blob: 1cb4c9aab8acbaf0988922596ab8811bf644d407
1 | /* |
2 | * Copyright (c) 2016 Paul B Mahol |
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 | * threshold video filter |
24 | */ |
25 | |
26 | #include "libavutil/imgutils.h" |
27 | #include "libavutil/internal.h" |
28 | #include "libavutil/opt.h" |
29 | #include "libavutil/pixdesc.h" |
30 | #include "avfilter.h" |
31 | #include "framesync.h" |
32 | #include "internal.h" |
33 | #include "video.h" |
34 | |
35 | typedef struct ThresholdContext { |
36 | const AVClass *class; |
37 | |
38 | int planes; |
39 | int bpc; |
40 | |
41 | int nb_planes; |
42 | int width[4], height[4]; |
43 | |
44 | void (*threshold)(const uint8_t *in, const uint8_t *threshold, |
45 | const uint8_t *min, const uint8_t *max, |
46 | uint8_t *out, |
47 | ptrdiff_t ilinesize, ptrdiff_t tlinesize, |
48 | ptrdiff_t flinesize, ptrdiff_t slinesize, |
49 | ptrdiff_t olinesize, |
50 | int w, int h); |
51 | |
52 | AVFrame *frames[4]; |
53 | FFFrameSync fs; |
54 | } ThresholdContext; |
55 | |
56 | #define OFFSET(x) offsetof(ThresholdContext, x) |
57 | #define FLAGS AV_OPT_FLAG_VIDEO_PARAM|AV_OPT_FLAG_FILTERING_PARAM |
58 | |
59 | static const AVOption threshold_options[] = { |
60 | { "planes", "set planes to filter", OFFSET(planes), AV_OPT_TYPE_INT, {.i64=15}, 0, 15, FLAGS}, |
61 | { NULL } |
62 | }; |
63 | |
64 | AVFILTER_DEFINE_CLASS(threshold); |
65 | |
66 | static int query_formats(AVFilterContext *ctx) |
67 | { |
68 | static const enum AVPixelFormat pix_fmts[] = { |
69 | AV_PIX_FMT_YUVA444P, AV_PIX_FMT_YUV444P, AV_PIX_FMT_YUV440P, |
70 | AV_PIX_FMT_YUVJ444P, AV_PIX_FMT_YUVJ440P, |
71 | AV_PIX_FMT_YUVA422P, AV_PIX_FMT_YUV422P, AV_PIX_FMT_YUVA420P, AV_PIX_FMT_YUV420P, |
72 | AV_PIX_FMT_YUVJ422P, AV_PIX_FMT_YUVJ420P, |
73 | AV_PIX_FMT_YUVJ411P, AV_PIX_FMT_YUV411P, AV_PIX_FMT_YUV410P, |
74 | AV_PIX_FMT_YUV420P9, AV_PIX_FMT_YUV422P9, AV_PIX_FMT_YUV444P9, |
75 | AV_PIX_FMT_YUV420P10, AV_PIX_FMT_YUV422P10, AV_PIX_FMT_YUV444P10, |
76 | AV_PIX_FMT_YUV420P16, AV_PIX_FMT_YUV422P16, AV_PIX_FMT_YUV444P16, |
77 | AV_PIX_FMT_YUVA420P9, AV_PIX_FMT_YUVA422P9, AV_PIX_FMT_YUVA444P9, |
78 | AV_PIX_FMT_YUVA420P10, AV_PIX_FMT_YUVA422P10, AV_PIX_FMT_YUVA444P10, |
79 | AV_PIX_FMT_YUVA420P16, AV_PIX_FMT_YUVA422P16, AV_PIX_FMT_YUVA444P16, |
80 | AV_PIX_FMT_GBRP, AV_PIX_FMT_GBRP9, AV_PIX_FMT_GBRP10, |
81 | AV_PIX_FMT_GBRP12, AV_PIX_FMT_GBRP14, AV_PIX_FMT_GBRP16, |
82 | AV_PIX_FMT_GBRAP, AV_PIX_FMT_GBRAP16, |
83 | AV_PIX_FMT_GRAY8, AV_PIX_FMT_GRAY10, |
84 | AV_PIX_FMT_GRAY12, AV_PIX_FMT_GRAY16, |
85 | AV_PIX_FMT_NONE |
86 | }; |
87 | |
88 | return ff_set_common_formats(ctx, ff_make_format_list(pix_fmts)); |
89 | } |
90 | |
91 | static int process_frame(FFFrameSync *fs) |
92 | { |
93 | AVFilterContext *ctx = fs->parent; |
94 | ThresholdContext *s = fs->opaque; |
95 | AVFilterLink *outlink = ctx->outputs[0]; |
96 | AVFrame *out, *in, *threshold, *min, *max; |
97 | int ret; |
98 | |
99 | if ((ret = ff_framesync_get_frame(&s->fs, 0, &in, 0)) < 0 || |
100 | (ret = ff_framesync_get_frame(&s->fs, 1, &threshold, 0)) < 0 || |
101 | (ret = ff_framesync_get_frame(&s->fs, 2, &min, 0)) < 0 || |
102 | (ret = ff_framesync_get_frame(&s->fs, 3, &max, 0)) < 0) |
103 | return ret; |
104 | |
105 | if (ctx->is_disabled) { |
106 | out = av_frame_clone(in); |
107 | if (!out) |
108 | return AVERROR(ENOMEM); |
109 | } else { |
110 | int p; |
111 | |
112 | out = ff_get_video_buffer(outlink, outlink->w, outlink->h); |
113 | if (!out) |
114 | return AVERROR(ENOMEM); |
115 | av_frame_copy_props(out, in); |
116 | |
117 | for (p = 0; p < s->nb_planes; p++) { |
118 | if (!(s->planes & (1 << p))) { |
119 | av_image_copy_plane(out->data[p], out->linesize[p], |
120 | in->data[p], in->linesize[p], |
121 | s->width[p] * s->bpc, |
122 | s->height[p]); |
123 | continue; |
124 | } |
125 | s->threshold(in->data[p], threshold->data[p], |
126 | min->data[p], max->data[p], |
127 | out->data[p], |
128 | in->linesize[p], threshold->linesize[p], |
129 | min->linesize[p], max->linesize[p], |
130 | out->linesize[p], |
131 | s->width[p], s->height[p]); |
132 | } |
133 | } |
134 | |
135 | out->pts = av_rescale_q(s->fs.pts, s->fs.time_base, outlink->time_base); |
136 | |
137 | return ff_filter_frame(outlink, out); |
138 | } |
139 | |
140 | static void threshold8(const uint8_t *in, const uint8_t *threshold, |
141 | const uint8_t *min, const uint8_t *max, |
142 | uint8_t *out, |
143 | ptrdiff_t ilinesize, ptrdiff_t tlinesize, |
144 | ptrdiff_t flinesize, ptrdiff_t slinesize, |
145 | ptrdiff_t olinesize, |
146 | int w, int h) |
147 | { |
148 | int x, y; |
149 | |
150 | for (y = 0; y < h; y++) { |
151 | for (x = 0; x < w; x++) { |
152 | out[x] = in[x] < threshold[x] ? min[x] : max[x]; |
153 | } |
154 | |
155 | in += ilinesize; |
156 | threshold += tlinesize; |
157 | min += flinesize; |
158 | max += flinesize; |
159 | out += olinesize; |
160 | } |
161 | } |
162 | |
163 | static void threshold16(const uint8_t *iin, const uint8_t *tthreshold, |
164 | const uint8_t *ffirst, const uint8_t *ssecond, |
165 | uint8_t *oout, |
166 | ptrdiff_t ilinesize, ptrdiff_t tlinesize, |
167 | ptrdiff_t flinesize, ptrdiff_t slinesize, |
168 | ptrdiff_t olinesize, |
169 | int w, int h) |
170 | { |
171 | const uint16_t *in = (const uint16_t *)iin; |
172 | const uint16_t *threshold = (const uint16_t *)tthreshold; |
173 | const uint16_t *min = (const uint16_t *)ffirst; |
174 | const uint16_t *max = (const uint16_t *)ssecond; |
175 | uint16_t *out = (uint16_t *)oout; |
176 | int x, y; |
177 | |
178 | for (y = 0; y < h; y++) { |
179 | for (x = 0; x < w; x++) { |
180 | out[x] = in[x] < threshold[x] ? min[x] : max[x]; |
181 | } |
182 | |
183 | in += ilinesize / 2; |
184 | threshold += tlinesize / 2; |
185 | min += flinesize / 2; |
186 | max += flinesize / 2; |
187 | out += olinesize / 2; |
188 | } |
189 | } |
190 | |
191 | static int config_input(AVFilterLink *inlink) |
192 | { |
193 | AVFilterContext *ctx = inlink->dst; |
194 | ThresholdContext *s = ctx->priv; |
195 | const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(inlink->format); |
196 | int vsub, hsub; |
197 | |
198 | s->nb_planes = av_pix_fmt_count_planes(inlink->format); |
199 | |
200 | hsub = desc->log2_chroma_w; |
201 | vsub = desc->log2_chroma_h; |
202 | s->height[1] = s->height[2] = AV_CEIL_RSHIFT(inlink->h, vsub); |
203 | s->height[0] = s->height[3] = inlink->h; |
204 | s->width[1] = s->width[2] = AV_CEIL_RSHIFT(inlink->w, hsub); |
205 | s->width[0] = s->width[3] = inlink->w; |
206 | |
207 | if (desc->comp[0].depth == 8) { |
208 | s->threshold = threshold8; |
209 | s->bpc = 1; |
210 | } else { |
211 | s->threshold = threshold16; |
212 | s->bpc = 2; |
213 | } |
214 | |
215 | return 0; |
216 | } |
217 | |
218 | static int config_output(AVFilterLink *outlink) |
219 | { |
220 | AVFilterContext *ctx = outlink->src; |
221 | ThresholdContext *s = ctx->priv; |
222 | AVFilterLink *base = ctx->inputs[0]; |
223 | AVFilterLink *threshold = ctx->inputs[1]; |
224 | AVFilterLink *min = ctx->inputs[2]; |
225 | AVFilterLink *max = ctx->inputs[3]; |
226 | FFFrameSyncIn *in; |
227 | int ret; |
228 | |
229 | if (base->format != threshold->format || |
230 | base->format != min->format || |
231 | base->format != max->format) { |
232 | av_log(ctx, AV_LOG_ERROR, "inputs must be of same pixel format\n"); |
233 | return AVERROR(EINVAL); |
234 | } |
235 | if (base->w != threshold->w || |
236 | base->h != threshold->h || |
237 | base->w != min->w || |
238 | base->h != min->h || |
239 | base->w != max->w || |
240 | base->h != max->h) { |
241 | av_log(ctx, AV_LOG_ERROR, "First input link %s parameters " |
242 | "(size %dx%d) do not match the corresponding " |
243 | "second input link %s parameters (%dx%d) " |
244 | "and/or third input link %s parameters (%dx%d) " |
245 | "and/or fourth input link %s parameters (%dx%d)\n", |
246 | ctx->input_pads[0].name, base->w, base->h, |
247 | ctx->input_pads[1].name, threshold->w, threshold->h, |
248 | ctx->input_pads[2].name, min->w, min->h, |
249 | ctx->input_pads[3].name, max->w, max->h); |
250 | return AVERROR(EINVAL); |
251 | } |
252 | |
253 | outlink->w = base->w; |
254 | outlink->h = base->h; |
255 | outlink->time_base = base->time_base; |
256 | outlink->sample_aspect_ratio = base->sample_aspect_ratio; |
257 | outlink->frame_rate = base->frame_rate; |
258 | |
259 | if ((ret = ff_framesync_init(&s->fs, ctx, 4)) < 0) |
260 | return ret; |
261 | |
262 | in = s->fs.in; |
263 | in[0].time_base = base->time_base; |
264 | in[1].time_base = threshold->time_base; |
265 | in[2].time_base = min->time_base; |
266 | in[3].time_base = max->time_base; |
267 | in[0].sync = 1; |
268 | in[0].before = EXT_STOP; |
269 | in[0].after = EXT_STOP; |
270 | in[1].sync = 1; |
271 | in[1].before = EXT_STOP; |
272 | in[1].after = EXT_STOP; |
273 | in[2].sync = 1; |
274 | in[2].before = EXT_STOP; |
275 | in[2].after = EXT_STOP; |
276 | in[3].sync = 1; |
277 | in[3].before = EXT_STOP; |
278 | in[3].after = EXT_STOP; |
279 | s->fs.opaque = s; |
280 | s->fs.on_event = process_frame; |
281 | |
282 | return ff_framesync_configure(&s->fs); |
283 | } |
284 | |
285 | static int filter_frame(AVFilterLink *inlink, AVFrame *buf) |
286 | { |
287 | ThresholdContext *s = inlink->dst->priv; |
288 | return ff_framesync_filter_frame(&s->fs, inlink, buf); |
289 | } |
290 | |
291 | static int request_frame(AVFilterLink *outlink) |
292 | { |
293 | ThresholdContext *s = outlink->src->priv; |
294 | return ff_framesync_request_frame(&s->fs, outlink); |
295 | } |
296 | |
297 | static av_cold void uninit(AVFilterContext *ctx) |
298 | { |
299 | ThresholdContext *s = ctx->priv; |
300 | |
301 | ff_framesync_uninit(&s->fs); |
302 | } |
303 | |
304 | static const AVFilterPad inputs[] = { |
305 | { |
306 | .name = "default", |
307 | .type = AVMEDIA_TYPE_VIDEO, |
308 | .filter_frame = filter_frame, |
309 | .config_props = config_input, |
310 | }, |
311 | { |
312 | .name = "threshold", |
313 | .type = AVMEDIA_TYPE_VIDEO, |
314 | .filter_frame = filter_frame, |
315 | }, |
316 | { |
317 | .name = "min", |
318 | .type = AVMEDIA_TYPE_VIDEO, |
319 | .filter_frame = filter_frame, |
320 | }, |
321 | { |
322 | .name = "max", |
323 | .type = AVMEDIA_TYPE_VIDEO, |
324 | .filter_frame = filter_frame, |
325 | }, |
326 | { NULL } |
327 | }; |
328 | |
329 | static const AVFilterPad outputs[] = { |
330 | { |
331 | .name = "default", |
332 | .type = AVMEDIA_TYPE_VIDEO, |
333 | .config_props = config_output, |
334 | .request_frame = request_frame, |
335 | }, |
336 | { NULL } |
337 | }; |
338 | |
339 | AVFilter ff_vf_threshold = { |
340 | .name = "threshold", |
341 | .description = NULL_IF_CONFIG_SMALL("Threshold first video stream using other video streams."), |
342 | .priv_size = sizeof(ThresholdContext), |
343 | .priv_class = &threshold_class, |
344 | .uninit = uninit, |
345 | .query_formats = query_formats, |
346 | .inputs = inputs, |
347 | .outputs = outputs, |
348 | .flags = AVFILTER_FLAG_SUPPORT_TIMELINE_INTERNAL, |
349 | }; |
350 |