libavfilter/vf_overlay.c

   1 /*
   2  * Copyright (c) 2010 Stefano Sabatini
   3  * Copyright (c) 2010 Baptiste Coudurier
   4  * Copyright (c) 2007 Bobby Bingham
   5  *
   6  * This file is part of Libav.
   7  *
   8  * Libav is free software; you can redistribute it and/or
   9  * modify it under the terms of the GNU Lesser General Public
  10  * License as published by the Free Software Foundation; either
  11  * version 2.1 of the License, or (at your option) any later version.
  12  *
  13  * Libav is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  16  * Lesser General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU Lesser General Public
  19  * License along with Libav; if not, write to the Free Software
  20  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  21  */
  22
  23 /**
  24  * @file
  25  * overlay one video on top of another
  26  */
  27
  28 #include "avfilter.h"
  29 #include "formats.h"
  30 #include "libavutil/common.h"
  31 #include "libavutil/eval.h"
  32 #include "libavutil/avstring.h"
  33 #include "libavutil/avassert.h"
  34 #include "libavutil/pixdesc.h"
  35 #include "libavutil/imgutils.h"
  36 #include "libavutil/mathematics.h"
  37 #include "internal.h"
  38 #include "video.h"
  39
  40 static const char *const var_names[] = {
  41     "E",
  42     "PHI",
  43     "PI",
  44     "main_w",    "W", ///< width  of the main    video
  45     "main_h",    "H", ///< height of the main    video
  46     "overlay_w", "w", ///< width  of the overlay video
  47     "overlay_h", "h", ///< height of the overlay video
  48     NULL
  49 };
  50
  51 enum var_name {
  52     VAR_E,
  53     VAR_PHI,
  54     VAR_PI,
  55     VAR_MAIN_W,    VAR_MW,
  56     VAR_MAIN_H,    VAR_MH,
  57     VAR_OVERLAY_W, VAR_OW,
  58     VAR_OVERLAY_H, VAR_OH,
  59     VAR_VARS_NB
  60 };
  61
  62 #define MAIN    0
  63 #define OVERLAY 1
  64
  65 typedef struct {
  66     int x, y;                   ///< position of overlayed picture
  67
  68     int max_plane_step[4];      ///< steps per pixel for each plane
  69     int hsub, vsub;             ///< chroma subsampling values
  70
  71     char x_expr[256], y_expr[256];
  72
  73     AVFilterBufferRef *main;
  74     AVFilterBufferRef *over_prev, *over_next;
  75 } OverlayContext;
  76
  77 static av_cold int init(AVFilterContext *ctx, const char *args)
  78 {
  79     OverlayContext *over = ctx->priv;
  80
  81     av_strlcpy(over->x_expr, "0", sizeof(over->x_expr));
  82     av_strlcpy(over->y_expr, "0", sizeof(over->y_expr));
  83
  84     if (args)
  85         sscanf(args, "%255[^:]:%255[^:]", over->x_expr, over->y_expr);
  86
  87     return 0;
  88 }
  89
  90 static av_cold void uninit(AVFilterContext *ctx)
  91 {
  92     OverlayContext *s = ctx->priv;
  93
  94     avfilter_unref_bufferp(&s->main);
  95     avfilter_unref_bufferp(&s->over_prev);
  96     avfilter_unref_bufferp(&s->over_next);
  97 }
  98
  99 static int query_formats(AVFilterContext *ctx)
 100 {
 101     const enum AVPixelFormat inout_pix_fmts[] = { AV_PIX_FMT_YUV420P,  AV_PIX_FMT_NONE };
 102     const enum AVPixelFormat blend_pix_fmts[] = { AV_PIX_FMT_YUVA420P, AV_PIX_FMT_NONE };
 103     AVFilterFormats *inout_formats = ff_make_format_list(inout_pix_fmts);
 104     AVFilterFormats *blend_formats = ff_make_format_list(blend_pix_fmts);
 105
 106     ff_formats_ref(inout_formats, &ctx->inputs [MAIN   ]->out_formats);
 107     ff_formats_ref(blend_formats, &ctx->inputs [OVERLAY]->out_formats);
 108     ff_formats_ref(inout_formats, &ctx->outputs[MAIN   ]->in_formats );
 109
 110     return 0;
 111 }
 112
 113 static int config_input_main(AVFilterLink *inlink)
 114 {
 115     OverlayContext *over = inlink->dst->priv;
 116     const AVPixFmtDescriptor *pix_desc = av_pix_fmt_desc_get(inlink->format);
 117
 118     av_image_fill_max_pixsteps(over->max_plane_step, NULL, pix_desc);
 119     over->hsub = pix_desc->log2_chroma_w;
 120     over->vsub = pix_desc->log2_chroma_h;
 121
 122     return 0;
 123 }
 124
 125 static int config_input_overlay(AVFilterLink *inlink)
 126 {
 127     AVFilterContext *ctx  = inlink->dst;
 128     OverlayContext  *over = inlink->dst->priv;
 129     char *expr;
 130     double var_values[VAR_VARS_NB], res;
 131     int ret;
 132
 133     /* Finish the configuration by evaluating the expressions
 134        now when both inputs are configured. */
 135     var_values[VAR_E  ] = M_E;
 136     var_values[VAR_PHI] = M_PHI;
 137     var_values[VAR_PI ] = M_PI;
 138
 139     var_values[VAR_MAIN_W   ] = var_values[VAR_MW] = ctx->inputs[MAIN   ]->w;
 140     var_values[VAR_MAIN_H   ] = var_values[VAR_MH] = ctx->inputs[MAIN   ]->h;
 141     var_values[VAR_OVERLAY_W] = var_values[VAR_OW] = ctx->inputs[OVERLAY]->w;
 142     var_values[VAR_OVERLAY_H] = var_values[VAR_OH] = ctx->inputs[OVERLAY]->h;
 143
 144     if ((ret = av_expr_parse_and_eval(&res, (expr = over->x_expr), var_names, var_values,
 145                                       NULL, NULL, NULL, NULL, NULL, 0, ctx)) < 0)
 146         goto fail;
 147     over->x = res;
 148     if ((ret = av_expr_parse_and_eval(&res, (expr = over->y_expr), var_names, var_values,
 149                                       NULL, NULL, NULL, NULL, NULL, 0, ctx)))
 150         goto fail;
 151     over->y = res;
 152     /* x may depend on y */
 153     if ((ret = av_expr_parse_and_eval(&res, (expr = over->x_expr), var_names, var_values,
 154                                       NULL, NULL, NULL, NULL, NULL, 0, ctx)) < 0)
 155         goto fail;
 156     over->x = res;
 157
 158     av_log(ctx, AV_LOG_VERBOSE,
 159            "main w:%d h:%d fmt:%s overlay x:%d y:%d w:%d h:%d fmt:%s\n",
 160            ctx->inputs[MAIN]->w, ctx->inputs[MAIN]->h,
 161            av_get_pix_fmt_name(ctx->inputs[MAIN]->format),
 162            over->x, over->y,
 163            ctx->inputs[OVERLAY]->w, ctx->inputs[OVERLAY]->h,
 164            av_get_pix_fmt_name(ctx->inputs[OVERLAY]->format));
 165
 166     if (over->x < 0 || over->y < 0 ||
 167         over->x + var_values[VAR_OVERLAY_W] > var_values[VAR_MAIN_W] ||
 168         over->y + var_values[VAR_OVERLAY_H] > var_values[VAR_MAIN_H]) {
 169         av_log(ctx, AV_LOG_ERROR,
 170                "Overlay area (%d,%d)<->(%d,%d) not within the main area (0,0)<->(%d,%d) or zero-sized\n",
 171                over->x, over->y,
 172                (int)(over->x + var_values[VAR_OVERLAY_W]),
 173                (int)(over->y + var_values[VAR_OVERLAY_H]),
 174                (int)var_values[VAR_MAIN_W], (int)var_values[VAR_MAIN_H]);
 175         return AVERROR(EINVAL);
 176     }
 177     return 0;
 178
 179 fail:
 180     av_log(NULL, AV_LOG_ERROR,
 181            "Error when evaluating the expression '%s'\n", expr);
 182     return ret;
 183 }
 184
 185 static int config_output(AVFilterLink *outlink)
 186 {
 187     AVFilterContext *ctx = outlink->src;
 188
 189     outlink->w = ctx->inputs[MAIN]->w;
 190     outlink->h = ctx->inputs[MAIN]->h;
 191     outlink->time_base = ctx->inputs[MAIN]->time_base;
 192
 193     return 0;
 194 }
 195
 196 static void blend_frame(AVFilterContext *ctx,
 197                         AVFilterBufferRef *dst, AVFilterBufferRef *src,
 198                         int x, int y)
 199 {
 200     OverlayContext *over = ctx->priv;
 201     int i, j, k;
 202     int width, height;
 203     int overlay_end_y = y + src->video->h;
 204     int end_y, start_y;
 205
 206     width = FFMIN(dst->video->w - x, src->video->w);
 207     end_y = FFMIN(dst->video->h, overlay_end_y);
 208     start_y = FFMAX(y, 0);
 209     height = end_y - start_y;
 210
 211     if (dst->format == AV_PIX_FMT_BGR24 || dst->format == AV_PIX_FMT_RGB24) {
 212         uint8_t *dp = dst->data[0] + x * 3 + start_y * dst->linesize[0];
 213         uint8_t *sp = src->data[0];
 214         int b = dst->format == AV_PIX_FMT_BGR24 ? 2 : 0;
 215         int r = dst->format == AV_PIX_FMT_BGR24 ? 0 : 2;
 216         if (y < 0)
 217             sp += -y * src->linesize[0];
 218         for (i = 0; i < height; i++) {
 219             uint8_t *d = dp, *s = sp;
 220             for (j = 0; j < width; j++) {
 221                 d[r] = (d[r] * (0xff - s[3]) + s[0] * s[3] + 128) >> 8;
 222                 d[1] = (d[1] * (0xff - s[3]) + s[1] * s[3] + 128) >> 8;
 223                 d[b] = (d[b] * (0xff - s[3]) + s[2] * s[3] + 128) >> 8;
 224                 d += 3;
 225                 s += 4;
 226             }
 227             dp += dst->linesize[0];
 228             sp += src->linesize[0];
 229         }
 230     } else {
 231         for (i = 0; i < 3; i++) {
 232             int hsub = i ? over->hsub : 0;
 233             int vsub = i ? over->vsub : 0;
 234             uint8_t *dp = dst->data[i] + (x >> hsub) +
 235                 (start_y >> vsub) * dst->linesize[i];
 236             uint8_t *sp = src->data[i];
 237             uint8_t *ap = src->data[3];
 238             int wp = FFALIGN(width, 1<<hsub) >> hsub;
 239             int hp = FFALIGN(height, 1<<vsub) >> vsub;
 240             if (y < 0) {
 241                 sp += ((-y) >> vsub) * src->linesize[i];
 242                 ap += -y * src->linesize[3];
 243             }
 244             for (j = 0; j < hp; j++) {
 245                 uint8_t *d = dp, *s = sp, *a = ap;
 246                 for (k = 0; k < wp; k++) {
 247                     // average alpha for color components, improve quality
 248                     int alpha_v, alpha_h, alpha;
 249                     if (hsub && vsub && j+1 < hp && k+1 < wp) {
 250                         alpha = (a[0] + a[src->linesize[3]] +
 251                                  a[1] + a[src->linesize[3]+1]) >> 2;
 252                     } else if (hsub || vsub) {
 253                         alpha_h = hsub && k+1 < wp ?
 254                             (a[0] + a[1]) >> 1 : a[0];
 255                         alpha_v = vsub && j+1 < hp ?
 256                             (a[0] + a[src->linesize[3]]) >> 1 : a[0];
 257                         alpha = (alpha_v + alpha_h) >> 1;
 258                     } else
 259                         alpha = a[0];
 260                     *d = (*d * (0xff - alpha) + *s++ * alpha + 128) >> 8;
 261                     d++;
 262                     a += 1 << hsub;
 263                 }
 264                 dp += dst->linesize[i];
 265                 sp += src->linesize[i];
 266                 ap += (1 << vsub) * src->linesize[3];
 267             }
 268         }
 269     }
 270 }
 271
 272 static int filter_frame_main(AVFilterLink *inlink, AVFilterBufferRef *frame)
 273 {
 274     OverlayContext *s = inlink->dst->priv;
 275
 276     av_assert0(!s->main);
 277     s->main         = frame;
 278
 279     return 0;
 280 }
 281
 282 static int filter_frame_overlay(AVFilterLink *inlink, AVFilterBufferRef *frame)
 283 {
 284     OverlayContext *s = inlink->dst->priv;
 285
 286     av_assert0(!s->over_next);
 287     s->over_next    = frame;
 288
 289     return 0;
 290 }
 291
 292 static int output_frame(AVFilterContext *ctx)
 293 {
 294     OverlayContext *s = ctx->priv;
 295     AVFilterLink *outlink = ctx->outputs[0];
 296     int ret = ff_filter_frame(outlink, s->main);
 297     s->main = NULL;
 298
 299     return ret;
 300 }
 301
 302 static int handle_overlay_eof(AVFilterContext *ctx)
 303 {
 304     OverlayContext *s = ctx->priv;
 305     if (s->over_prev)
 306         blend_frame(ctx, s->main, s->over_prev, s->x, s->y);
 307     return output_frame(ctx);
 308 }
 309
 310 static int request_frame(AVFilterLink *outlink)
 311 {
 312     AVFilterContext *ctx = outlink->src;
 313     OverlayContext    *s = ctx->priv;
 314     AVRational tb_main = ctx->inputs[MAIN]->time_base;
 315     AVRational tb_over = ctx->inputs[OVERLAY]->time_base;
 316     int ret = 0;
 317
 318     /* get a frame on the main input */
 319     if (!s->main) {
 320         ret = ff_request_frame(ctx->inputs[MAIN]);
 321         if (ret < 0)
 322             return ret;
 323     }
 324
 325     /* get a new frame on the overlay input, on EOF
 326      * reuse previous */
 327     if (!s->over_next) {
 328         ret = ff_request_frame(ctx->inputs[OVERLAY]);
 329         if (ret == AVERROR_EOF)
 330            return handle_overlay_eof(ctx);
 331         else if (ret < 0)
 332             return ret;
 333     }
 334
 335     while (s->main->pts != AV_NOPTS_VALUE &&
 336            s->over_next->pts != AV_NOPTS_VALUE &&
 337            av_compare_ts(s->over_next->pts, tb_over, s->main->pts, tb_main) < 0) {
 338         avfilter_unref_bufferp(&s->over_prev);
 339         FFSWAP(AVFilterBufferRef*, s->over_prev, s->over_next);
 340
 341         ret = ff_request_frame(ctx->inputs[OVERLAY]);
 342         if (ret == AVERROR_EOF)
 343             return handle_overlay_eof(ctx);
 344         else if (ret < 0)
 345             return ret;
 346     }
 347
 348     if (s->main->pts == AV_NOPTS_VALUE ||
 349         s->over_next->pts == AV_NOPTS_VALUE ||
 350         !av_compare_ts(s->over_next->pts, tb_over, s->main->pts, tb_main)) {
 351         blend_frame(ctx, s->main, s->over_next, s->x, s->y);
 352         avfilter_unref_bufferp(&s->over_prev);
 353         FFSWAP(AVFilterBufferRef*, s->over_prev, s->over_next);
 354     } else if (s->over_prev) {
 355         blend_frame(ctx, s->main, s->over_prev, s->x, s->y);
 356     }
 357
 358     return output_frame(ctx);
 359 }
 360
 361 static const AVFilterPad avfilter_vf_overlay_inputs[] = {
 362     {
 363         .name         = "main",
 364         .type         = AVMEDIA_TYPE_VIDEO,
 365         .config_props = config_input_main,
 366         .filter_frame = filter_frame_main,
 367         .min_perms    = AV_PERM_READ,
 368         .rej_perms    = AV_PERM_REUSE2 | AV_PERM_PRESERVE,
 369         .needs_fifo   = 1,
 370     },
 371     {
 372         .name         = "overlay",
 373         .type         = AVMEDIA_TYPE_VIDEO,
 374         .config_props = config_input_overlay,
 375         .filter_frame = filter_frame_overlay,
 376         .min_perms    = AV_PERM_READ,
 377         .rej_perms    = AV_PERM_REUSE2,
 378         .needs_fifo   = 1,
 379     },
 380     { NULL }
 381 };
 382
 383 static const AVFilterPad avfilter_vf_overlay_outputs[] = {
 384     {
 385         .name          = "default",
 386         .type          = AVMEDIA_TYPE_VIDEO,
 387         .config_props  = config_output,
 388         .request_frame = request_frame,
 389     },
 390     { NULL }
 391 };
 392
 393 AVFilter avfilter_vf_overlay = {
 394     .name      = "overlay",
 395     .description = NULL_IF_CONFIG_SMALL("Overlay a video source on top of the input."),
 396
 397     .init      = init,
 398     .uninit    = uninit,
 399
 400     .priv_size = sizeof(OverlayContext),
 401
 402     .query_formats = query_formats,
 403
 404     .inputs    = avfilter_vf_overlay_inputs,
 405     .outputs   = avfilter_vf_overlay_outputs,
 406 };