1 /*
2 * Copyright (c) 2007 Bobby Bingham
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 /**
22 * @file
23 * video crop filter
24 */
25
26 #include <stdio.h>
27
28 #include "avfilter.h"
29 #include "formats.h"
30 #include "internal.h"
31 #include "video.h"
32 #include "libavutil/eval.h"
33 #include "libavutil/avstring.h"
34 #include "libavutil/internal.h"
35 #include "libavutil/libm.h"
36 #include "libavutil/imgutils.h"
37 #include "libavutil/mathematics.h"
38 #include "libavutil/opt.h"
39
40 static const char *const var_names[] = {
41 "in_w", "iw", ///< width of the input video
42 "in_h", "ih", ///< height of the input video
43 "out_w", "ow", ///< width of the cropped video
44 "out_h", "oh", ///< height of the cropped video
45 "a",
46 "sar",
47 "dar",
48 "hsub",
49 "vsub",
50 "x",
51 "y",
52 "n", ///< number of frame
53 "pos", ///< position in the file
54 "t", ///< timestamp expressed in seconds
55 NULL
56 };
57
58 enum var_name {
59 VAR_IN_W, VAR_IW,
60 VAR_IN_H, VAR_IH,
61 VAR_OUT_W, VAR_OW,
62 VAR_OUT_H, VAR_OH,
63 VAR_A,
64 VAR_SAR,
65 VAR_DAR,
66 VAR_HSUB,
67 VAR_VSUB,
68 VAR_X,
69 VAR_Y,
70 VAR_N,
71 VAR_POS,
72 VAR_T,
73 VAR_VARS_NB
74 };
75
76 typedef struct CropContext {
77 const AVClass *class;
78 int x; ///< x offset of the non-cropped area with respect to the input area
79 int y; ///< y offset of the non-cropped area with respect to the input area
80 int w; ///< width of the cropped area
81 int h; ///< height of the cropped area
82
83 AVRational out_sar; ///< output sample aspect ratio
84 int keep_aspect; ///< keep display aspect ratio when cropping
85 int exact; ///< exact cropping, for subsampled formats
86
87 int max_step[4]; ///< max pixel step for each plane, expressed as a number of bytes
88 int hsub, vsub; ///< chroma subsampling
89 char *x_expr, *y_expr, *w_expr, *h_expr;
90 AVExpr *x_pexpr, *y_pexpr; /* parsed expressions for x and y */
91 double var_values[VAR_VARS_NB];
92 } CropContext;
93
query_formats(AVFilterContext * ctx)94 static int query_formats(AVFilterContext *ctx)
95 {
96 int reject_flags = AV_PIX_FMT_FLAG_BITSTREAM | FF_PIX_FMT_FLAG_SW_FLAT_SUB;
97
98 return ff_set_common_formats(ctx, ff_formats_pixdesc_filter(0, reject_flags));
99 }
100
uninit(AVFilterContext * ctx)101 static av_cold void uninit(AVFilterContext *ctx)
102 {
103 CropContext *s = ctx->priv;
104
105 av_expr_free(s->x_pexpr);
106 s->x_pexpr = NULL;
107 av_expr_free(s->y_pexpr);
108 s->y_pexpr = NULL;
109 }
110
normalize_double(int * n,double d)111 static inline int normalize_double(int *n, double d)
112 {
113 int ret = 0;
114
115 if (isnan(d)) {
116 ret = AVERROR(EINVAL);
117 } else if (d > INT_MAX || d < INT_MIN) {
118 *n = d > INT_MAX ? INT_MAX : INT_MIN;
119 ret = AVERROR(EINVAL);
120 } else
121 *n = lrint(d);
122
123 return ret;
124 }
125
config_input(AVFilterLink * link)126 static int config_input(AVFilterLink *link)
127 {
128 AVFilterContext *ctx = link->dst;
129 CropContext *s = ctx->priv;
130 const AVPixFmtDescriptor *pix_desc = av_pix_fmt_desc_get(link->format);
131 int ret;
132 const char *expr;
133 double res;
134
135 s->var_values[VAR_IN_W] = s->var_values[VAR_IW] = ctx->inputs[0]->w;
136 s->var_values[VAR_IN_H] = s->var_values[VAR_IH] = ctx->inputs[0]->h;
137 s->var_values[VAR_A] = (float) link->w / link->h;
138 s->var_values[VAR_SAR] = link->sample_aspect_ratio.num ? av_q2d(link->sample_aspect_ratio) : 1;
139 s->var_values[VAR_DAR] = s->var_values[VAR_A] * s->var_values[VAR_SAR];
140 s->var_values[VAR_HSUB] = 1<<pix_desc->log2_chroma_w;
141 s->var_values[VAR_VSUB] = 1<<pix_desc->log2_chroma_h;
142 s->var_values[VAR_X] = NAN;
143 s->var_values[VAR_Y] = NAN;
144 s->var_values[VAR_OUT_W] = s->var_values[VAR_OW] = NAN;
145 s->var_values[VAR_OUT_H] = s->var_values[VAR_OH] = NAN;
146 s->var_values[VAR_N] = 0;
147 s->var_values[VAR_T] = NAN;
148 s->var_values[VAR_POS] = NAN;
149
150 av_image_fill_max_pixsteps(s->max_step, NULL, pix_desc);
151
152 if (pix_desc->flags & AV_PIX_FMT_FLAG_HWACCEL) {
153 s->hsub = 1;
154 s->vsub = 1;
155 } else {
156 s->hsub = pix_desc->log2_chroma_w;
157 s->vsub = pix_desc->log2_chroma_h;
158 }
159
160 av_expr_parse_and_eval(&res, (expr = s->w_expr),
161 var_names, s->var_values,
162 NULL, NULL, NULL, NULL, NULL, 0, ctx);
163 s->var_values[VAR_OUT_W] = s->var_values[VAR_OW] = res;
164 if ((ret = av_expr_parse_and_eval(&res, (expr = s->h_expr),
165 var_names, s->var_values,
166 NULL, NULL, NULL, NULL, NULL, 0, ctx)) < 0)
167 goto fail_expr;
168 s->var_values[VAR_OUT_H] = s->var_values[VAR_OH] = res;
169 /* evaluate again ow as it may depend on oh */
170 if ((ret = av_expr_parse_and_eval(&res, (expr = s->w_expr),
171 var_names, s->var_values,
172 NULL, NULL, NULL, NULL, NULL, 0, ctx)) < 0)
173 goto fail_expr;
174
175 s->var_values[VAR_OUT_W] = s->var_values[VAR_OW] = res;
176 if (normalize_double(&s->w, s->var_values[VAR_OUT_W]) < 0 ||
177 normalize_double(&s->h, s->var_values[VAR_OUT_H]) < 0) {
178 av_log(ctx, AV_LOG_ERROR,
179 "Too big value or invalid expression for out_w/ow or out_h/oh. "
180 "Maybe the expression for out_w:'%s' or for out_h:'%s' is self-referencing.\n",
181 s->w_expr, s->h_expr);
182 return AVERROR(EINVAL);
183 }
184
185 if (!s->exact) {
186 s->w &= ~((1 << s->hsub) - 1);
187 s->h &= ~((1 << s->vsub) - 1);
188 }
189
190 av_expr_free(s->x_pexpr);
191 av_expr_free(s->y_pexpr);
192 s->x_pexpr = s->y_pexpr = NULL;
193 if ((ret = av_expr_parse(&s->x_pexpr, s->x_expr, var_names,
194 NULL, NULL, NULL, NULL, 0, ctx)) < 0 ||
195 (ret = av_expr_parse(&s->y_pexpr, s->y_expr, var_names,
196 NULL, NULL, NULL, NULL, 0, ctx)) < 0)
197 return AVERROR(EINVAL);
198
199 if (s->keep_aspect) {
200 AVRational dar = av_mul_q(link->sample_aspect_ratio,
201 (AVRational){ link->w, link->h });
202 av_reduce(&s->out_sar.num, &s->out_sar.den,
203 dar.num * s->h, dar.den * s->w, INT_MAX);
204 } else
205 s->out_sar = link->sample_aspect_ratio;
206
207 av_log(ctx, AV_LOG_VERBOSE, "w:%d h:%d sar:%d/%d -> w:%d h:%d sar:%d/%d\n",
208 link->w, link->h, link->sample_aspect_ratio.num, link->sample_aspect_ratio.den,
209 s->w, s->h, s->out_sar.num, s->out_sar.den);
210
211 if (s->w <= 0 || s->h <= 0 ||
212 s->w > link->w || s->h > link->h) {
213 av_log(ctx, AV_LOG_ERROR,
214 "Invalid too big or non positive size for width '%d' or height '%d'\n",
215 s->w, s->h);
216 return AVERROR(EINVAL);
217 }
218
219 /* set default, required in the case the first computed value for x/y is NAN */
220 s->x = (link->w - s->w) / 2;
221 s->y = (link->h - s->h) / 2;
222 if (!s->exact) {
223 s->x &= ~((1 << s->hsub) - 1);
224 s->y &= ~((1 << s->vsub) - 1);
225 }
226 return 0;
227
228 fail_expr:
229 av_log(ctx, AV_LOG_ERROR, "Error when evaluating the expression '%s'\n", expr);
230 return ret;
231 }
232
config_output(AVFilterLink * link)233 static int config_output(AVFilterLink *link)
234 {
235 CropContext *s = link->src->priv;
236 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(link->format);
237
238 if (desc->flags & AV_PIX_FMT_FLAG_HWACCEL) {
239 // Hardware frames adjust the cropping regions rather than
240 // changing the frame size.
241 } else {
242 link->w = s->w;
243 link->h = s->h;
244 }
245 link->sample_aspect_ratio = s->out_sar;
246
247 return 0;
248 }
249
filter_frame(AVFilterLink * link,AVFrame * frame)250 static int filter_frame(AVFilterLink *link, AVFrame *frame)
251 {
252 AVFilterContext *ctx = link->dst;
253 CropContext *s = ctx->priv;
254 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(link->format);
255 int i;
256
257 s->var_values[VAR_N] = link->frame_count_out;
258 s->var_values[VAR_T] = frame->pts == AV_NOPTS_VALUE ?
259 NAN : frame->pts * av_q2d(link->time_base);
260 s->var_values[VAR_POS] = frame->pkt_pos == -1 ?
261 NAN : frame->pkt_pos;
262 s->var_values[VAR_X] = av_expr_eval(s->x_pexpr, s->var_values, NULL);
263 s->var_values[VAR_Y] = av_expr_eval(s->y_pexpr, s->var_values, NULL);
264 /* It is necessary if x is expressed from y */
265 s->var_values[VAR_X] = av_expr_eval(s->x_pexpr, s->var_values, NULL);
266
267 normalize_double(&s->x, s->var_values[VAR_X]);
268 normalize_double(&s->y, s->var_values[VAR_Y]);
269
270 if (s->x < 0)
271 s->x = 0;
272 if (s->y < 0)
273 s->y = 0;
274 if ((unsigned)s->x + (unsigned)s->w > link->w)
275 s->x = link->w - s->w;
276 if ((unsigned)s->y + (unsigned)s->h > link->h)
277 s->y = link->h - s->h;
278 if (!s->exact) {
279 s->x &= ~((1 << s->hsub) - 1);
280 s->y &= ~((1 << s->vsub) - 1);
281 }
282
283 av_log(ctx, AV_LOG_TRACE, "n:%d t:%f pos:%f x:%d y:%d x+w:%d y+h:%d\n",
284 (int)s->var_values[VAR_N], s->var_values[VAR_T], s->var_values[VAR_POS],
285 s->x, s->y, s->x+s->w, s->y+s->h);
286
287 if (desc->flags & AV_PIX_FMT_FLAG_HWACCEL) {
288 frame->crop_top += s->y;
289 frame->crop_left += s->x;
290 frame->crop_bottom = frame->height - frame->crop_top - frame->crop_bottom - s->h;
291 frame->crop_right = frame->width - frame->crop_left - frame->crop_right - s->w;
292 } else {
293 frame->width = s->w;
294 frame->height = s->h;
295
296 frame->data[0] += s->y * frame->linesize[0];
297 frame->data[0] += s->x * s->max_step[0];
298
299 if (!(desc->flags & AV_PIX_FMT_FLAG_PAL)) {
300 for (i = 1; i < 3; i ++) {
301 if (frame->data[i]) {
302 frame->data[i] += (s->y >> s->vsub) * frame->linesize[i];
303 frame->data[i] += (s->x * s->max_step[i]) >> s->hsub;
304 }
305 }
306 }
307
308 /* alpha plane */
309 if (frame->data[3]) {
310 frame->data[3] += s->y * frame->linesize[3];
311 frame->data[3] += s->x * s->max_step[3];
312 }
313 }
314
315 return ff_filter_frame(link->dst->outputs[0], frame);
316 }
317
process_command(AVFilterContext * ctx,const char * cmd,const char * args,char * res,int res_len,int flags)318 static int process_command(AVFilterContext *ctx, const char *cmd, const char *args,
319 char *res, int res_len, int flags)
320 {
321 CropContext *s = ctx->priv;
322 int ret;
323
324 if ( !strcmp(cmd, "out_w") || !strcmp(cmd, "w")
325 || !strcmp(cmd, "out_h") || !strcmp(cmd, "h")
326 || !strcmp(cmd, "x") || !strcmp(cmd, "y")) {
327
328 int old_x = s->x;
329 int old_y = s->y;
330 int old_w = s->w;
331 int old_h = s->h;
332
333 AVFilterLink *outlink = ctx->outputs[0];
334 AVFilterLink *inlink = ctx->inputs[0];
335
336 av_opt_set(s, cmd, args, 0);
337
338 if ((ret = config_input(inlink)) < 0) {
339 s->x = old_x;
340 s->y = old_y;
341 s->w = old_w;
342 s->h = old_h;
343 return ret;
344 }
345
346 ret = config_output(outlink);
347
348 } else
349 ret = AVERROR(ENOSYS);
350
351 return ret;
352 }
353
354 #define OFFSET(x) offsetof(CropContext, x)
355 #define FLAGS AV_OPT_FLAG_FILTERING_PARAM|AV_OPT_FLAG_VIDEO_PARAM
356 #define TFLAGS AV_OPT_FLAG_FILTERING_PARAM|AV_OPT_FLAG_VIDEO_PARAM|AV_OPT_FLAG_RUNTIME_PARAM
357
358 static const AVOption crop_options[] = {
359 { "out_w", "set the width crop area expression", OFFSET(w_expr), AV_OPT_TYPE_STRING, {.str = "iw"}, 0, 0, TFLAGS },
360 { "w", "set the width crop area expression", OFFSET(w_expr), AV_OPT_TYPE_STRING, {.str = "iw"}, 0, 0, TFLAGS },
361 { "out_h", "set the height crop area expression", OFFSET(h_expr), AV_OPT_TYPE_STRING, {.str = "ih"}, 0, 0, TFLAGS },
362 { "h", "set the height crop area expression", OFFSET(h_expr), AV_OPT_TYPE_STRING, {.str = "ih"}, 0, 0, TFLAGS },
363 { "x", "set the x crop area expression", OFFSET(x_expr), AV_OPT_TYPE_STRING, {.str = "(in_w-out_w)/2"}, 0, 0, TFLAGS },
364 { "y", "set the y crop area expression", OFFSET(y_expr), AV_OPT_TYPE_STRING, {.str = "(in_h-out_h)/2"}, 0, 0, TFLAGS },
365 { "keep_aspect", "keep aspect ratio", OFFSET(keep_aspect), AV_OPT_TYPE_BOOL, {.i64=0}, 0, 1, FLAGS },
366 { "exact", "do exact cropping", OFFSET(exact), AV_OPT_TYPE_BOOL, {.i64=0}, 0, 1, FLAGS },
367 { NULL }
368 };
369
370 AVFILTER_DEFINE_CLASS(crop);
371
372 static const AVFilterPad avfilter_vf_crop_inputs[] = {
373 {
374 .name = "default",
375 .type = AVMEDIA_TYPE_VIDEO,
376 .filter_frame = filter_frame,
377 .config_props = config_input,
378 },
379 };
380
381 static const AVFilterPad avfilter_vf_crop_outputs[] = {
382 {
383 .name = "default",
384 .type = AVMEDIA_TYPE_VIDEO,
385 .config_props = config_output,
386 },
387 };
388
389 const AVFilter ff_vf_crop = {
390 .name = "crop",
391 .description = NULL_IF_CONFIG_SMALL("Crop the input video."),
392 .priv_size = sizeof(CropContext),
393 .priv_class = &crop_class,
394 .uninit = uninit,
395 FILTER_INPUTS(avfilter_vf_crop_inputs),
396 FILTER_OUTPUTS(avfilter_vf_crop_outputs),
397 FILTER_QUERY_FUNC(query_formats),
398 .process_command = process_command,
399 };
400