diff --git a/doc/filters.texi b/doc/filters.texi index e325bb519b..216ce42ea1 100644 --- a/doc/filters.texi +++ b/doc/filters.texi @@ -10388,7 +10388,7 @@ settb=AVTB @end itemize @section showcqt -Convert input audio to a video output (at full HD resolution), representing ++Convert input audio to a video output representing frequency spectrum logarithmically (using constant Q transform with Brown-Puckette algorithm), with musical tone scale, from E0 to D#10 (10 octaves). @@ -10416,6 +10416,10 @@ Specify gamma. Lower gamma makes the spectrum more contrast, higher gamma makes the spectrum having more range. Acceptable value is [1.0, 7.0]. Default value is @code{3.0}. +@item fullhd +If set to 1 (the default), the video size is 1920x1080 (full HD), +if set to 0, the video size is 960x540. Use this option to make CPU usage lower. + @item fps Specify video fps. Default value is @code{25}. @@ -10441,6 +10445,12 @@ Same as above, but with frame rate 30 fps: ffplay -f lavfi 'amovie=a.mp3, asplit [a][out1]; [a] showcqt=fps=30:count=5 [out0]' @end example +@item +Playing at 960x540 and lower CPU usage: +@example +ffplay -f lavfi 'amovie=a.mp3, asplit [a][out1]; [a] showcqt=fullhd=0:count=3 [out0]' +@end example + @item A1 and its harmonics: A1, A2, (near)E3, A3: @example diff --git a/libavfilter/avf_showcqt.c b/libavfilter/avf_showcqt.c index 8b4309c709..1e59bd82c0 100644 --- a/libavfilter/avf_showcqt.c +++ b/libavfilter/avf_showcqt.c @@ -56,11 +56,11 @@ typedef struct { FFTComplex *fft_data; FFTComplex *fft_result_left; FFTComplex *fft_result_right; + uint8_t *spectogram; SparseCoeff *coeff_sort; SparseCoeff *coeffs[VIDEO_WIDTH]; int coeffs_len[VIDEO_WIDTH]; uint8_t font_color[VIDEO_WIDTH]; - uint8_t spectogram[SPECTOGRAM_HEIGHT][VIDEO_WIDTH][3]; int64_t frame_count; int spectogram_count; int spectogram_index; @@ -70,6 +70,7 @@ typedef struct { double volume; double timeclamp; /* lower timeclamp, time-accurate, higher timeclamp, freq-accurate (at low freq)*/ float coeffclamp; /* lower coeffclamp, more precise, higher coeffclamp, faster */ + int fullhd; /* if true, output video is at full HD resolution, otherwise it will be halved */ float gamma; /* lower gamma, more contrast, higher gamma, more range */ int fps; /* the required fps is so strict, so it's enough to be int, but 24000/1001 etc cannot be encoded */ int count; /* fps * count = transform rate */ @@ -83,6 +84,7 @@ static const AVOption showcqt_options[] = { { "timeclamp", "set timeclamp", OFFSET(timeclamp), AV_OPT_TYPE_DOUBLE, { .dbl = 0.17 }, 0.1, 1.0, FLAGS }, { "coeffclamp", "set coeffclamp", OFFSET(coeffclamp), AV_OPT_TYPE_FLOAT, { .dbl = 1 }, 0.1, 10, FLAGS }, { "gamma", "set gamma", OFFSET(gamma), AV_OPT_TYPE_FLOAT, { .dbl = 3 }, 1, 7, FLAGS }, + { "fullhd", "set full HD resolution", OFFSET(fullhd), AV_OPT_TYPE_INT, { .i64 = 1 }, 0, 1, FLAGS }, { "fps", "set video fps", OFFSET(fps), AV_OPT_TYPE_INT, { .i64 = 25 }, 10, 100, FLAGS }, { "count", "set number of transform per frame", OFFSET(count), AV_OPT_TYPE_INT, { .i64 = 6 }, 1, 30, FLAGS }, { NULL } @@ -103,6 +105,7 @@ static av_cold void uninit(AVFilterContext *ctx) av_freep(&s->fft_result_left); av_freep(&s->fft_result_right); av_freep(&s->coeff_sort); + av_freep(&s->spectogram); av_frame_free(&s->outpicref); } @@ -160,6 +163,10 @@ static int config_output(AVFilterLink *outlink) int rate = inlink->sample_rate; double max_len = rate * (double) s->timeclamp; int64_t start_time, end_time; + int video_scale = s->fullhd ? 2 : 1; + int video_width = (VIDEO_WIDTH/2) * video_scale; + int video_height = (VIDEO_HEIGHT/2) * video_scale; + int spectogram_height = (SPECTOGRAM_HEIGHT/2) * video_scale; s->fft_bits = ceil(log2(max_len)); fft_len = 1 << s->fft_bits; @@ -179,9 +186,9 @@ static int config_output(AVFilterLink *outlink) return AVERROR(ENOMEM); /* initializing font */ - for (x = 0; x < VIDEO_WIDTH; x++) { - if (x >= (12*3+8)*16 && x < (12*4+8)*16) { - float fx = (x-(12*3+8)*16) * (1.0f/192.0f); + for (x = 0; x < video_width; x++) { + if (x >= (12*3+8)*8*video_scale && x < (12*4+8)*8*video_scale) { + float fx = (x-(12*3+8)*8*video_scale) * (2.0f/(192.0f*video_scale)); float sv = sinf(M_PI*fx); s->font_color[x] = sv*sv*255.0f + 0.5f; } else { @@ -273,21 +280,24 @@ static int config_output(AVFilterLink *outlink) end_time = av_gettime_relative(); av_log(ctx, AV_LOG_INFO, "Elapsed time %.6f s (fft_len=%u, num_coeffs=%u)\n", 1e-6 * (end_time-start_time), fft_len, num_coeffs); - outlink->w = VIDEO_WIDTH; - outlink->h = VIDEO_HEIGHT; + outlink->w = video_width; + outlink->h = video_height; s->req_fullfilled = 0; s->spectogram_index = 0; s->frame_count = 0; s->spectogram_count = 0; s->remaining_fill = fft_len >> 1; - memset(s->spectogram, 0, VIDEO_WIDTH * SPECTOGRAM_HEIGHT * 3); memset(s->fft_data, 0, fft_len * sizeof(*s->fft_data)); s->outpicref = ff_get_video_buffer(outlink, outlink->w, outlink->h); if (!s->outpicref) return AVERROR(ENOMEM); + s->spectogram = av_calloc(spectogram_height, s->outpicref->linesize[0]); + if (!s->spectogram) + return AVERROR(ENOMEM); + outlink->sample_aspect_ratio = av_make_q(1, 1); outlink->time_base = av_make_q(1, s->fps); outlink->frame_rate = av_make_q(s->fps, 1); @@ -302,6 +312,12 @@ static int plot_cqt(AVFilterLink *inlink) int fft_len = 1 << s->fft_bits; FFTSample result[VIDEO_WIDTH][4]; int x, y, ret = 0; + int linesize = s->outpicref->linesize[0]; + int video_scale = s->fullhd ? 2 : 1; + int video_width = (VIDEO_WIDTH/2) * video_scale; + int spectogram_height = (SPECTOGRAM_HEIGHT/2) * video_scale; + int spectogram_start = (SPECTOGRAM_START/2) * video_scale; + int font_height = (FONT_HEIGHT/2) * video_scale; /* real part contains left samples, imaginary part contains right samples */ memcpy(s->fft_result_left, s->fft_data, fft_len * sizeof(*s->fft_data)); @@ -352,26 +368,34 @@ static int plot_cqt(AVFilterLink *inlink) result[x][2] = 255.0f * powf(FFMIN(1.0f,result[x][2]), g); } - for (x = 0; x < VIDEO_WIDTH; x++) { - s->spectogram[s->spectogram_index][x][0] = result[x][0] + 0.5f; - s->spectogram[s->spectogram_index][x][1] = result[x][1] + 0.5f; - s->spectogram[s->spectogram_index][x][2] = result[x][2] + 0.5f; + if (!s->fullhd) { + for (x = 0; x < video_width; x++) { + result[x][0] = 0.5f * (result[2*x][0] + result[2*x+1][0]); + result[x][1] = 0.5f * (result[2*x][1] + result[2*x+1][1]); + result[x][2] = 0.5f * (result[2*x][2] + result[2*x+1][2]); + result[x][3] = 0.5f * (result[2*x][3] + result[2*x+1][3]); + } + } + + for (x = 0; x < video_width; x++) { + s->spectogram[s->spectogram_index*linesize + 3*x] = result[x][0] + 0.5f; + s->spectogram[s->spectogram_index*linesize + 3*x + 1] = result[x][1] + 0.5f; + s->spectogram[s->spectogram_index*linesize + 3*x + 2] = result[x][2] + 0.5f; } /* drawing */ if (!s->spectogram_count) { uint8_t *data = (uint8_t*) s->outpicref->data[0]; - int linesize = s->outpicref->linesize[0]; float rcp_result[VIDEO_WIDTH]; - for (x = 0; x < VIDEO_WIDTH; x++) + for (x = 0; x < video_width; x++) rcp_result[x] = 1.0f / (result[x][3]+0.0001f); /* drawing bar */ - for (y = 0; y < SPECTOGRAM_HEIGHT; y++) { - float height = (SPECTOGRAM_HEIGHT - y) * (1.0f/SPECTOGRAM_HEIGHT); + for (y = 0; y < spectogram_height; y++) { + float height = (spectogram_height - y) * (1.0f/spectogram_height); uint8_t *lineptr = data + y * linesize; - for (x = 0; x < VIDEO_WIDTH; x++) { + for (x = 0; x < video_width; x++) { float mul; if (result[x][3] <= height) { *lineptr++ = 0; @@ -384,35 +408,39 @@ static int plot_cqt(AVFilterLink *inlink) *lineptr++ = mul * result[x][2] + 0.5f; } } - } /* drawing font */ - for (y = 0; y < FONT_HEIGHT; y++) { - uint8_t *lineptr = data + (SPECTOGRAM_HEIGHT + y) * linesize; - memcpy(lineptr, s->spectogram[s->spectogram_index], VIDEO_WIDTH*3); + for (y = 0; y < font_height; y++) { + uint8_t *lineptr = data + (spectogram_height + y) * linesize; + memcpy(lineptr, s->spectogram + s->spectogram_index * linesize, video_width*3); } - for (x = 0; x < VIDEO_WIDTH; x += VIDEO_WIDTH/10) { + for (x = 0; x < video_width; x += video_width/10) { int u; static const char str[] = "EF G A BC D "; - uint8_t *startptr = data + SPECTOGRAM_HEIGHT * linesize + x * 3; + uint8_t *startptr = data + spectogram_height * linesize + x * 3; for (u = 0; str[u]; u++) { int v; for (v = 0; v < 16; v++) { - uint8_t *p = startptr + 2 * v * linesize + 16 * 3 * u; - int ux = x + 16 * u; + uint8_t *p = startptr + v * linesize * video_scale + 8 * 3 * u * video_scale; + int ux = x + 8 * u * video_scale; int mask; for (mask = 0x80; mask; mask >>= 1) { if (mask & avpriv_vga16_font[str[u] * 16 + v]) { - p[0] = p[linesize] = 255 - s->font_color[ux]; - p[1] = p[linesize+1] = 0; - p[2] = p[linesize+2] = s->font_color[ux]; - p[3] = p[linesize+3] = 255 - s->font_color[ux+1]; - p[4] = p[linesize+4] = 0; - p[5] = p[linesize+5] = s->font_color[ux+1]; + p[0] = 255 - s->font_color[ux]; + p[1] = 0; + p[2] = s->font_color[ux]; + if (video_scale == 2) { + p[linesize] = p[0]; + p[linesize+1] = p[1]; + p[linesize+2] = p[2]; + p[3] = p[linesize+3] = 255 - s->font_color[ux+1]; + p[4] = p[linesize+4] = 0; + p[5] = p[linesize+5] = s->font_color[ux+1]; + } } - p += 6; - ux += 2; + p += 3 * video_scale; + ux += video_scale; } } } @@ -420,18 +448,14 @@ static int plot_cqt(AVFilterLink *inlink) } /* drawing spectogram/sonogram */ - if (linesize == VIDEO_WIDTH * 3) { - int total_length = VIDEO_WIDTH * SPECTOGRAM_HEIGHT * 3; - int back_length = VIDEO_WIDTH * s->spectogram_index * 3; - data += SPECTOGRAM_START * VIDEO_WIDTH * 3; - memcpy(data, s->spectogram[s->spectogram_index], total_length - back_length); - data += total_length - back_length; - if (back_length) - memcpy(data, s->spectogram[0], back_length); - } else { - for (y = 0; y < SPECTOGRAM_HEIGHT; y++) - memcpy(data + (SPECTOGRAM_START + y) * linesize, s->spectogram[(s->spectogram_index + y) % SPECTOGRAM_HEIGHT], VIDEO_WIDTH * 3); - } + int total_length = linesize * spectogram_height; + int back_length = linesize * s->spectogram_index; + data += spectogram_start * linesize; + memcpy(data, s->spectogram + s->spectogram_index*linesize, total_length - back_length); + + data += total_length - back_length; + if (back_length) + memcpy(data, s->spectogram, back_length); s->outpicref->pts = s->frame_count; ret = ff_filter_frame(outlink, av_frame_clone(s->outpicref)); @@ -439,7 +463,7 @@ static int plot_cqt(AVFilterLink *inlink) s->frame_count++; } s->spectogram_count = (s->spectogram_count + 1) % s->count; - s->spectogram_index = (s->spectogram_index + SPECTOGRAM_HEIGHT - 1) % SPECTOGRAM_HEIGHT; + s->spectogram_index = (s->spectogram_index + spectogram_height - 1) % spectogram_height; return ret; }