| Line | Branch | Exec | Source |
|---|---|---|---|
| 1 | /* | ||
| 2 | * Copyright (c) 2003 Rich Felker | ||
| 3 | * Copyright (c) 2012 Stefano Sabatini | ||
| 4 | * Copyright (c) 2026 Dawid Stachowiak | ||
| 5 | * | ||
| 6 | * This file is part of FFmpeg. | ||
| 7 | * | ||
| 8 | * FFmpeg is free software; you can redistribute it and/or modify | ||
| 9 | * it under the terms of the GNU General Public License as published by | ||
| 10 | * the Free Software Foundation; either version 2 of the License, or | ||
| 11 | * (at your option) any later version. | ||
| 12 | * | ||
| 13 | * FFmpeg is distributed in the hope that it will be useful, | ||
| 14 | * but WITHOUT ANY WARRANTY; without even the implied warranty of | ||
| 15 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | ||
| 16 | * GNU General Public License for more details. | ||
| 17 | * | ||
| 18 | * You should have received a copy of the GNU General Public License along | ||
| 19 | * with FFmpeg; if not, write to the Free Software Foundation, Inc., | ||
| 20 | * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA. | ||
| 21 | */ | ||
| 22 | |||
| 23 | /** | ||
| 24 | * @file mpdecimate filter, ported from libmpcodecs/vf_decimate.c by | ||
| 25 | * Rich Felker. | ||
| 26 | */ | ||
| 27 | |||
| 28 | #include "libavutil/opt.h" | ||
| 29 | #include "libavutil/pixdesc.h" | ||
| 30 | #include "libavutil/pixelutils.h" | ||
| 31 | #include "libavutil/timestamp.h" | ||
| 32 | #include "avfilter.h" | ||
| 33 | #include "filters.h" | ||
| 34 | #include "video.h" | ||
| 35 | |||
| 36 | typedef enum { | ||
| 37 | DECIMATE_DROP, ///< similar frame, past keep threshold — drop it | ||
| 38 | DECIMATE_KEEP_UPDATE, ///< keep frame and update reference (frame is different, or first frame) | ||
| 39 | DECIMATE_KEEP_NO_UPDATE,///< keep frame without updating reference (similar frame under keep threshold, or forced keep due to max_drop_count) | ||
| 40 | } DecimateResult; | ||
| 41 | |||
| 42 | typedef struct DecimateContext { | ||
| 43 | const AVClass *class; | ||
| 44 | int mode; ///< 0: drop similar frames, 1: drop similar and unique frames | ||
| 45 | int lo, hi; ///< lower and higher threshold number of differences | ||
| 46 | ///< values for 8x8 blocks | ||
| 47 | |||
| 48 | float frac; ///< threshold of changed pixels over the total fraction | ||
| 49 | |||
| 50 | int max_drop_count; ///< for mode 0: if positive: maximum number of sequential frames to drop | ||
| 51 | ///< for mode 0: if negative: minimum number of frames between two drops | ||
| 52 | |||
| 53 | int drop_count; ///< if positive: number of frames sequentially dropped | ||
| 54 | ///< if negative: number of sequential frames which were not dropped | ||
| 55 | |||
| 56 | int max_keep_count; ///< for mode 0: number of similar frames to ignore before to start dropping them | ||
| 57 | int keep_count; ///< for mode 0: number of similar frames already ignored | ||
| 58 | |||
| 59 | int min_dup_count; ///< for mode 1: minimum number of previous frames that need to be duplicated to keep frame | ||
| 60 | int dup_count; ///< for mode 1: number of duplicated frames | ||
| 61 | |||
| 62 | int hsub, vsub; ///< chroma subsampling values | ||
| 63 | AVFrame *ref; ///< reference picture | ||
| 64 | av_pixelutils_sad_fn sad; ///< sum of absolute difference function | ||
| 65 | } DecimateContext; | ||
| 66 | |||
| 67 | #define OFFSET(x) offsetof(DecimateContext, x) | ||
| 68 | #define FLAGS AV_OPT_FLAG_VIDEO_PARAM|AV_OPT_FLAG_FILTERING_PARAM | ||
| 69 | |||
| 70 | static const AVOption mpdecimate_options[] = { | ||
| 71 | { "max", "for mode 0: set the maximum number of consecutive dropped frames (positive), or the minimum interval between dropped frames (negative)", | ||
| 72 | OFFSET(max_drop_count), AV_OPT_TYPE_INT, {.i64=0}, INT_MIN, INT_MAX, FLAGS }, | ||
| 73 | { "keep", "for mode 0: set the number of similar consecutive frames to be kept before starting to drop similar frames", | ||
| 74 | OFFSET(max_keep_count), AV_OPT_TYPE_INT, {.i64=0}, 0, INT_MAX, FLAGS }, | ||
| 75 | { "hi", "set high dropping threshold", OFFSET(hi), AV_OPT_TYPE_INT, {.i64=64*12}, INT_MIN, INT_MAX, FLAGS }, | ||
| 76 | { "lo", "set low dropping threshold", OFFSET(lo), AV_OPT_TYPE_INT, {.i64=64*5}, INT_MIN, INT_MAX, FLAGS }, | ||
| 77 | { "frac", "set fraction dropping threshold", OFFSET(frac), AV_OPT_TYPE_FLOAT, {.dbl=0.33}, 0, 1, FLAGS }, | ||
| 78 | { "mode", "0: drop similar frames, 1: drop similar and unique frames", OFFSET(mode), AV_OPT_TYPE_INT, {.i64=0}, 0, 1, FLAGS }, | ||
| 79 | { "min", "for mode 1: set minimum number of previous frames that need to be duplicated to keep current frame", | ||
| 80 | OFFSET(min_dup_count), AV_OPT_TYPE_INT, {.i64=1}, 1, INT_MAX, FLAGS }, | ||
| 81 | { NULL } | ||
| 82 | }; | ||
| 83 | |||
| 84 | AVFILTER_DEFINE_CLASS(mpdecimate); | ||
| 85 | |||
| 86 | /** | ||
| 87 | * Return 1 if the two planes are different, 0 otherwise. | ||
| 88 | */ | ||
| 89 | 617 | static int diff_planes(AVFilterContext *ctx, | |
| 90 | uint8_t *cur, int cur_linesize, | ||
| 91 | uint8_t *ref, int ref_linesize, | ||
| 92 | int w, int h) | ||
| 93 | { | ||
| 94 | 617 | DecimateContext *decimate = ctx->priv; | |
| 95 | |||
| 96 | int x, y; | ||
| 97 | 617 | int d, c = 0; | |
| 98 | 617 | int t = (w/16)*(h/16)*decimate->frac; | |
| 99 | |||
| 100 | /* compute difference for blocks of 8x8 bytes */ | ||
| 101 |
2/2✓ Branch 0 taken 21181 times.
✓ Branch 1 taken 540 times.
|
21721 | for (y = 0; y < h-7; y += 4) { |
| 102 |
2/2✓ Branch 0 taken 1209144 times.
✓ Branch 1 taken 21104 times.
|
1230248 | for (x = 8; x < w-7; x += 4) { |
| 103 | 1209144 | d = decimate->sad(cur + y*cur_linesize + x, cur_linesize, | |
| 104 | 1209144 | ref + y*ref_linesize + x, ref_linesize); | |
| 105 |
2/2✓ Branch 0 taken 77 times.
✓ Branch 1 taken 1209067 times.
|
1209144 | if (d > decimate->hi) { |
| 106 | 77 | av_log(ctx, AV_LOG_DEBUG, "%d>=hi ", d); | |
| 107 | 77 | return 1; | |
| 108 | } | ||
| 109 |
2/2✓ Branch 0 taken 180 times.
✓ Branch 1 taken 1208887 times.
|
1209067 | if (d > decimate->lo) { |
| 110 | 180 | c++; | |
| 111 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 180 times.
|
180 | if (c > t) { |
| 112 | ✗ | av_log(ctx, AV_LOG_DEBUG, "lo:%d>=%d ", c, t); | |
| 113 | ✗ | return 1; | |
| 114 | } | ||
| 115 | } | ||
| 116 | } | ||
| 117 | } | ||
| 118 | |||
| 119 | 540 | av_log(ctx, AV_LOG_DEBUG, "lo:%d<%d ", c, t); | |
| 120 | 540 | return 0; | |
| 121 | } | ||
| 122 | |||
| 123 | /** | ||
| 124 | * Tell if the frame is different with respect to the reference frame ref. | ||
| 125 | */ | ||
| 126 | 257 | static int is_frame_different(AVFilterContext *ctx, | |
| 127 | AVFrame *cur, AVFrame *ref) | ||
| 128 | { | ||
| 129 | 257 | DecimateContext *decimate = ctx->priv; | |
| 130 | int plane; | ||
| 131 | |||
| 132 |
3/4✓ Branch 0 taken 617 times.
✓ Branch 1 taken 180 times.
✓ Branch 2 taken 617 times.
✗ Branch 3 not taken.
|
797 | for (plane = 0; ref->data[plane] && ref->linesize[plane]; plane++) { |
| 133 | /* use 8x8 SAD even on subsampled planes. The blocks won't match up with | ||
| 134 | * luma blocks, but hopefully nobody is depending on this to catch | ||
| 135 | * localized chroma changes that wouldn't exceed the thresholds when | ||
| 136 | * diluted by using what's effectively a larger block size. | ||
| 137 | */ | ||
| 138 |
4/4✓ Branch 0 taken 437 times.
✓ Branch 1 taken 180 times.
✓ Branch 2 taken 180 times.
✓ Branch 3 taken 257 times.
|
617 | int vsub = plane == 1 || plane == 2 ? decimate->vsub : 0; |
| 139 |
4/4✓ Branch 0 taken 437 times.
✓ Branch 1 taken 180 times.
✓ Branch 2 taken 180 times.
✓ Branch 3 taken 257 times.
|
617 | int hsub = plane == 1 || plane == 2 ? decimate->hsub : 0; |
| 140 |
2/2✓ Branch 0 taken 77 times.
✓ Branch 1 taken 540 times.
|
617 | if (diff_planes(ctx, |
| 141 | cur->data[plane], cur->linesize[plane], | ||
| 142 | ref->data[plane], ref->linesize[plane], | ||
| 143 | 617 | AV_CEIL_RSHIFT(ref->width, hsub), | |
| 144 | 617 | AV_CEIL_RSHIFT(ref->height, vsub))) | |
| 145 | 77 | return 1; | |
| 146 | } | ||
| 147 | |||
| 148 | 180 | return 0; | |
| 149 | } | ||
| 150 | |||
| 151 | /** | ||
| 152 | * Tell if the frame should be decimated, for example if it is no much | ||
| 153 | * different with respect to the reference frame ref. | ||
| 154 | */ | ||
| 155 | 185 | static DecimateResult decimate_frame(AVFilterContext *ctx, AVFrame *cur, AVFrame *ref) | |
| 156 | { | ||
| 157 | 185 | DecimateContext *decimate = ctx->priv; | |
| 158 | 185 | int is_similar = is_frame_different(ctx, cur, ref) == 0; | |
| 159 | |||
| 160 |
2/2✓ Branch 0 taken 45 times.
✓ Branch 1 taken 140 times.
|
185 | if (!is_similar) { |
| 161 | 45 | return DECIMATE_KEEP_UPDATE; | |
| 162 | } | ||
| 163 | /* Frame is similar - check if we must keep it due to drop limits */ | ||
| 164 |
2/2✓ Branch 0 taken 45 times.
✓ Branch 1 taken 95 times.
|
140 | if (decimate->max_drop_count > 0 && |
| 165 |
2/2✓ Branch 0 taken 10 times.
✓ Branch 1 taken 35 times.
|
45 | decimate->drop_count >= decimate->max_drop_count) |
| 166 | 10 | return DECIMATE_KEEP_NO_UPDATE; | |
| 167 |
2/2✓ Branch 0 taken 45 times.
✓ Branch 1 taken 85 times.
|
130 | if (decimate->max_drop_count < 0 && |
| 168 |
2/2✓ Branch 0 taken 30 times.
✓ Branch 1 taken 15 times.
|
45 | (decimate->drop_count - 1) > decimate->max_drop_count) |
| 169 | 30 | return DECIMATE_KEEP_NO_UPDATE; | |
| 170 | |||
| 171 | /* Frame is similar - check if we must keep it due to keep option */ | ||
| 172 |
4/4✓ Branch 0 taken 40 times.
✓ Branch 1 taken 60 times.
✓ Branch 2 taken 30 times.
✓ Branch 3 taken 10 times.
|
100 | if (decimate->max_keep_count > 0 && decimate->keep_count > -1 && |
| 173 |
2/2✓ Branch 0 taken 20 times.
✓ Branch 1 taken 10 times.
|
30 | decimate->keep_count < decimate->max_keep_count) { |
| 174 | 20 | decimate->keep_count++; | |
| 175 | 20 | return DECIMATE_KEEP_NO_UPDATE; | |
| 176 | } | ||
| 177 | 80 | return DECIMATE_DROP; | |
| 178 | } | ||
| 179 | |||
| 180 | 14 | static av_cold int init(AVFilterContext *ctx) | |
| 181 | { | ||
| 182 | 14 | DecimateContext *decimate = ctx->priv; | |
| 183 | |||
| 184 | 14 | decimate->sad = av_pixelutils_get_sad_fn(3, 3, 0, ctx); // 8x8, not aligned on blocksize | |
| 185 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 14 times.
|
14 | if (!decimate->sad) |
| 186 | ✗ | return AVERROR(EINVAL); | |
| 187 | |||
| 188 |
2/2✓ Branch 0 taken 4 times.
✓ Branch 1 taken 10 times.
|
14 | if (decimate->mode == 1) { |
| 189 | 4 | av_log(ctx, AV_LOG_VERBOSE, "min_dup_count:%d hi:%d lo:%d frac:%f\n", | |
| 190 | decimate->min_dup_count, decimate->hi, decimate->lo, | ||
| 191 | 4 | decimate->frac); | |
| 192 | } else { | ||
| 193 | 10 | av_log(ctx, AV_LOG_VERBOSE, "max_drop_count:%d hi:%d lo:%d frac:%f\n", | |
| 194 | 10 | decimate->max_drop_count, decimate->hi, decimate->lo, decimate->frac); | |
| 195 | } | ||
| 196 | 14 | return 0; | |
| 197 | } | ||
| 198 | |||
| 199 | 14 | static av_cold void uninit(AVFilterContext *ctx) | |
| 200 | { | ||
| 201 | 14 | DecimateContext *decimate = ctx->priv; | |
| 202 | 14 | av_frame_free(&decimate->ref); | |
| 203 | 14 | } | |
| 204 | |||
| 205 | static const enum AVPixelFormat pix_fmts[] = { | ||
| 206 | AV_PIX_FMT_YUV444P, AV_PIX_FMT_YUV422P, | ||
| 207 | AV_PIX_FMT_YUV420P, AV_PIX_FMT_YUV411P, | ||
| 208 | AV_PIX_FMT_YUV410P, AV_PIX_FMT_YUV440P, | ||
| 209 | AV_PIX_FMT_YUVJ444P, AV_PIX_FMT_YUVJ422P, | ||
| 210 | AV_PIX_FMT_YUVJ420P, AV_PIX_FMT_YUVJ440P, | ||
| 211 | AV_PIX_FMT_YUVA420P, | ||
| 212 | |||
| 213 | AV_PIX_FMT_GBRP, | ||
| 214 | |||
| 215 | AV_PIX_FMT_YUVA444P, | ||
| 216 | AV_PIX_FMT_YUVA422P, | ||
| 217 | |||
| 218 | AV_PIX_FMT_NONE | ||
| 219 | }; | ||
| 220 | |||
| 221 | 7 | static int config_input(AVFilterLink *inlink) | |
| 222 | { | ||
| 223 | 7 | AVFilterContext *ctx = inlink->dst; | |
| 224 | 7 | DecimateContext *decimate = ctx->priv; | |
| 225 | 7 | const AVPixFmtDescriptor *pix_desc = av_pix_fmt_desc_get(inlink->format); | |
| 226 | 7 | decimate->hsub = pix_desc->log2_chroma_w; | |
| 227 | 7 | decimate->vsub = pix_desc->log2_chroma_h; | |
| 228 | |||
| 229 | 7 | return 0; | |
| 230 | } | ||
| 231 | |||
| 232 | 190 | static int filter_frame_mode_0(AVFilterLink *inlink, AVFrame *cur) | |
| 233 | { | ||
| 234 | 190 | DecimateContext *decimate = inlink->dst->priv; | |
| 235 | 190 | AVFilterLink *outlink = inlink->dst->outputs[0]; | |
| 236 | 190 | AVFrame *out = NULL; | |
| 237 | int ret; | ||
| 238 |
2/2✓ Branch 0 taken 185 times.
✓ Branch 1 taken 5 times.
|
190 | DecimateResult result = decimate->ref ? decimate_frame(inlink->dst, cur, decimate->ref) : DECIMATE_KEEP_UPDATE; |
| 239 | |||
| 240 |
3/4✓ Branch 0 taken 80 times.
✓ Branch 1 taken 60 times.
✓ Branch 2 taken 50 times.
✗ Branch 3 not taken.
|
190 | switch (result) { |
| 241 | 80 | case DECIMATE_DROP: | |
| 242 | 80 | decimate->drop_count = FFMAX(1, decimate->drop_count+1); | |
| 243 | 80 | decimate->keep_count = -1; | |
| 244 | 80 | break; | |
| 245 | 60 | case DECIMATE_KEEP_NO_UPDATE: | |
| 246 | 60 | decimate->drop_count = FFMIN(-1, decimate->drop_count-1); | |
| 247 | 60 | out = cur; | |
| 248 | 60 | break; | |
| 249 | 50 | case DECIMATE_KEEP_UPDATE: | |
| 250 | 50 | out = av_frame_clone(cur); | |
| 251 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 50 times.
|
50 | if (!out) { |
| 252 | ✗ | av_frame_free(&cur); | |
| 253 | ✗ | return AVERROR(ENOMEM); | |
| 254 | } | ||
| 255 | 50 | av_frame_free(&decimate->ref); | |
| 256 | 50 | decimate->ref = cur; | |
| 257 | 50 | decimate->drop_count = FFMIN(-1, decimate->drop_count-1); | |
| 258 | 50 | decimate->keep_count = 0; | |
| 259 | 50 | break; | |
| 260 | } | ||
| 261 | |||
| 262 |
2/2✓ Branch 0 taken 80 times.
✓ Branch 1 taken 110 times.
|
190 | av_log(inlink->dst, AV_LOG_DEBUG, |
| 263 | "%s pts:%s pts_time:%s drop_count:%d keep_count:%d\n", | ||
| 264 | result == DECIMATE_DROP ? "drop" : "keep", | ||
| 265 | 190 | av_ts2str(cur->pts), av_ts2timestr(cur->pts, &inlink->time_base), | |
| 266 | decimate->drop_count, | ||
| 267 | decimate->keep_count); | ||
| 268 | |||
| 269 |
2/2✓ Branch 0 taken 80 times.
✓ Branch 1 taken 110 times.
|
190 | if (result == DECIMATE_DROP) { |
| 270 | 80 | av_frame_free(&cur); | |
| 271 | 80 | return 0; | |
| 272 | } | ||
| 273 | |||
| 274 | 110 | ret = ff_filter_frame(outlink, out); | |
| 275 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 110 times.
|
110 | if (ret < 0) |
| 276 | ✗ | return ret; | |
| 277 | |||
| 278 | 110 | return 0; | |
| 279 | } | ||
| 280 | |||
| 281 | 74 | static int filter_frame_mode_1(AVFilterLink *inlink, AVFrame *cur) | |
| 282 | { | ||
| 283 | 74 | DecimateContext *decimate = inlink->dst->priv; | |
| 284 | 74 | AVFilterLink *outlink = inlink->dst->outputs[0]; | |
| 285 | int ret; | ||
| 286 | 74 | AVFrame *out = NULL; | |
| 287 | |||
| 288 |
4/4✓ Branch 0 taken 72 times.
✓ Branch 1 taken 2 times.
✓ Branch 3 taken 32 times.
✓ Branch 4 taken 40 times.
|
74 | if (!decimate->ref || is_frame_different(inlink->dst, cur, decimate->ref)) { |
| 289 | 34 | AVFrame *ref = av_frame_clone(cur); | |
| 290 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 34 times.
|
34 | if (!ref) { |
| 291 | ✗ | av_frame_free(&cur); | |
| 292 | ✗ | return AVERROR(ENOMEM); | |
| 293 | } | ||
| 294 | 34 | av_frame_free(&decimate->ref); | |
| 295 | 34 | decimate->ref = ref; | |
| 296 | 34 | decimate->dup_count = 0; | |
| 297 | } else { | ||
| 298 | 40 | decimate->dup_count++; | |
| 299 | } | ||
| 300 | |||
| 301 | 74 | av_log(inlink->dst, AV_LOG_DEBUG, | |
| 302 | "%s pts:%s pts_time:%s dup_count:%d\n", | ||
| 303 | 74 | decimate->dup_count == | |
| 304 |
2/2✓ Branch 1 taken 12 times.
✓ Branch 2 taken 62 times.
|
74 | decimate->min_dup_count ? "keep" : "drop", av_ts2str(cur->pts), |
| 305 | 74 | av_ts2timestr(cur->pts, &inlink->time_base), | |
| 306 | decimate->dup_count); | ||
| 307 | |||
| 308 |
2/2✓ Branch 0 taken 12 times.
✓ Branch 1 taken 62 times.
|
74 | if (decimate->dup_count == decimate->min_dup_count) { |
| 309 | 12 | out = av_frame_clone(decimate->ref); | |
| 310 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 12 times.
|
12 | if (!out) { |
| 311 | ✗ | av_frame_free(&cur); | |
| 312 | ✗ | return AVERROR(ENOMEM); | |
| 313 | } | ||
| 314 | 12 | ret = ff_filter_frame(outlink, out); | |
| 315 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 12 times.
|
12 | if (ret < 0) { |
| 316 | ✗ | av_frame_free(&cur); | |
| 317 | ✗ | return ret; | |
| 318 | } | ||
| 319 | } | ||
| 320 | 74 | av_frame_free(&cur); | |
| 321 | |||
| 322 | 74 | return 0; | |
| 323 | } | ||
| 324 | |||
| 325 | 264 | static int filter_frame(AVFilterLink *inlink, AVFrame *cur) | |
| 326 | { | ||
| 327 | 264 | DecimateContext *decimate = inlink->dst->priv; | |
| 328 |
2/2✓ Branch 0 taken 190 times.
✓ Branch 1 taken 74 times.
|
264 | return decimate->mode == 0 ? filter_frame_mode_0(inlink, cur) : filter_frame_mode_1(inlink, cur); |
| 329 | } | ||
| 330 | |||
| 331 | static const AVFilterPad mpdecimate_inputs[] = { | ||
| 332 | { | ||
| 333 | .name = "default", | ||
| 334 | .type = AVMEDIA_TYPE_VIDEO, | ||
| 335 | .config_props = config_input, | ||
| 336 | .filter_frame = filter_frame, | ||
| 337 | }, | ||
| 338 | }; | ||
| 339 | |||
| 340 | const FFFilter ff_vf_mpdecimate = { | ||
| 341 | .p.name = "mpdecimate", | ||
| 342 | .p.description = NULL_IF_CONFIG_SMALL("Remove near-duplicate frames."), | ||
| 343 | .p.priv_class = &mpdecimate_class, | ||
| 344 | .init = init, | ||
| 345 | .uninit = uninit, | ||
| 346 | .priv_size = sizeof(DecimateContext), | ||
| 347 | FILTER_INPUTS(mpdecimate_inputs), | ||
| 348 | FILTER_OUTPUTS(ff_video_default_filterpad), | ||
| 349 | FILTER_PIXFMTS_ARRAY(pix_fmts), | ||
| 350 | }; | ||
| 351 |