summaryrefslogtreecommitdiff
path: root/libavfilter/vf_perspective.c (plain)
blob: 92495097cc5515994a344be2a0558073c5293227
1/*
2 * Copyright (c) 2002 Michael Niedermayer <michaelni@gmx.at>
3 * Copyright (c) 2013 Paul B Mahol
4 *
5 * This file is part of FFmpeg.
6 *
7 * FFmpeg is free software; you can redistribute it and/or
8 * modify it under the terms of the GNU General Public
9 * License as published by the Free Software Foundation; either
10 * version 2 of the License, or (at your option) any later version.
11 *
12 * FFmpeg is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 * GNU General Public License for more details.
16 *
17 * You should have received a copy of the GNU General Public License along
18 * with FFmpeg; if not, write to the Free Software
19 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
20 */
21
22#include "libavutil/avassert.h"
23#include "libavutil/eval.h"
24#include "libavutil/imgutils.h"
25#include "libavutil/pixdesc.h"
26#include "libavutil/opt.h"
27#include "avfilter.h"
28#include "formats.h"
29#include "internal.h"
30#include "video.h"
31
32#define SUB_PIXEL_BITS 8
33#define SUB_PIXELS (1 << SUB_PIXEL_BITS)
34#define COEFF_BITS 11
35
36#define LINEAR 0
37#define CUBIC 1
38
39typedef struct PerspectiveContext {
40 const AVClass *class;
41 char *expr_str[4][2];
42 double ref[4][2];
43 int32_t (*pv)[2];
44 int32_t coeff[SUB_PIXELS][4];
45 int interpolation;
46 int linesize[4];
47 int height[4];
48 int hsub, vsub;
49 int nb_planes;
50 int sense;
51 int eval_mode;
52
53 int (*perspective)(AVFilterContext *ctx,
54 void *arg, int job, int nb_jobs);
55} PerspectiveContext;
56
57#define OFFSET(x) offsetof(PerspectiveContext, x)
58#define FLAGS AV_OPT_FLAG_FILTERING_PARAM|AV_OPT_FLAG_VIDEO_PARAM
59
60enum PERSPECTIVESense {
61 PERSPECTIVE_SENSE_SOURCE = 0, ///< coordinates give locations in source of corners of destination.
62 PERSPECTIVE_SENSE_DESTINATION = 1, ///< coordinates give locations in destination of corners of source.
63};
64
65enum EvalMode {
66 EVAL_MODE_INIT,
67 EVAL_MODE_FRAME,
68 EVAL_MODE_NB
69};
70
71static const AVOption perspective_options[] = {
72 { "x0", "set top left x coordinate", OFFSET(expr_str[0][0]), AV_OPT_TYPE_STRING, {.str="0"}, 0, 0, FLAGS },
73 { "y0", "set top left y coordinate", OFFSET(expr_str[0][1]), AV_OPT_TYPE_STRING, {.str="0"}, 0, 0, FLAGS },
74 { "x1", "set top right x coordinate", OFFSET(expr_str[1][0]), AV_OPT_TYPE_STRING, {.str="W"}, 0, 0, FLAGS },
75 { "y1", "set top right y coordinate", OFFSET(expr_str[1][1]), AV_OPT_TYPE_STRING, {.str="0"}, 0, 0, FLAGS },
76 { "x2", "set bottom left x coordinate", OFFSET(expr_str[2][0]), AV_OPT_TYPE_STRING, {.str="0"}, 0, 0, FLAGS },
77 { "y2", "set bottom left y coordinate", OFFSET(expr_str[2][1]), AV_OPT_TYPE_STRING, {.str="H"}, 0, 0, FLAGS },
78 { "x3", "set bottom right x coordinate", OFFSET(expr_str[3][0]), AV_OPT_TYPE_STRING, {.str="W"}, 0, 0, FLAGS },
79 { "y3", "set bottom right y coordinate", OFFSET(expr_str[3][1]), AV_OPT_TYPE_STRING, {.str="H"}, 0, 0, FLAGS },
80 { "interpolation", "set interpolation", OFFSET(interpolation), AV_OPT_TYPE_INT, {.i64=LINEAR}, 0, 1, FLAGS, "interpolation" },
81 { "linear", "", 0, AV_OPT_TYPE_CONST, {.i64=LINEAR}, 0, 0, FLAGS, "interpolation" },
82 { "cubic", "", 0, AV_OPT_TYPE_CONST, {.i64=CUBIC}, 0, 0, FLAGS, "interpolation" },
83 { "sense", "specify the sense of the coordinates", OFFSET(sense), AV_OPT_TYPE_INT, {.i64=PERSPECTIVE_SENSE_SOURCE}, 0, 1, FLAGS, "sense"},
84 { "source", "specify locations in source to send to corners in destination",
85 0, AV_OPT_TYPE_CONST, {.i64=PERSPECTIVE_SENSE_SOURCE}, 0, 0, FLAGS, "sense"},
86 { "destination", "specify locations in destination to send corners of source",
87 0, AV_OPT_TYPE_CONST, {.i64=PERSPECTIVE_SENSE_DESTINATION}, 0, 0, FLAGS, "sense"},
88 { "eval", "specify when to evaluate expressions", OFFSET(eval_mode), AV_OPT_TYPE_INT, {.i64 = EVAL_MODE_INIT}, 0, EVAL_MODE_NB-1, FLAGS, "eval" },
89 { "init", "eval expressions once during initialization", 0, AV_OPT_TYPE_CONST, {.i64=EVAL_MODE_INIT}, .flags = FLAGS, .unit = "eval" },
90 { "frame", "eval expressions per-frame", 0, AV_OPT_TYPE_CONST, {.i64=EVAL_MODE_FRAME}, .flags = FLAGS, .unit = "eval" },
91
92 { NULL }
93};
94
95AVFILTER_DEFINE_CLASS(perspective);
96
97static int query_formats(AVFilterContext *ctx)
98{
99 static const enum AVPixelFormat pix_fmts[] = {
100 AV_PIX_FMT_YUVA444P, AV_PIX_FMT_YUVA422P, AV_PIX_FMT_YUVA420P,
101 AV_PIX_FMT_YUVJ444P, AV_PIX_FMT_YUVJ440P, AV_PIX_FMT_YUVJ422P,AV_PIX_FMT_YUVJ420P, AV_PIX_FMT_YUVJ411P,
102 AV_PIX_FMT_YUV444P, AV_PIX_FMT_YUV440P, AV_PIX_FMT_YUV422P, AV_PIX_FMT_YUV420P, AV_PIX_FMT_YUV411P, AV_PIX_FMT_YUV410P,
103 AV_PIX_FMT_GBRP, AV_PIX_FMT_GBRAP, AV_PIX_FMT_GRAY8, AV_PIX_FMT_NONE
104 };
105
106 AVFilterFormats *fmts_list = ff_make_format_list(pix_fmts);
107 if (!fmts_list)
108 return AVERROR(ENOMEM);
109 return ff_set_common_formats(ctx, fmts_list);
110}
111
112static inline double get_coeff(double d)
113{
114 double coeff, A = -0.60;
115
116 d = fabs(d);
117
118 if (d < 1.0)
119 coeff = (1.0 - (A + 3.0) * d * d + (A + 2.0) * d * d * d);
120 else if (d < 2.0)
121 coeff = (-4.0 * A + 8.0 * A * d - 5.0 * A * d * d + A * d * d * d);
122 else
123 coeff = 0.0;
124
125 return coeff;
126}
127
128static const char *const var_names[] = { "W", "H", "in", "on", NULL };
129enum { VAR_W, VAR_H, VAR_IN, VAR_ON, VAR_VARS_NB };
130
131static int calc_persp_luts(AVFilterContext *ctx, AVFilterLink *inlink)
132{
133 PerspectiveContext *s = ctx->priv;
134 AVFilterLink *outlink = ctx->outputs[0];
135 double (*ref)[2] = s->ref;
136
137 double values[VAR_VARS_NB] = { [VAR_W] = inlink->w, [VAR_H] = inlink->h,
138 [VAR_IN] = inlink->frame_count_out + 1,
139 [VAR_ON] = outlink->frame_count_in + 1 };
140 const int h = values[VAR_H];
141 const int w = values[VAR_W];
142 double x0, x1, x2, x3, x4, x5, x6, x7, x8, q;
143 double t0, t1, t2, t3;
144 int x, y, i, j, ret;
145
146 for (i = 0; i < 4; i++) {
147 for (j = 0; j < 2; j++) {
148 if (!s->expr_str[i][j])
149 return AVERROR(EINVAL);
150 ret = av_expr_parse_and_eval(&s->ref[i][j], s->expr_str[i][j],
151 var_names, &values[0],
152 NULL, NULL, NULL, NULL,
153 0, 0, ctx);
154 if (ret < 0)
155 return ret;
156 }
157 }
158
159 switch (s->sense) {
160 case PERSPECTIVE_SENSE_SOURCE:
161 x6 = ((ref[0][0] - ref[1][0] - ref[2][0] + ref[3][0]) *
162 (ref[2][1] - ref[3][1]) -
163 ( ref[0][1] - ref[1][1] - ref[2][1] + ref[3][1]) *
164 (ref[2][0] - ref[3][0])) * h;
165 x7 = ((ref[0][1] - ref[1][1] - ref[2][1] + ref[3][1]) *
166 (ref[1][0] - ref[3][0]) -
167 ( ref[0][0] - ref[1][0] - ref[2][0] + ref[3][0]) *
168 (ref[1][1] - ref[3][1])) * w;
169 q = ( ref[1][0] - ref[3][0]) * (ref[2][1] - ref[3][1]) -
170 ( ref[2][0] - ref[3][0]) * (ref[1][1] - ref[3][1]);
171
172 x0 = q * (ref[1][0] - ref[0][0]) * h + x6 * ref[1][0];
173 x1 = q * (ref[2][0] - ref[0][0]) * w + x7 * ref[2][0];
174 x2 = q * ref[0][0] * w * h;
175 x3 = q * (ref[1][1] - ref[0][1]) * h + x6 * ref[1][1];
176 x4 = q * (ref[2][1] - ref[0][1]) * w + x7 * ref[2][1];
177 x5 = q * ref[0][1] * w * h;
178 x8 = q * w * h;
179 break;
180 case PERSPECTIVE_SENSE_DESTINATION:
181 t0 = ref[0][0] * (ref[3][1] - ref[1][1]) +
182 ref[1][0] * (ref[0][1] - ref[3][1]) +
183 ref[3][0] * (ref[1][1] - ref[0][1]);
184 t1 = ref[1][0] * (ref[2][1] - ref[3][1]) +
185 ref[2][0] * (ref[3][1] - ref[1][1]) +
186 ref[3][0] * (ref[1][1] - ref[2][1]);
187 t2 = ref[0][0] * (ref[3][1] - ref[2][1]) +
188 ref[2][0] * (ref[0][1] - ref[3][1]) +
189 ref[3][0] * (ref[2][1] - ref[0][1]);
190 t3 = ref[0][0] * (ref[1][1] - ref[2][1]) +
191 ref[1][0] * (ref[2][1] - ref[0][1]) +
192 ref[2][0] * (ref[0][1] - ref[1][1]);
193
194 x0 = t0 * t1 * w * (ref[2][1] - ref[0][1]);
195 x1 = t0 * t1 * w * (ref[0][0] - ref[2][0]);
196 x2 = t0 * t1 * w * (ref[0][1] * ref[2][0] - ref[0][0] * ref[2][1]);
197 x3 = t1 * t2 * h * (ref[1][1] - ref[0][1]);
198 x4 = t1 * t2 * h * (ref[0][0] - ref[1][0]);
199 x5 = t1 * t2 * h * (ref[0][1] * ref[1][0] - ref[0][0] * ref[1][1]);
200 x6 = t1 * t2 * (ref[1][1] - ref[0][1]) +
201 t0 * t3 * (ref[2][1] - ref[3][1]);
202 x7 = t1 * t2 * (ref[0][0] - ref[1][0]) +
203 t0 * t3 * (ref[3][0] - ref[2][0]);
204 x8 = t1 * t2 * (ref[0][1] * ref[1][0] - ref[0][0] * ref[1][1]) +
205 t0 * t3 * (ref[2][0] * ref[3][1] - ref[2][1] * ref[3][0]);
206 break;
207 default:
208 av_assert0(0);
209 }
210
211 for (y = 0; y < h; y++){
212 for (x = 0; x < w; x++){
213 int u, v;
214
215 u = lrint(SUB_PIXELS * (x0 * x + x1 * y + x2) /
216 (x6 * x + x7 * y + x8));
217 v = lrint(SUB_PIXELS * (x3 * x + x4 * y + x5) /
218 (x6 * x + x7 * y + x8));
219
220 s->pv[x + y * w][0] = u;
221 s->pv[x + y * w][1] = v;
222 }
223 }
224
225 return 0;
226}
227
228static int config_input(AVFilterLink *inlink)
229{
230 AVFilterContext *ctx = inlink->dst;
231 PerspectiveContext *s = ctx->priv;
232 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(inlink->format);
233 int h = inlink->h;
234 int w = inlink->w;
235 int i, j, ret;
236 s->hsub = desc->log2_chroma_w;
237 s->vsub = desc->log2_chroma_h;
238 s->nb_planes = av_pix_fmt_count_planes(inlink->format);
239 if ((ret = av_image_fill_linesizes(s->linesize, inlink->format, inlink->w)) < 0)
240 return ret;
241
242 s->height[1] = s->height[2] = AV_CEIL_RSHIFT(inlink->h, desc->log2_chroma_h);
243 s->height[0] = s->height[3] = inlink->h;
244
245 s->pv = av_realloc_f(s->pv, w * h, 2 * sizeof(*s->pv));
246 if (!s->pv)
247 return AVERROR(ENOMEM);
248
249 if (s->eval_mode == EVAL_MODE_INIT) {
250 if ((ret = calc_persp_luts(ctx, inlink)) < 0) {
251 return ret;
252 }
253 }
254
255 for (i = 0; i < SUB_PIXELS; i++){
256 double d = i / (double)SUB_PIXELS;
257 double temp[4];
258 double sum = 0;
259
260 for (j = 0; j < 4; j++)
261 temp[j] = get_coeff(j - d - 1);
262
263 for (j = 0; j < 4; j++)
264 sum += temp[j];
265
266 for (j = 0; j < 4; j++)
267 s->coeff[i][j] = lrint((1 << COEFF_BITS) * temp[j] / sum);
268 }
269
270 return 0;
271}
272
273typedef struct ThreadData {
274 uint8_t *dst;
275 int dst_linesize;
276 uint8_t *src;
277 int src_linesize;
278 int w, h;
279 int hsub, vsub;
280} ThreadData;
281
282static int resample_cubic(AVFilterContext *ctx, void *arg,
283 int job, int nb_jobs)
284{
285 PerspectiveContext *s = ctx->priv;
286 ThreadData *td = arg;
287 uint8_t *dst = td->dst;
288 int dst_linesize = td->dst_linesize;
289 uint8_t *src = td->src;
290 int src_linesize = td->src_linesize;
291 int w = td->w;
292 int h = td->h;
293 int hsub = td->hsub;
294 int vsub = td->vsub;
295 int start = (h * job) / nb_jobs;
296 int end = (h * (job+1)) / nb_jobs;
297 const int linesize = s->linesize[0];
298 int x, y;
299
300 for (y = start; y < end; y++) {
301 int sy = y << vsub;
302 for (x = 0; x < w; x++) {
303 int u, v, subU, subV, sum, sx;
304
305 sx = x << hsub;
306 u = s->pv[sx + sy * linesize][0] >> hsub;
307 v = s->pv[sx + sy * linesize][1] >> vsub;
308 subU = u & (SUB_PIXELS - 1);
309 subV = v & (SUB_PIXELS - 1);
310 u >>= SUB_PIXEL_BITS;
311 v >>= SUB_PIXEL_BITS;
312
313 if (u > 0 && v > 0 && u < w - 2 && v < h - 2){
314 const int index = u + v*src_linesize;
315 const int a = s->coeff[subU][0];
316 const int b = s->coeff[subU][1];
317 const int c = s->coeff[subU][2];
318 const int d = s->coeff[subU][3];
319
320 sum = s->coeff[subV][0] * (a * src[index - 1 - src_linesize] + b * src[index - 0 - src_linesize] +
321 c * src[index + 1 - src_linesize] + d * src[index + 2 - src_linesize]) +
322 s->coeff[subV][1] * (a * src[index - 1 ] + b * src[index - 0 ] +
323 c * src[index + 1 ] + d * src[index + 2 ]) +
324 s->coeff[subV][2] * (a * src[index - 1 + src_linesize] + b * src[index - 0 + src_linesize] +
325 c * src[index + 1 + src_linesize] + d * src[index + 2 + src_linesize]) +
326 s->coeff[subV][3] * (a * src[index - 1 + 2 * src_linesize] + b * src[index - 0 + 2 * src_linesize] +
327 c * src[index + 1 + 2 * src_linesize] + d * src[index + 2 + 2 * src_linesize]);
328 } else {
329 int dx, dy;
330
331 sum = 0;
332
333 for (dy = 0; dy < 4; dy++) {
334 int iy = v + dy - 1;
335
336 if (iy < 0)
337 iy = 0;
338 else if (iy >= h)
339 iy = h-1;
340 for (dx = 0; dx < 4; dx++) {
341 int ix = u + dx - 1;
342
343 if (ix < 0)
344 ix = 0;
345 else if (ix >= w)
346 ix = w - 1;
347
348 sum += s->coeff[subU][dx] * s->coeff[subV][dy] * src[ ix + iy * src_linesize];
349 }
350 }
351 }
352
353 sum = (sum + (1<<(COEFF_BITS * 2 - 1))) >> (COEFF_BITS * 2);
354 sum = av_clip_uint8(sum);
355 dst[x + y * dst_linesize] = sum;
356 }
357 }
358 return 0;
359}
360
361static int resample_linear(AVFilterContext *ctx, void *arg,
362 int job, int nb_jobs)
363{
364 PerspectiveContext *s = ctx->priv;
365 ThreadData *td = arg;
366 uint8_t *dst = td->dst;
367 int dst_linesize = td->dst_linesize;
368 uint8_t *src = td->src;
369 int src_linesize = td->src_linesize;
370 int w = td->w;
371 int h = td->h;
372 int hsub = td->hsub;
373 int vsub = td->vsub;
374 int start = (h * job) / nb_jobs;
375 int end = (h * (job+1)) / nb_jobs;
376 const int linesize = s->linesize[0];
377 int x, y;
378
379 for (y = start; y < end; y++){
380 int sy = y << vsub;
381 for (x = 0; x < w; x++){
382 int u, v, subU, subV, sum, sx, index, subUI, subVI;
383
384 sx = x << hsub;
385 u = s->pv[sx + sy * linesize][0] >> hsub;
386 v = s->pv[sx + sy * linesize][1] >> vsub;
387 subU = u & (SUB_PIXELS - 1);
388 subV = v & (SUB_PIXELS - 1);
389 u >>= SUB_PIXEL_BITS;
390 v >>= SUB_PIXEL_BITS;
391
392 index = u + v * src_linesize;
393 subUI = SUB_PIXELS - subU;
394 subVI = SUB_PIXELS - subV;
395
396 if ((unsigned)u < (unsigned)(w - 1)){
397 if((unsigned)v < (unsigned)(h - 1)){
398 sum = subVI * (subUI * src[index] + subU * src[index + 1]) +
399 subV * (subUI * src[index + src_linesize] + subU * src[index + src_linesize + 1]);
400 sum = (sum + (1 << (SUB_PIXEL_BITS * 2 - 1)))>> (SUB_PIXEL_BITS * 2);
401 } else {
402 if (v < 0)
403 v = 0;
404 else
405 v = h - 1;
406 index = u + v * src_linesize;
407 sum = subUI * src[index] + subU * src[index + 1];
408 sum = (sum + (1 << (SUB_PIXEL_BITS - 1))) >> SUB_PIXEL_BITS;
409 }
410 } else {
411 if (u < 0)
412 u = 0;
413 else
414 u = w - 1;
415 if ((unsigned)v < (unsigned)(h - 1)){
416 index = u + v * src_linesize;
417 sum = subVI * src[index] + subV * src[index + src_linesize];
418 sum = (sum + (1 << (SUB_PIXEL_BITS - 1))) >> SUB_PIXEL_BITS;
419 } else {
420 if (v < 0)
421 v = 0;
422 else
423 v = h - 1;
424 index = u + v * src_linesize;
425 sum = src[index];
426 }
427 }
428
429 sum = av_clip_uint8(sum);
430 dst[x + y * dst_linesize] = sum;
431 }
432 }
433 return 0;
434}
435
436static av_cold int init(AVFilterContext *ctx)
437{
438 PerspectiveContext *s = ctx->priv;
439
440 switch (s->interpolation) {
441 case LINEAR: s->perspective = resample_linear; break;
442 case CUBIC: s->perspective = resample_cubic; break;
443 }
444
445 return 0;
446}
447
448static int filter_frame(AVFilterLink *inlink, AVFrame *frame)
449{
450 AVFilterContext *ctx = inlink->dst;
451 AVFilterLink *outlink = ctx->outputs[0];
452 PerspectiveContext *s = ctx->priv;
453 AVFrame *out;
454 int plane;
455 int ret;
456
457 out = ff_get_video_buffer(outlink, outlink->w, outlink->h);
458 if (!out) {
459 av_frame_free(&frame);
460 return AVERROR(ENOMEM);
461 }
462 av_frame_copy_props(out, frame);
463
464 if (s->eval_mode == EVAL_MODE_FRAME) {
465 if ((ret = calc_persp_luts(ctx, inlink)) < 0) {
466 av_frame_free(&out);
467 return ret;
468 }
469 }
470
471 for (plane = 0; plane < s->nb_planes; plane++) {
472 int hsub = plane == 1 || plane == 2 ? s->hsub : 0;
473 int vsub = plane == 1 || plane == 2 ? s->vsub : 0;
474 ThreadData td = {.dst = out->data[plane],
475 .dst_linesize = out->linesize[plane],
476 .src = frame->data[plane],
477 .src_linesize = frame->linesize[plane],
478 .w = s->linesize[plane],
479 .h = s->height[plane],
480 .hsub = hsub,
481 .vsub = vsub };
482 ctx->internal->execute(ctx, s->perspective, &td, NULL, FFMIN(td.h, ff_filter_get_nb_threads(ctx)));
483 }
484
485 av_frame_free(&frame);
486 return ff_filter_frame(outlink, out);
487}
488
489static av_cold void uninit(AVFilterContext *ctx)
490{
491 PerspectiveContext *s = ctx->priv;
492
493 av_freep(&s->pv);
494}
495
496static const AVFilterPad perspective_inputs[] = {
497 {
498 .name = "default",
499 .type = AVMEDIA_TYPE_VIDEO,
500 .filter_frame = filter_frame,
501 .config_props = config_input,
502 },
503 { NULL }
504};
505
506static const AVFilterPad perspective_outputs[] = {
507 {
508 .name = "default",
509 .type = AVMEDIA_TYPE_VIDEO,
510 },
511 { NULL }
512};
513
514AVFilter ff_vf_perspective = {
515 .name = "perspective",
516 .description = NULL_IF_CONFIG_SMALL("Correct the perspective of video."),
517 .priv_size = sizeof(PerspectiveContext),
518 .init = init,
519 .uninit = uninit,
520 .query_formats = query_formats,
521 .inputs = perspective_inputs,
522 .outputs = perspective_outputs,
523 .priv_class = &perspective_class,
524 .flags = AVFILTER_FLAG_SUPPORT_TIMELINE_GENERIC | AVFILTER_FLAG_SLICE_THREADS,
525};
526
generated by cgit at 2024-05-17 04:20:52 (GMT)