1 /*
2 * Copyright (C) 2012 Michael Niedermayer <michaelni@gmx.at>
3 *
4 * This file is part of FFmpeg.
5 *
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.
10 *
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.
15 *
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
19 */
20
21 #include <float.h> /* FLT_MAX */
22
23 #include "libavutil/common.h"
24 #include "libavutil/opt.h"
25 #include "internal.h"
26 #include "vf_idet.h"
27
28 #define OFFSET(x) offsetof(IDETContext, x)
29 #define FLAGS AV_OPT_FLAG_VIDEO_PARAM|AV_OPT_FLAG_FILTERING_PARAM
30
31 static const AVOption idet_options[] = {
32 { "intl_thres", "set interlacing threshold", OFFSET(interlace_threshold), AV_OPT_TYPE_FLOAT, {.dbl = 1.04}, -1, FLT_MAX, FLAGS },
33 { "prog_thres", "set progressive threshold", OFFSET(progressive_threshold), AV_OPT_TYPE_FLOAT, {.dbl = 1.5}, -1, FLT_MAX, FLAGS },
34 { "rep_thres", "set repeat threshold", OFFSET(repeat_threshold), AV_OPT_TYPE_FLOAT, {.dbl = 3.0}, -1, FLT_MAX, FLAGS },
35 { "half_life", "half life of cumulative statistics", OFFSET(half_life), AV_OPT_TYPE_FLOAT, {.dbl = 0.0}, -1, INT_MAX, FLAGS },
36 { "analyze_interlaced_flag", "set number of frames to use to determine if the interlace flag is accurate", OFFSET(analyze_interlaced_flag), AV_OPT_TYPE_INT, {.i64 = 0 }, 0, INT_MAX, FLAGS },
37 { NULL }
38 };
39
40 AVFILTER_DEFINE_CLASS(idet);
41
type2str(Type type)42 static const char *type2str(Type type)
43 {
44 switch(type) {
45 case TFF : return "tff";
46 case BFF : return "bff";
47 case PROGRESSIVE : return "progressive";
48 case UNDETERMINED : return "undetermined";
49 }
50 return NULL;
51 }
52
53 #define PRECISION 1048576
54
uintpow(uint64_t b,unsigned int e)55 static uint64_t uintpow(uint64_t b,unsigned int e)
56 {
57 uint64_t r=1;
58 while(e--) r*=b;
59 return r;
60 }
61
av_dict_set_fxp(AVDictionary ** pm,const char * key,uint64_t value,unsigned int digits,int flags)62 static int av_dict_set_fxp(AVDictionary **pm, const char *key, uint64_t value, unsigned int digits,
63 int flags)
64 {
65 char valuestr[44];
66 uint64_t print_precision = uintpow(10, digits);
67
68 value = av_rescale(value, print_precision, PRECISION);
69
70 snprintf(valuestr, sizeof(valuestr), "%"PRId64".%0*"PRId64,
71 value / print_precision, digits, value % print_precision);
72
73 return av_dict_set(pm, key, valuestr, flags);
74 }
75
rep2str(RepeatedField repeated_field)76 static const char *rep2str(RepeatedField repeated_field)
77 {
78 switch(repeated_field) {
79 case REPEAT_NONE : return "neither";
80 case REPEAT_TOP : return "top";
81 case REPEAT_BOTTOM : return "bottom";
82 }
83 return NULL;
84 }
85
ff_idet_filter_line_c(const uint8_t * a,const uint8_t * b,const uint8_t * c,int w)86 int ff_idet_filter_line_c(const uint8_t *a, const uint8_t *b, const uint8_t *c, int w)
87 {
88 int x;
89 int ret=0;
90
91 for(x=0; x<w; x++){
92 int v = (*a++ + *c++) - 2 * *b++;
93 ret += FFABS(v);
94 }
95
96 return ret;
97 }
98
ff_idet_filter_line_c_16bit(const uint16_t * a,const uint16_t * b,const uint16_t * c,int w)99 int ff_idet_filter_line_c_16bit(const uint16_t *a, const uint16_t *b, const uint16_t *c, int w)
100 {
101 int x;
102 int ret=0;
103
104 for(x=0; x<w; x++){
105 int v = (*a++ + *c++) - 2 * *b++;
106 ret += FFABS(v);
107 }
108
109 return ret;
110 }
111
filter(AVFilterContext * ctx)112 static void filter(AVFilterContext *ctx)
113 {
114 IDETContext *idet = ctx->priv;
115 int y, i;
116 int64_t alpha[2]={0};
117 int64_t delta=0;
118 int64_t gamma[2]={0};
119 Type type, best_type;
120 RepeatedField repeat;
121 int match = 0;
122 AVDictionary **metadata = &idet->cur->metadata;
123
124 for (i = 0; i < idet->csp->nb_components; i++) {
125 int w = idet->cur->width;
126 int h = idet->cur->height;
127 int refs = idet->cur->linesize[i];
128
129 if (i && i<3) {
130 w = AV_CEIL_RSHIFT(w, idet->csp->log2_chroma_w);
131 h = AV_CEIL_RSHIFT(h, idet->csp->log2_chroma_h);
132 }
133
134 for (y = 2; y < h - 2; y++) {
135 uint8_t *prev = &idet->prev->data[i][y*refs];
136 uint8_t *cur = &idet->cur ->data[i][y*refs];
137 uint8_t *next = &idet->next->data[i][y*refs];
138 alpha[ y &1] += idet->filter_line(cur-refs, prev, cur+refs, w);
139 alpha[(y^1)&1] += idet->filter_line(cur-refs, next, cur+refs, w);
140 delta += idet->filter_line(cur-refs, cur, cur+refs, w);
141 gamma[(y^1)&1] += idet->filter_line(cur , prev, cur , w);
142 }
143 }
144
145 if (alpha[0] > idet->interlace_threshold * alpha[1]){
146 type = TFF;
147 }else if(alpha[1] > idet->interlace_threshold * alpha[0]){
148 type = BFF;
149 }else if(alpha[1] > idet->progressive_threshold * delta){
150 type = PROGRESSIVE;
151 }else{
152 type = UNDETERMINED;
153 }
154
155 if ( gamma[0] > idet->repeat_threshold * gamma[1] ){
156 repeat = REPEAT_TOP;
157 } else if ( gamma[1] > idet->repeat_threshold * gamma[0] ){
158 repeat = REPEAT_BOTTOM;
159 } else {
160 repeat = REPEAT_NONE;
161 }
162
163 memmove(idet->history+1, idet->history, HIST_SIZE-1);
164 idet->history[0] = type;
165 best_type = UNDETERMINED;
166 for(i=0; i<HIST_SIZE; i++){
167 if(idet->history[i] != UNDETERMINED){
168 if(best_type == UNDETERMINED)
169 best_type = idet->history[i];
170
171 if(idet->history[i] == best_type) {
172 match++;
173 }else{
174 match=0;
175 break;
176 }
177 }
178 }
179 if(idet->last_type == UNDETERMINED){
180 if(match ) idet->last_type = best_type;
181 }else{
182 if(match>2) idet->last_type = best_type;
183 }
184
185 if (idet->last_type == TFF){
186 idet->cur->top_field_first = 1;
187 idet->cur->interlaced_frame = 1;
188 }else if(idet->last_type == BFF){
189 idet->cur->top_field_first = 0;
190 idet->cur->interlaced_frame = 1;
191 }else if(idet->last_type == PROGRESSIVE){
192 idet->cur->interlaced_frame = 0;
193 }
194
195 for(i=0; i<3; i++)
196 idet->repeats[i] = av_rescale(idet->repeats [i], idet->decay_coefficient, PRECISION);
197
198 for(i=0; i<4; i++){
199 idet->prestat [i] = av_rescale(idet->prestat [i], idet->decay_coefficient, PRECISION);
200 idet->poststat[i] = av_rescale(idet->poststat[i], idet->decay_coefficient, PRECISION);
201 }
202
203 idet->total_repeats [ repeat] ++;
204 idet->repeats [ repeat] += PRECISION;
205
206 idet->total_prestat [ type] ++;
207 idet->prestat [ type] += PRECISION;
208
209 idet->total_poststat[idet->last_type] ++;
210 idet->poststat [idet->last_type] += PRECISION;
211
212 av_log(ctx, AV_LOG_DEBUG, "Repeated Field:%12s, Single frame:%12s, Multi frame:%12s\n",
213 rep2str(repeat), type2str(type), type2str(idet->last_type));
214
215 av_dict_set (metadata, "lavfi.idet.repeated.current_frame", rep2str(repeat), 0);
216 av_dict_set_fxp(metadata, "lavfi.idet.repeated.neither", idet->repeats[REPEAT_NONE], 2, 0);
217 av_dict_set_fxp(metadata, "lavfi.idet.repeated.top", idet->repeats[REPEAT_TOP], 2, 0);
218 av_dict_set_fxp(metadata, "lavfi.idet.repeated.bottom", idet->repeats[REPEAT_BOTTOM], 2, 0);
219
220 av_dict_set (metadata, "lavfi.idet.single.current_frame", type2str(type), 0);
221 av_dict_set_fxp(metadata, "lavfi.idet.single.tff", idet->prestat[TFF], 2 , 0);
222 av_dict_set_fxp(metadata, "lavfi.idet.single.bff", idet->prestat[BFF], 2, 0);
223 av_dict_set_fxp(metadata, "lavfi.idet.single.progressive", idet->prestat[PROGRESSIVE], 2, 0);
224 av_dict_set_fxp(metadata, "lavfi.idet.single.undetermined", idet->prestat[UNDETERMINED], 2, 0);
225
226 av_dict_set (metadata, "lavfi.idet.multiple.current_frame", type2str(idet->last_type), 0);
227 av_dict_set_fxp(metadata, "lavfi.idet.multiple.tff", idet->poststat[TFF], 2, 0);
228 av_dict_set_fxp(metadata, "lavfi.idet.multiple.bff", idet->poststat[BFF], 2, 0);
229 av_dict_set_fxp(metadata, "lavfi.idet.multiple.progressive", idet->poststat[PROGRESSIVE], 2, 0);
230 av_dict_set_fxp(metadata, "lavfi.idet.multiple.undetermined", idet->poststat[UNDETERMINED], 2, 0);
231 }
232
filter_frame(AVFilterLink * link,AVFrame * picref)233 static int filter_frame(AVFilterLink *link, AVFrame *picref)
234 {
235 AVFilterContext *ctx = link->dst;
236 IDETContext *idet = ctx->priv;
237
238 // initial frame(s) and not interlaced, just pass through for
239 // the analyze_interlaced_flag mode
240 if (idet->analyze_interlaced_flag &&
241 !picref->interlaced_frame &&
242 !idet->next) {
243 return ff_filter_frame(ctx->outputs[0], picref);
244 }
245 if (idet->analyze_interlaced_flag_done) {
246 if (picref->interlaced_frame && idet->interlaced_flag_accuracy < 0)
247 picref->interlaced_frame = 0;
248 return ff_filter_frame(ctx->outputs[0], picref);
249 }
250
251 av_frame_free(&idet->prev);
252
253 if( picref->width != link->w
254 || picref->height != link->h
255 || picref->format != link->format) {
256 link->dst->inputs[0]->format = picref->format;
257 link->dst->inputs[0]->w = picref->width;
258 link->dst->inputs[0]->h = picref->height;
259
260 av_frame_free(&idet->cur );
261 av_frame_free(&idet->next);
262 }
263
264 idet->prev = idet->cur;
265 idet->cur = idet->next;
266 idet->next = picref;
267
268 if (!idet->cur &&
269 !(idet->cur = av_frame_clone(idet->next)))
270 return AVERROR(ENOMEM);
271
272 if (!idet->prev)
273 return 0;
274
275 if (!idet->csp)
276 idet->csp = av_pix_fmt_desc_get(link->format);
277 if (idet->csp->comp[0].depth > 8){
278 idet->filter_line = (ff_idet_filter_func)ff_idet_filter_line_c_16bit;
279 #if ARCH_X86
280 ff_idet_init_x86(idet, 1);
281 #endif
282 }
283
284 if (idet->analyze_interlaced_flag) {
285 if (idet->cur->interlaced_frame) {
286 idet->cur->interlaced_frame = 0;
287 filter(ctx);
288 if (idet->last_type == PROGRESSIVE) {
289 idet->interlaced_flag_accuracy --;
290 idet->analyze_interlaced_flag --;
291 } else if (idet->last_type != UNDETERMINED) {
292 idet->interlaced_flag_accuracy ++;
293 idet->analyze_interlaced_flag --;
294 }
295 if (idet->analyze_interlaced_flag == 1) {
296 ff_filter_frame(ctx->outputs[0], av_frame_clone(idet->cur));
297
298 if (idet->next->interlaced_frame && idet->interlaced_flag_accuracy < 0)
299 idet->next->interlaced_frame = 0;
300 idet->analyze_interlaced_flag_done = 1;
301 av_log(ctx, AV_LOG_INFO, "Final flag accuracy %d\n", idet->interlaced_flag_accuracy);
302 return ff_filter_frame(ctx->outputs[0], av_frame_clone(idet->next));
303 }
304 }
305 } else {
306 filter(ctx);
307 }
308
309 return ff_filter_frame(ctx->outputs[0], av_frame_clone(idet->cur));
310 }
311
request_frame(AVFilterLink * link)312 static int request_frame(AVFilterLink *link)
313 {
314 AVFilterContext *ctx = link->src;
315 IDETContext *idet = ctx->priv;
316 int ret;
317
318 if (idet->eof)
319 return AVERROR_EOF;
320
321 ret = ff_request_frame(link->src->inputs[0]);
322
323 if (ret == AVERROR_EOF && idet->cur && !idet->analyze_interlaced_flag_done) {
324 AVFrame *next = av_frame_clone(idet->next);
325
326 if (!next)
327 return AVERROR(ENOMEM);
328
329 ret = filter_frame(link->src->inputs[0], next);
330 idet->eof = 1;
331 }
332
333 return ret;
334 }
335
uninit(AVFilterContext * ctx)336 static av_cold void uninit(AVFilterContext *ctx)
337 {
338 IDETContext *idet = ctx->priv;
339
340 av_log(ctx, AV_LOG_INFO, "Repeated Fields: Neither:%6"PRId64" Top:%6"PRId64" Bottom:%6"PRId64"\n",
341 idet->total_repeats[REPEAT_NONE],
342 idet->total_repeats[REPEAT_TOP],
343 idet->total_repeats[REPEAT_BOTTOM]
344 );
345 av_log(ctx, AV_LOG_INFO, "Single frame detection: TFF:%6"PRId64" BFF:%6"PRId64" Progressive:%6"PRId64" Undetermined:%6"PRId64"\n",
346 idet->total_prestat[TFF],
347 idet->total_prestat[BFF],
348 idet->total_prestat[PROGRESSIVE],
349 idet->total_prestat[UNDETERMINED]
350 );
351 av_log(ctx, AV_LOG_INFO, "Multi frame detection: TFF:%6"PRId64" BFF:%6"PRId64" Progressive:%6"PRId64" Undetermined:%6"PRId64"\n",
352 idet->total_poststat[TFF],
353 idet->total_poststat[BFF],
354 idet->total_poststat[PROGRESSIVE],
355 idet->total_poststat[UNDETERMINED]
356 );
357
358 av_frame_free(&idet->prev);
359 av_frame_free(&idet->cur );
360 av_frame_free(&idet->next);
361 }
362
363 static const enum AVPixelFormat pix_fmts[] = {
364 AV_PIX_FMT_YUV420P,
365 AV_PIX_FMT_YUV422P,
366 AV_PIX_FMT_YUV444P,
367 AV_PIX_FMT_YUV410P,
368 AV_PIX_FMT_YUV411P,
369 AV_PIX_FMT_GRAY8,
370 AV_PIX_FMT_YUVJ420P,
371 AV_PIX_FMT_YUVJ422P,
372 AV_PIX_FMT_YUVJ444P,
373 AV_PIX_FMT_GRAY16,
374 AV_PIX_FMT_YUV440P,
375 AV_PIX_FMT_YUVJ440P,
376 AV_PIX_FMT_YUV420P9,
377 AV_PIX_FMT_YUV422P9,
378 AV_PIX_FMT_YUV444P9,
379 AV_PIX_FMT_YUV420P10,
380 AV_PIX_FMT_YUV422P10,
381 AV_PIX_FMT_YUV444P10,
382 AV_PIX_FMT_YUV420P12,
383 AV_PIX_FMT_YUV422P12,
384 AV_PIX_FMT_YUV444P12,
385 AV_PIX_FMT_YUV420P14,
386 AV_PIX_FMT_YUV422P14,
387 AV_PIX_FMT_YUV444P14,
388 AV_PIX_FMT_YUV420P16,
389 AV_PIX_FMT_YUV422P16,
390 AV_PIX_FMT_YUV444P16,
391 AV_PIX_FMT_YUVA420P,
392 AV_PIX_FMT_YUVA422P,
393 AV_PIX_FMT_YUVA444P,
394 AV_PIX_FMT_NONE
395 };
396
init(AVFilterContext * ctx)397 static av_cold int init(AVFilterContext *ctx)
398 {
399 IDETContext *idet = ctx->priv;
400
401 idet->eof = 0;
402 idet->last_type = UNDETERMINED;
403 memset(idet->history, UNDETERMINED, HIST_SIZE);
404
405 if( idet->half_life > 0 )
406 idet->decay_coefficient = lrint( PRECISION * exp2(-1.0 / idet->half_life) );
407 else
408 idet->decay_coefficient = PRECISION;
409
410 idet->filter_line = ff_idet_filter_line_c;
411
412 #if ARCH_X86
413 ff_idet_init_x86(idet, 0);
414 #endif
415
416 return 0;
417 }
418
419 static const AVFilterPad idet_inputs[] = {
420 {
421 .name = "default",
422 .type = AVMEDIA_TYPE_VIDEO,
423 .filter_frame = filter_frame,
424 },
425 };
426
427 static const AVFilterPad idet_outputs[] = {
428 {
429 .name = "default",
430 .type = AVMEDIA_TYPE_VIDEO,
431 .request_frame = request_frame
432 },
433 };
434
435 const AVFilter ff_vf_idet = {
436 .name = "idet",
437 .description = NULL_IF_CONFIG_SMALL("Interlace detect Filter."),
438 .priv_size = sizeof(IDETContext),
439 .init = init,
440 .uninit = uninit,
441 .flags = AVFILTER_FLAG_METADATA_ONLY,
442 FILTER_INPUTS(idet_inputs),
443 FILTER_OUTPUTS(idet_outputs),
444 FILTER_PIXFMTS_ARRAY(pix_fmts),
445 .priv_class = &idet_class,
446 };
447