts + pts;
318 // set stereo3d side data
319 stereo = av_stereo3d_create_side_data(s->input_views[i]);
321 return AVERROR(ENOMEM);
322 stereo->type = s->format;
323 stereo->view = i == LEFT ? AV_STEREO3D_VIEW_LEFT
324 : AV_STEREO3D_VIEW_RIGHT;
326 // filter the frame and immediately relinquish its pointer
327 ret = ff_filter_frame(outlink, s->input_views[i]);
328 s->input_views[i] = NULL;
334 AVFrame *dst = ff_get_video_buffer(outlink, outlink->w, outlink->h);
336 return AVERROR(ENOMEM);
338 spatial_frame_pack(outlink, dst);
340 // get any property from the original frame
341 ret = av_frame_copy_props(dst, s->input_views[LEFT]);
347 for (i = 0; i < 2; i++)
348 av_frame_free(&s->input_views[i]);
350 // set stereo3d side data
351 stereo = av_stereo3d_create_side_data(dst);
354 return AVERROR(ENOMEM);
356 stereo->type = s->format;
358 return ff_filter_frame(outlink, dst);
362 #define OFFSET(x) offsetof(FramepackContext, x)
363 #define VF AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_FILTERING_PARAM
364 static const AVOption framepack_options[] = {
365 { "format", "Frame pack output format", OFFSET(format), AV_OPT_TYPE_INT,
366 { .i64 = AV_STEREO3D_SIDEBYSIDE }, 0, INT_MAX, .flags = VF, .unit = "format" },
367 { "sbs", "Views are packed next to each other", 0, AV_OPT_TYPE_CONST,
368 { .i64 = AV_STEREO3D_SIDEBYSIDE }, INT_MIN, INT_MAX, .flags = VF, .unit = "format" },
369 { "tab", "Views are packed on top of each other", 0, AV_OPT_TYPE_CONST,
370 { .i64 = AV_STEREO3D_TOPBOTTOM }, INT_MIN, INT_MAX, .flags = VF, .unit = "format" },
371 { "frameseq", "Views are one after the other", 0, AV_OPT_TYPE_CONST,
372 { .i64 = AV_STEREO3D_FRAMESEQUENCE }, INT_MIN, INT_MAX, .flags = VF, .unit = "format" },
373 { "lines", "Views are interleaved by lines", 0, AV_OPT_TYPE_CONST,
374 { .i64 = AV_STEREO3D_LINES }, INT_MIN, INT_MAX, .flags = VF, .unit = "format" },
375 { "columns", "Views are interleaved by columns", 0, AV_OPT_TYPE_CONST,
376 { .i64 = AV_STEREO3D_COLUMNS }, INT_MIN, INT_MAX, .flags = VF, .unit = "format" },
380 AVFILTER_DEFINE_CLASS(framepack);
382 static const AVFilterPad framepack_inputs[] = {
385 .type = AVMEDIA_TYPE_VIDEO,
386 .filter_frame = filter_frame_left,
391 .type = AVMEDIA_TYPE_VIDEO,
392 .filter_frame = filter_frame_right,
398 static const AVFilterPad framepack_outputs[] = {
401 .type = AVMEDIA_TYPE_VIDEO,
402 .config_props = config_output,
403 .request_frame = request_frame,
408 AVFilter ff_vf_framepack = {
410 .description = NULL_IF_CONFIG_SMALL("Generate a frame packed stereoscopic video."),
411 .priv_size = sizeof(FramepackContext),
412 .priv_class = &framepack_class,
413 .query_formats = query_formats,
414 .inputs = framepack_inputs,
415 .outputs = framepack_outputs,
416 .uninit = framepack_uninit,