FFmpeg coverage

Directory: ../../../ffmpeg/
File: src/libavfilter/vf_scdet.c
Date: 2026-08-31 17:12:07
Exec Total Coverage
Lines: 67 70 95.7%
Functions: 5 5 100.0%
Branches: 33 42 78.6%
Line Branch Exec Source
1 /*
2 * This file is part of FFmpeg.
3 *
4 * FFmpeg is free software; you can redistribute it and/or
5 * modify it under the terms of the GNU Lesser General Public
6 * License as published by the Free Software Foundation; either
7 * version 2.1 of the License, or (at your option) any later version.
8 *
9 * FFmpeg is distributed in the hope that it will be useful,
10 * but WITHOUT ANY WARRANTY; without even the implied warranty of
11 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
12 * Lesser General Public License for more details.
13 *
14 * You should have received a copy of the GNU Lesser General Public
15 * License along with FFmpeg; if not, write to the Free Software
16 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
17 */
18
19 /**
20 * @file
21 * video scene change detection filter
22 */
23
24 #include "libavutil/imgutils.h"
25 #include "libavutil/opt.h"
26 #include "libavutil/pixdesc.h"
27 #include "libavutil/timestamp.h"
28
29 #include "avfilter.h"
30 #include "filters.h"
31 #include "scene_sad.h"
32 #include "video.h"
33
34 typedef struct SCDetContext {
35 const AVClass *class;
36
37 ptrdiff_t width[4];
38 ptrdiff_t height[4];
39 int nb_planes;
40 int bitdepth;
41 ff_scene_sad_fn sad;
42 double prev_mafd;
43 double scene_score;
44 AVFrame *prev_picref;
45 double threshold;
46 int sc_pass;
47 } SCDetContext;
48
49 #define OFFSET(x) offsetof(SCDetContext, x)
50 #define V AV_OPT_FLAG_VIDEO_PARAM
51 #define F AV_OPT_FLAG_FILTERING_PARAM
52
53 static const AVOption scdet_options[] = {
54 { "threshold", "set scene change detect threshold", OFFSET(threshold), AV_OPT_TYPE_DOUBLE, {.dbl = 10.}, 0, 100., V|F },
55 { "t", "set scene change detect threshold", OFFSET(threshold), AV_OPT_TYPE_DOUBLE, {.dbl = 10.}, 0, 100., V|F },
56 { "sc_pass", "Set the flag to pass scene change frames", OFFSET(sc_pass), AV_OPT_TYPE_BOOL, {.i64 = 0 }, 0, 1, V|F },
57 { "s", "Set the flag to pass scene change frames", OFFSET(sc_pass), AV_OPT_TYPE_BOOL, {.i64 = 0 }, 0, 1, V|F },
58 {NULL}
59 };
60
61 AVFILTER_DEFINE_CLASS(scdet);
62
63 static const enum AVPixelFormat pix_fmts[] = {
64 AV_PIX_FMT_RGB24, AV_PIX_FMT_BGR24,
65 AV_PIX_FMT_ARGB, AV_PIX_FMT_RGBA,
66 AV_PIX_FMT_ABGR, AV_PIX_FMT_BGRA,
67 AV_PIX_FMT_RGB48, AV_PIX_FMT_BGR48,
68 AV_PIX_FMT_RGBA64, AV_PIX_FMT_BGRA64,
69
70 AV_PIX_FMT_GBRP, AV_PIX_FMT_GBRP9, AV_PIX_FMT_GBRP10,
71 AV_PIX_FMT_GBRP12, AV_PIX_FMT_GBRP14, AV_PIX_FMT_GBRP16,
72 AV_PIX_FMT_GBRAP, AV_PIX_FMT_GBRAP10, AV_PIX_FMT_GBRAP12,
73 AV_PIX_FMT_GBRAP14, AV_PIX_FMT_GBRAP16,
74 AV_PIX_FMT_GBRP10MSB, AV_PIX_FMT_GBRP12MSB,
75
76 AV_PIX_FMT_GRAY8, AV_PIX_FMT_GRAY9, AV_PIX_FMT_GRAY10,
77 AV_PIX_FMT_GRAY12, AV_PIX_FMT_GRAY14, AV_PIX_FMT_GRAY16,
78 AV_PIX_FMT_YA8, AV_PIX_FMT_YA16,
79
80 AV_PIX_FMT_YUV410P, AV_PIX_FMT_YUV411P, AV_PIX_FMT_YUVJ411P,
81 AV_PIX_FMT_YUV420P, AV_PIX_FMT_YUVJ420P,
82 AV_PIX_FMT_YUV422P, AV_PIX_FMT_YUVJ422P,
83 AV_PIX_FMT_YUV440P, AV_PIX_FMT_YUVJ440P,
84 AV_PIX_FMT_YUV444P, AV_PIX_FMT_YUVJ444P,
85 AV_PIX_FMT_YUV420P9, AV_PIX_FMT_YUV422P9, AV_PIX_FMT_YUV444P9,
86 AV_PIX_FMT_YUV420P10, AV_PIX_FMT_YUV422P10,
87 AV_PIX_FMT_YUV440P10, AV_PIX_FMT_YUV444P10,
88 AV_PIX_FMT_YUV420P12, AV_PIX_FMT_YUV422P12,
89 AV_PIX_FMT_YUV440P12, AV_PIX_FMT_YUV444P12,
90 AV_PIX_FMT_YUV420P14, AV_PIX_FMT_YUV422P14, AV_PIX_FMT_YUV444P14,
91 AV_PIX_FMT_YUV420P16, AV_PIX_FMT_YUV422P16, AV_PIX_FMT_YUV444P16,
92
93 AV_PIX_FMT_YUVA420P, AV_PIX_FMT_YUVA422P, AV_PIX_FMT_YUVA444P,
94 AV_PIX_FMT_YUVA420P9, AV_PIX_FMT_YUVA422P9, AV_PIX_FMT_YUVA444P9,
95 AV_PIX_FMT_YUVA420P10, AV_PIX_FMT_YUVA422P10, AV_PIX_FMT_YUVA444P10,
96 AV_PIX_FMT_YUVA422P12, AV_PIX_FMT_YUVA444P12,
97 AV_PIX_FMT_YUVA420P16, AV_PIX_FMT_YUVA422P16, AV_PIX_FMT_YUVA444P16,
98
99 AV_PIX_FMT_YUV444P10MSB, AV_PIX_FMT_YUV444P12MSB,
100
101 AV_PIX_FMT_NV12, AV_PIX_FMT_NV21, AV_PIX_FMT_NV16,
102 AV_PIX_FMT_NV24, AV_PIX_FMT_NV42, AV_PIX_FMT_NV20,
103 AV_PIX_FMT_P010, AV_PIX_FMT_P210, AV_PIX_FMT_P410,
104 AV_PIX_FMT_P012, AV_PIX_FMT_P212, AV_PIX_FMT_P412,
105 AV_PIX_FMT_P016, AV_PIX_FMT_P216, AV_PIX_FMT_P416,
106
107 AV_PIX_FMT_NONE
108 };
109
110 2 static int config_input(AVFilterLink *inlink)
111 {
112 2 AVFilterContext *ctx = inlink->dst;
113 2 SCDetContext *s = ctx->priv;
114 2 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(inlink->format);
115 6 int is_yuv = !(desc->flags & AV_PIX_FMT_FLAG_RGB) &&
116
2/4
✓ Branch 0 taken 2 times.
✗ Branch 1 not taken.
✓ Branch 2 taken 2 times.
✗ Branch 3 not taken.
 4 (desc->flags & AV_PIX_FMT_FLAG_PLANAR) &&
117
1/2
✓ Branch 0 taken 2 times.
✗ Branch 1 not taken.
 2 desc->nb_components >= 3;
118
119 /* SAD scores are normalize by the bit-depth, so MSB-aligned formats are
120 * equivalent to their corresponding full depth representation */
121 2 s->bitdepth = desc->comp[0].depth + desc->comp[0].shift;
122
1/2
✗ Branch 0 not taken.
✓ Branch 1 taken 2 times.
 2 s->nb_planes = is_yuv ? 1 : av_pix_fmt_count_planes(inlink->format);
123
124
2/2
✓ Branch 0 taken 8 times.
✓ Branch 1 taken 2 times.
 10 for (int plane = 0; plane < 4; plane++) {
125 8 ptrdiff_t line_size = av_image_get_linesize(inlink->format, inlink->w, plane);
126 8 s->width[plane] = line_size >> (s->bitdepth > 8);
127
4/4
✓ Branch 0 taken 6 times.
✓ Branch 1 taken 2 times.
✓ Branch 2 taken 2 times.
✓ Branch 3 taken 4 times.
 8 s->height[plane] = inlink->h >> ((plane == 1 || plane == 2) ? desc->log2_chroma_h : 0);
128 }
129
130 2 s->sad = ff_scene_sad_get_fn(s->bitdepth);
131
1/2
✗ Branch 0 not taken.
✓ Branch 1 taken 2 times.
 2 if (!s->sad)
132 ✗ return AVERROR(EINVAL);
133
134 2 return 0;
135 }
136
137 3 static av_cold void uninit(AVFilterContext *ctx)
138 {
139 3 SCDetContext *s = ctx->priv;
140
141 3 av_frame_free(&s->prev_picref);
142 3 }
143
144 1700 static double get_scene_score(AVFilterContext *ctx, AVFrame *frame)
145 {
146 1700 double ret = 0;
147 1700 SCDetContext *s = ctx->priv;
148 1700 AVFrame *prev_picref = s->prev_picref;
149
150
3/4
✓ Branch 0 taken 1698 times.
✓ Branch 1 taken 2 times.
✓ Branch 2 taken 1698 times.
✗ Branch 3 not taken.
 1700 if (prev_picref && frame->height == prev_picref->height
151
1/2
✓ Branch 0 taken 1698 times.
✗ Branch 1 not taken.
 1698 && frame->width == prev_picref->width) {
152 1698 uint64_t sad = 0;
153 double mafd, diff;
154 1698 uint64_t count = 0;
155
156
2/2
✓ Branch 0 taken 1698 times.
✓ Branch 1 taken 1698 times.
 3396 for (int plane = 0; plane < s->nb_planes; plane++) {
157 uint64_t plane_sad;
158 1698 s->sad(prev_picref->data[plane], prev_picref->linesize[plane],
159 1698 frame->data[plane], frame->linesize[plane],
160 s->width[plane], s->height[plane], &plane_sad);
161 1698 sad += plane_sad;
162 1698 count += s->width[plane] * s->height[plane];
163 }
164
165 1698 mafd = (double)sad * 100. / count / (1ULL << s->bitdepth);
166 1698 diff = fabs(mafd - s->prev_mafd);
167
2/2
✓ Branch 0 taken 1441 times.
✓ Branch 1 taken 257 times.
 1698 ret = av_clipf(FFMIN(mafd, diff), 0, 100.);
168 1698 s->prev_mafd = mafd;
169 1698 av_frame_free(&prev_picref);
170 }
171 1700 s->prev_picref = av_frame_clone(frame);
172 1700 return ret;
173 }
174
175 3416 static int set_meta(SCDetContext *s, AVFrame *frame, const char *key, const char *value)
176 {
177 3416 return av_dict_set(&frame->metadata, key, value, 0);
178 }
179
180 2107 static int activate(AVFilterContext *ctx)
181 {
182 int ret;
183 2107 AVFilterLink *inlink = ctx->inputs[0];
184 2107 AVFilterLink *outlink = ctx->outputs[0];
185 2107 SCDetContext *s = ctx->priv;
186 AVFrame *frame;
187
188
2/2
✓ Branch 1 taken 1 times.
✓ Branch 2 taken 2106 times.
 2107 FF_FILTER_FORWARD_STATUS_BACK(outlink, inlink);
189
190 2106 ret = ff_inlink_consume_frame(inlink, &frame);
191
1/2
✗ Branch 0 not taken.
✓ Branch 1 taken 2106 times.
 2106 if (ret < 0)
192 ✗ return ret;
193
194
2/2
✓ Branch 0 taken 1700 times.
✓ Branch 1 taken 406 times.
 2106 if (frame) {
195 char buf[64];
196 1700 s->scene_score = get_scene_score(ctx, frame);
197 1700 snprintf(buf, sizeof(buf), "%0.3f", s->prev_mafd);
198 1700 set_meta(s, frame, "lavfi.scd.mafd", buf);
199 1700 snprintf(buf, sizeof(buf), "%0.3f", s->scene_score);
200 1700 set_meta(s, frame, "lavfi.scd.score", buf);
201
202
2/2
✓ Branch 0 taken 16 times.
✓ Branch 1 taken 1684 times.
 1700 if (s->scene_score >= s->threshold) {
203 16 av_log(ctx, AV_LOG_INFO, "lavfi.scd.score: %.3f, lavfi.scd.time: %s\n",
204 16 s->scene_score, av_ts2timestr(frame->pts, &inlink->time_base));
205 16 set_meta(s, frame, "lavfi.scd.time",
206 16 av_ts2timestr(frame->pts, &inlink->time_base));
207 }
208
2/2
✓ Branch 0 taken 1308 times.
✓ Branch 1 taken 392 times.
 1700 if (s->sc_pass) {
209
2/2
✓ Branch 0 taken 11 times.
✓ Branch 1 taken 1297 times.
 1308 if (s->scene_score >= s->threshold)
210 403 return ff_filter_frame(outlink, frame);
211 else {
212 1297 av_frame_free(&frame);
213 }
214 } else
215 392 return ff_filter_frame(outlink, frame);
216 }
217
218
2/2
✓ Branch 1 taken 2 times.
✓ Branch 2 taken 1701 times.
 1703 FF_FILTER_FORWARD_STATUS(inlink, outlink);
219
1/2
✓ Branch 1 taken 1701 times.
✗ Branch 2 not taken.
 1701 FF_FILTER_FORWARD_WANTED(outlink, inlink);
220
221 ✗ return FFERROR_NOT_READY;
222 }
223
224 static const AVFilterPad scdet_inputs[] = {
225 {
226 .name = "default",
227 .type = AVMEDIA_TYPE_VIDEO,
228 .config_props = config_input,
229 },
230 };
231
232 const FFFilter ff_vf_scdet = {
233 .p.name = "scdet",
234 .p.description = NULL_IF_CONFIG_SMALL("Detect video scene change"),
235 .p.priv_class = &scdet_class,
236 .p.flags = AVFILTER_FLAG_METADATA_ONLY,
237 .priv_size = sizeof(SCDetContext),
238 .uninit = uninit,
239 FILTER_INPUTS(scdet_inputs),
240 FILTER_OUTPUTS(ff_video_default_filterpad),
241 FILTER_PIXFMTS_ARRAY(pix_fmts),
242 .activate = activate,
243 };
244