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 AVFilterFormats *formats = NULL;
97 int ret;
98
99 ret = ff_formats_pixdesc_filter(&formats, 0, AV_PIX_FMT_FLAG_BITSTREAM | FF_PIX_FMT_FLAG_SW_FLAT_SUB);
100 if (ret < 0)
101 return ret;
102 return ff_set_common_formats(ctx, formats);
103 }
104
uninit(AVFilterContext * ctx)105 static av_cold void uninit(AVFilterContext *ctx)
106 {
107 CropContext *s = ctx->priv;
108
109 av_expr_free(s->x_pexpr);
110 s->x_pexpr = NULL;
111 av_expr_free(s->y_pexpr);
112 s->y_pexpr = NULL;
113 }
114
normalize_double(int * n,double d)115 static inline int normalize_double(int *n, double d)
116 {
117 int ret = 0;
118
119 if (isnan(d)) {
120 ret = AVERROR(EINVAL);
121 } else if (d > INT_MAX || d < INT_MIN) {
122 *n = d > INT_MAX ? INT_MAX : INT_MIN;
123 ret = AVERROR(EINVAL);
124 } else
125 *n = lrint(d);
126
127 return ret;
128 }
129
config_input(AVFilterLink * link)130 static int config_input(AVFilterLink *link)
131 {
132 AVFilterContext *ctx = link->dst;
133 CropContext *s = ctx->priv;
134 const AVPixFmtDescriptor *pix_desc = av_pix_fmt_desc_get(link->format);
135 int ret;
136 const char *expr;
137 double res;
138
139 s->var_values[VAR_IN_W] = s->var_values[VAR_IW] = ctx->inputs[0]->w;
140 s->var_values[VAR_IN_H] = s->var_values[VAR_IH] = ctx->inputs[0]->h;
141 s->var_values[VAR_A] = (float) link->w / link->h;
142 s->var_values[VAR_SAR] = link->sample_aspect_ratio.num ? av_q2d(link->sample_aspect_ratio) : 1;
143 s->var_values[VAR_DAR] = s->var_values[VAR_A] * s->var_values[VAR_SAR];
144 s->var_values[VAR_HSUB] = 1<<pix_desc->log2_chroma_w;
145 s->var_values[VAR_VSUB] = 1<<pix_desc->log2_chroma_h;
146 s->var_values[VAR_X] = NAN;
147 s->var_values[VAR_Y] = NAN;
148 s->var_values[VAR_OUT_W] = s->var_values[VAR_OW] = NAN;
149 s->var_values[VAR_OUT_H] = s->var_values[VAR_OH] = NAN;
150 s->var_values[VAR_N] = 0;
151 s->var_values[VAR_T] = NAN;
152 s->var_values[VAR_POS] = NAN;
153
154 av_image_fill_max_pixsteps(s->max_step, NULL, pix_desc);
155
156 if (pix_desc->flags & AV_PIX_FMT_FLAG_HWACCEL) {
157 s->hsub = 1;
158 s->vsub = 1;
159 } else {
160 s->hsub = pix_desc->log2_chroma_w;
161 s->vsub = pix_desc->log2_chroma_h;
162 }
163
164 av_expr_parse_and_eval(&res, (expr = s->w_expr),
165 var_names, s->var_values,
166 NULL, NULL, NULL, NULL, NULL, 0, ctx);
167 s->var_values[VAR_OUT_W] = s->var_values[VAR_OW] = res;
168 if ((ret = av_expr_parse_and_eval(&res, (expr = s->h_expr),
169 var_names, s->var_values,
170 NULL, NULL, NULL, NULL, NULL, 0, ctx)) < 0)
171 goto fail_expr;
172 s->var_values[VAR_OUT_H] = s->var_values[VAR_OH] = res;
173 /* evaluate again ow as it may depend on oh */
174 if ((ret = av_expr_parse_and_eval(&res, (expr = s->w_expr),
175 var_names, s->var_values,
176 NULL, NULL, NULL, NULL, NULL, 0, ctx)) < 0)
177 goto fail_expr;
178
179 s->var_values[VAR_OUT_W] = s->var_values[VAR_OW] = res;
180 if (normalize_double(&s->w, s->var_values[VAR_OUT_W]) < 0 ||
181 normalize_double(&s->h, s->var_values[VAR_OUT_H]) < 0) {
182 av_log(ctx, AV_LOG_ERROR,
183 "Too big value or invalid expression for out_w/ow or out_h/oh. "
184 "Maybe the expression for out_w:'%s' or for out_h:'%s' is self-referencing.\n",
185 s->w_expr, s->h_expr);
186 return AVERROR(EINVAL);
187 }
188
189 if (!s->exact) {
190 s->w &= ~((1 << s->hsub) - 1);
191 s->h &= ~((1 << s->vsub) - 1);
192 }
193
194 av_expr_free(s->x_pexpr);
195 av_expr_free(s->y_pexpr);
196 s->x_pexpr = s->y_pexpr = NULL;
197 if ((ret = av_expr_parse(&s->x_pexpr, s->x_expr, var_names,
198 NULL, NULL, NULL, NULL, 0, ctx)) < 0 ||
199 (ret = av_expr_parse(&s->y_pexpr, s->y_expr, var_names,
200 NULL, NULL, NULL, NULL, 0, ctx)) < 0)
201 return AVERROR(EINVAL);
202
203 if (s->keep_aspect) {
204 AVRational dar = av_mul_q(link->sample_aspect_ratio,
205 (AVRational){ link->w, link->h });
206 av_reduce(&s->out_sar.num, &s->out_sar.den,
207 dar.num * s->h, dar.den * s->w, INT_MAX);
208 } else
209 s->out_sar = link->sample_aspect_ratio;
210
211 av_log(ctx, AV_LOG_VERBOSE, "w:%d h:%d sar:%d/%d -> w:%d h:%d sar:%d/%d\n",
212 link->w, link->h, link->sample_aspect_ratio.num, link->sample_aspect_ratio.den,
213 s->w, s->h, s->out_sar.num, s->out_sar.den);
214
215 if (s->w <= 0 || s->h <= 0 ||
216 s->w > link->w || s->h > link->h) {
217 av_log(ctx, AV_LOG_ERROR,
218 "Invalid too big or non positive size for width '%d' or height '%d'\n",
219 s->w, s->h);
220 return AVERROR(EINVAL);
221 }
222
223 /* set default, required in the case the first computed value for x/y is NAN */
224 s->x = (link->w - s->w) / 2;
225 s->y = (link->h - s->h) / 2;
226 if (!s->exact) {
227 s->x &= ~((1 << s->hsub) - 1);
228 s->y &= ~((1 << s->vsub) - 1);
229 }
230 return 0;
231
232 fail_expr:
233 av_log(ctx, AV_LOG_ERROR, "Error when evaluating the expression '%s'\n", expr);
234 return ret;
235 }
236
config_output(AVFilterLink * link)237 static int config_output(AVFilterLink *link)
238 {
239 CropContext *s = link->src->priv;
240 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(link->format);
241
242 if (desc->flags & AV_PIX_FMT_FLAG_HWACCEL) {
243 // Hardware frames adjust the cropping regions rather than
244 // changing the frame size.
245 } else {
246 link->w = s->w;
247 link->h = s->h;
248 }
249 link->sample_aspect_ratio = s->out_sar;
250
251 return 0;
252 }
253
filter_frame(AVFilterLink * link,AVFrame * frame)254 static int filter_frame(AVFilterLink *link, AVFrame *frame)
255 {
256 AVFilterContext *ctx = link->dst;
257 CropContext *s = ctx->priv;
258 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(link->format);
259 int i;
260
261 s->var_values[VAR_N] = link->frame_count_out;
262 s->var_values[VAR_T] = frame->pts == AV_NOPTS_VALUE ?
263 NAN : frame->pts * av_q2d(link->time_base);
264 s->var_values[VAR_POS] = frame->pkt_pos == -1 ?
265 NAN : frame->pkt_pos;
266 s->var_values[VAR_X] = av_expr_eval(s->x_pexpr, s->var_values, NULL);
267 s->var_values[VAR_Y] = av_expr_eval(s->y_pexpr, s->var_values, NULL);
268 /* It is necessary if x is expressed from y */
269 s->var_values[VAR_X] = av_expr_eval(s->x_pexpr, s->var_values, NULL);
270
271 normalize_double(&s->x, s->var_values[VAR_X]);
272 normalize_double(&s->y, s->var_values[VAR_Y]);
273
274 if (s->x < 0)
275 s->x = 0;
276 if (s->y < 0)
277 s->y = 0;
278 if ((unsigned)s->x + (unsigned)s->w > link->w)
279 s->x = link->w - s->w;
280 if ((unsigned)s->y + (unsigned)s->h > link->h)
281 s->y = link->h - s->h;
282 if (!s->exact) {
283 s->x &= ~((1 << s->hsub) - 1);
284 s->y &= ~((1 << s->vsub) - 1);
285 }
286
287 av_log(ctx, AV_LOG_TRACE, "n:%d t:%f pos:%f x:%d y:%d x+w:%d y+h:%d\n",
288 (int)s->var_values[VAR_N], s->var_values[VAR_T], s->var_values[VAR_POS],
289 s->x, s->y, s->x+s->w, s->y+s->h);
290
291 if (desc->flags & AV_PIX_FMT_FLAG_HWACCEL) {
292 frame->crop_top += s->y;
293 frame->crop_left += s->x;
294 frame->crop_bottom = frame->height - frame->crop_top - frame->crop_bottom - s->h;
295 frame->crop_right = frame->width - frame->crop_left - frame->crop_right - s->w;
296 } else {
297 frame->width = s->w;
298 frame->height = s->h;
299
300 frame->data[0] += s->y * frame->linesize[0];
301 frame->data[0] += s->x * s->max_step[0];
302
303 if (!(desc->flags & AV_PIX_FMT_FLAG_PAL || desc->flags & FF_PSEUDOPAL)) {
304 for (i = 1; i < 3; i ++) {
305 if (frame->data[i]) {
306 frame->data[i] += (s->y >> s->vsub) * frame->linesize[i];
307 frame->data[i] += (s->x * s->max_step[i]) >> s->hsub;
308 }
309 }
310 }
311
312 /* alpha plane */
313 if (frame->data[3]) {
314 frame->data[3] += s->y * frame->linesize[3];
315 frame->data[3] += s->x * s->max_step[3];
316 }
317 }
318
319 return ff_filter_frame(link->dst->outputs[0], frame);
320 }
321
process_command(AVFilterContext * ctx,const char * cmd,const char * args,char * res,int res_len,int flags)322 static int process_command(AVFilterContext *ctx, const char *cmd, const char *args,
323 char *res, int res_len, int flags)
324 {
325 CropContext *s = ctx->priv;
326 int ret;
327
328 if ( !strcmp(cmd, "out_w") || !strcmp(cmd, "w")
329 || !strcmp(cmd, "out_h") || !strcmp(cmd, "h")
330 || !strcmp(cmd, "x") || !strcmp(cmd, "y")) {
331
332 int old_x = s->x;
333 int old_y = s->y;
334 int old_w = s->w;
335 int old_h = s->h;
336
337 AVFilterLink *outlink = ctx->outputs[0];
338 AVFilterLink *inlink = ctx->inputs[0];
339
340 av_opt_set(s, cmd, args, 0);
341
342 if ((ret = config_input(inlink)) < 0) {
343 s->x = old_x;
344 s->y = old_y;
345 s->w = old_w;
346 s->h = old_h;
347 return ret;
348 }
349
350 ret = config_output(outlink);
351
352 } else
353 ret = AVERROR(ENOSYS);
354
355 return ret;
356 }
357
358 #define OFFSET(x) offsetof(CropContext, x)
359 #define FLAGS AV_OPT_FLAG_FILTERING_PARAM|AV_OPT_FLAG_VIDEO_PARAM
360 #define TFLAGS AV_OPT_FLAG_FILTERING_PARAM|AV_OPT_FLAG_VIDEO_PARAM|AV_OPT_FLAG_RUNTIME_PARAM
361
362 static const AVOption crop_options[] = {
363 { "out_w", "set the width crop area expression", OFFSET(w_expr), AV_OPT_TYPE_STRING, {.str = "iw"}, 0, 0, TFLAGS },
364 { "w", "set the width crop area expression", OFFSET(w_expr), AV_OPT_TYPE_STRING, {.str = "iw"}, 0, 0, TFLAGS },
365 { "out_h", "set the height crop area expression", OFFSET(h_expr), AV_OPT_TYPE_STRING, {.str = "ih"}, 0, 0, TFLAGS },
366 { "h", "set the height crop area expression", OFFSET(h_expr), AV_OPT_TYPE_STRING, {.str = "ih"}, 0, 0, TFLAGS },
367 { "x", "set the x crop area expression", OFFSET(x_expr), AV_OPT_TYPE_STRING, {.str = "(in_w-out_w)/2"}, 0, 0, TFLAGS },
368 { "y", "set the y crop area expression", OFFSET(y_expr), AV_OPT_TYPE_STRING, {.str = "(in_h-out_h)/2"}, 0, 0, TFLAGS },
369 { "keep_aspect", "keep aspect ratio", OFFSET(keep_aspect), AV_OPT_TYPE_BOOL, {.i64=0}, 0, 1, FLAGS },
370 { "exact", "do exact cropping", OFFSET(exact), AV_OPT_TYPE_BOOL, {.i64=0}, 0, 1, FLAGS },
371 { NULL }
372 };
373
374 AVFILTER_DEFINE_CLASS(crop);
375
376 static const AVFilterPad avfilter_vf_crop_inputs[] = {
377 {
378 .name = "default",
379 .type = AVMEDIA_TYPE_VIDEO,
380 .filter_frame = filter_frame,
381 .config_props = config_input,
382 },
383 { NULL }
384 };
385
386 static const AVFilterPad avfilter_vf_crop_outputs[] = {
387 {
388 .name = "default",
389 .type = AVMEDIA_TYPE_VIDEO,
390 .config_props = config_output,
391 },
392 { NULL }
393 };
394
395 AVFilter ff_vf_crop = {
396 .name = "crop",
397 .description = NULL_IF_CONFIG_SMALL("Crop the input video."),
398 .priv_size = sizeof(CropContext),
399 .priv_class = &crop_class,
400 .query_formats = query_formats,
401 .uninit = uninit,
402 .inputs = avfilter_vf_crop_inputs,
403 .outputs = avfilter_vf_crop_outputs,
404 .process_command = process_command,
405 };
406