July 2021 - ffmpeg-cvslog - lists.ffmpeg.org

libavcodec/mpeg12dec.c: Switch to init_get_bits8 and checks return value
by Thierry Foucu 28 Jul '21

28 Jul '21

ffmpeg | branch: master | Thierry Foucu <tfoucu(a)gmail.com> | Tue Jul 27 11:25:04 2021 -0700| [fc300613bec1e55974d45f955265b72cbf9b5683] | committer: Michael Niedermayer libavcodec/mpeg12dec.c: Switch to init_get_bits8 and checks return value Signed-off-by: Michael Niedermayer <michael(a)niedermayer.cc> > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=fc300613bec1e5597… --- libavcodec/mpeg12dec.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) diff --git a/libavcodec/mpeg12dec.c b/libavcodec/mpeg12dec.c index b27ed5bd6d..858dca660c 100644 --- a/libavcodec/mpeg12dec.c +++ b/libavcodec/mpeg12dec.c @@ -1340,9 +1340,11 @@ static int mpeg1_decode_picture(AVCodecContext *avctx, const uint8_t *buf, { Mpeg1Context *s1 = avctx->priv_data; MpegEncContext *s = &s1->mpeg_enc_ctx; - int ref, f_code, vbv_delay; + int ref, f_code, vbv_delay, ret; - init_get_bits(&s->gb, buf, buf_size * 8); + ret = init_get_bits8(&s->gb, buf, buf_size); + if (ret < 0) + return ret; ref = get_bits(&s->gb, 10); /* temporal ref */ s->pict_type = get_bits(&s->gb, 3);

1 0

avfilter/avf_showcqt: switch to TX FFT from avutil
by Paul B Mahol 27 Jul '21

27 Jul '21

ffmpeg | branch: master | Paul B Mahol <onemda(a)gmail.com> | Tue Jul 27 21:16:28 2021 +0200| [0068b3d0f0bc06b2b083eb729bf84f1a3196a2a9] | committer: Paul B Mahol avfilter/avf_showcqt: switch to TX FFT from avutil > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=0068b3d0f0bc06b2b… --- configure | 5 ++--- libavfilter/avf_showcqt.c | 33 +++++++++++++++++---------------- libavfilter/avf_showcqt.h | 16 +++++++++------- libavfilter/x86/avf_showcqt_init.c | 2 +- 4 files changed, 29 insertions(+), 27 deletions(-) diff --git a/configure b/configure index d0da55fbd0..b61a189b07 100755 --- a/configure +++ b/configure @@ -3652,9 +3652,8 @@ scale_qsv_filter_deps="libmfx" scdet_filter_select="scene_sad" select_filter_select="scene_sad" sharpness_vaapi_filter_deps="vaapi" -showcqt_filter_deps="avcodec avformat swscale" +showcqt_filter_deps="avformat swscale" showcqt_filter_suggest="libfontconfig libfreetype" -showcqt_filter_select="fft" showfreqs_filter_deps="avcodec" showfreqs_filter_select="fft" showspatial_filter_deps="avcodec" @@ -7281,7 +7280,7 @@ enabled removelogo_filter && prepend avfilter_deps "avformat avcodec swscale" enabled sab_filter && prepend avfilter_deps "swscale" enabled scale_filter && prepend avfilter_deps "swscale" enabled scale2ref_filter && prepend avfilter_deps "swscale" -enabled showcqt_filter && prepend avfilter_deps "avformat avcodec swscale" +enabled showcqt_filter && prepend avfilter_deps "avformat swscale" enabled showfreqs_filter && prepend avfilter_deps "avcodec" enabled signature_filter && prepend avfilter_deps "avcodec avformat" enabled smartblur_filter && prepend avfilter_deps "swscale" diff --git a/libavfilter/avf_showcqt.c b/libavfilter/avf_showcqt.c index f1d8f16b32..aa9900ebe0 100644 --- a/libavfilter/avf_showcqt.c +++ b/libavfilter/avf_showcqt.c @@ -19,7 +19,7 @@ */ #include "config.h" -#include "libavcodec/avfft.h" +#include "libavutil/tx.h" #include "libavutil/channel_layout.h" #include "libavutil/opt.h" #include "libavutil/xga_font_data.h" @@ -144,13 +144,13 @@ static void common_uninit(ShowCQTContext *s) av_frame_free(&s->axis_frame); av_frame_free(&s->sono_frame); - av_fft_end(s->fft_ctx); - s->fft_ctx = NULL; + av_tx_uninit(&s->fft_ctx); if (s->coeffs) for (k = 0; k < s->cqt_len; k++) av_freep(&s->coeffs[k].val); av_freep(&s->coeffs); av_freep(&s->fft_data); + av_freep(&s->fft_input); av_freep(&s->fft_result); av_freep(&s->cqt_result); av_freep(&s->attack_data); @@ -267,15 +267,15 @@ error: return ret; } -static void cqt_calc(FFTComplex *dst, const FFTComplex *src, const Coeffs *coeffs, +static void cqt_calc(AVComplexFloat *dst, const AVComplexFloat *src, const Coeffs *coeffs, int len, int fft_len) { int k, x, i, j; for (k = 0; k < len; k++) { - FFTComplex l, r, a = {0,0}, b = {0,0}; + AVComplexFloat l, r, a = {0,0}, b = {0,0}; for (x = 0; x < coeffs[k].len; x++) { - FFTSample u = coeffs[k].val[x]; + float u = coeffs[k].val[x]; i = coeffs[k].start + x; j = fft_len - i; a.re += u * src[i].re; @@ -730,7 +730,7 @@ static float calculate_gamma(float v, float g) return expf(logf(v) / g); } -static void rgb_from_cqt(ColorFloat *c, const FFTComplex *v, float g, int len, float cscheme[6]) +static void rgb_from_cqt(ColorFloat *c, const AVComplexFloat *v, float g, int len, float cscheme[6]) { int x; for (x = 0; x < len; x++) { @@ -740,7 +740,7 @@ static void rgb_from_cqt(ColorFloat *c, const FFTComplex *v, float g, int len, f } } -static void yuv_from_cqt(ColorFloat *c, const FFTComplex *v, float gamma, int len, float cm[3][3], float cscheme[6]) +static void yuv_from_cqt(ColorFloat *c, const AVComplexFloat *v, float gamma, int len, float cm[3][3], float cscheme[6]) { int x; for (x = 0; x < len; x++) { @@ -1110,7 +1110,7 @@ static void process_cqt(ShowCQTContext *s) if (s->fcount > 1) { float rcp_fcount = 1.0f / s->fcount; for (x = 0; x < s->width; x++) { - FFTComplex result = {0.0f, 0.0f}; + AVComplexFloat result = {0.0f, 0.0f}; for (i = 0; i < s->fcount; i++) { result.re += s->cqt_result[s->fcount * x + i].re; result.im += s->cqt_result[s->fcount * x + i].im; @@ -1139,17 +1139,16 @@ static int plot_cqt(AVFilterContext *ctx, AVFrame **frameout) last_time = av_gettime_relative(); - memcpy(s->fft_result, s->fft_data, s->fft_len * sizeof(*s->fft_data)); + memcpy(s->fft_input, s->fft_data, s->fft_len * sizeof(*s->fft_data)); if (s->attack_data) { int k; for (k = 0; k < s->remaining_fill_max; k++) { - s->fft_result[s->fft_len/2+k].re *= s->attack_data[k]; - s->fft_result[s->fft_len/2+k].im *= s->attack_data[k]; + s->fft_input[s->fft_len/2+k].re *= s->attack_data[k]; + s->fft_input[s->fft_len/2+k].im *= s->attack_data[k]; } } - av_fft_permute(s->fft_ctx, s->fft_result); - av_fft_calc(s->fft_ctx, s->fft_result); + s->tx_fn(s->fft_ctx, s->fft_result, s->fft_input, sizeof(float)); s->fft_result[s->fft_len] = s->fft_result[0]; UPDATE_TIME(s->fft_time); @@ -1355,6 +1354,7 @@ static int config_output(AVFilterLink *outlink) AVFilterContext *ctx = outlink->src; AVFilterLink *inlink = ctx->inputs[0]; ShowCQTContext *s = ctx->priv; + float scale; int ret; common_uninit(s); @@ -1380,9 +1380,10 @@ static int config_output(AVFilterLink *outlink) s->fft_len = 1 << s->fft_bits; av_log(ctx, AV_LOG_INFO, "fft_len = %d, cqt_len = %d.\n", s->fft_len, s->cqt_len); - s->fft_ctx = av_fft_init(s->fft_bits, 0); + ret = av_tx_init(&s->fft_ctx, &s->tx_fn, AV_TX_FLOAT_FFT, 0, s->fft_len, &scale, 0); s->fft_data = av_calloc(s->fft_len, sizeof(*s->fft_data)); - s->fft_result = av_calloc(s->fft_len + 64, sizeof(*s->fft_result)); + s->fft_input = av_calloc(FFALIGN(s->fft_len + 64, 256), sizeof(*s->fft_input)); + s->fft_result = av_calloc(FFALIGN(s->fft_len + 64, 256), sizeof(*s->fft_result)); s->cqt_result = av_malloc_array(s->cqt_len, sizeof(*s->cqt_result)); if (!s->fft_ctx || !s->fft_data || !s->fft_result || !s->cqt_result) return AVERROR(ENOMEM); diff --git a/libavfilter/avf_showcqt.h b/libavfilter/avf_showcqt.h index 0cef5a2003..b6c1e0e108 100644 --- a/libavfilter/avf_showcqt.h +++ b/libavfilter/avf_showcqt.h @@ -21,12 +21,12 @@ #ifndef AVFILTER_SHOWCQT_H #define AVFILTER_SHOWCQT_H -#include "libavcodec/avfft.h" +#include "libavutil/tx.h" #include "avfilter.h" #include "internal.h" typedef struct Coeffs { - FFTSample *val; + float *val; int start, len; } Coeffs; @@ -58,11 +58,13 @@ typedef struct ShowCQTContext { int remaining_fill_max; int64_t next_pts; double *freq; - FFTContext *fft_ctx; + AVTXContext *fft_ctx; + av_tx_fn tx_fn; Coeffs *coeffs; - FFTComplex *fft_data; - FFTComplex *fft_result; - FFTComplex *cqt_result; + AVComplexFloat *fft_data; + AVComplexFloat *fft_input; + AVComplexFloat *fft_result; + AVComplexFloat *cqt_result; float *attack_data; int fft_bits; int fft_len; @@ -76,7 +78,7 @@ typedef struct ShowCQTContext { float cmatrix[3][3]; float cscheme_v[6]; /* callback */ - void (*cqt_calc)(FFTComplex *dst, const FFTComplex *src, const Coeffs *coeffs, + void (*cqt_calc)(AVComplexFloat *dst, const AVComplexFloat *src, const Coeffs *coeffs, int len, int fft_len); void (*permute_coeffs)(float *v, int len); void (*draw_bar)(AVFrame *out, const float *h, const float *rcp_h, diff --git a/libavfilter/x86/avf_showcqt_init.c b/libavfilter/x86/avf_showcqt_init.c index 0cc164c352..2320b30239 100644 --- a/libavfilter/x86/avf_showcqt_init.c +++ b/libavfilter/x86/avf_showcqt_init.c @@ -24,7 +24,7 @@ #include "libavfilter/avf_showcqt.h" #define DECLARE_CQT_CALC(type) \ -void ff_showcqt_cqt_calc_##type(FFTComplex *dst, const FFTComplex *src, \ +void ff_showcqt_cqt_calc_##type(AVComplexFloat *dst, const AVComplexFloat *src, \ const Coeffs *coeffs, int len, int fft_len) DECLARE_CQT_CALC(sse);

1 0

avfilter/af_afftdn: switch to TX FFT from avutil
by Paul B Mahol 27 Jul '21

27 Jul '21

ffmpeg | branch: master | Paul B Mahol <onemda(a)gmail.com> | Tue Jul 27 20:48:40 2021 +0200| [925d41ebd47bb21a55de4cc1668277076349ce0e] | committer: Paul B Mahol avfilter/af_afftdn: switch to TX FFT from avutil > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=925d41ebd47bb21a5… --- configure | 3 -- libavfilter/af_afftdn.c | 89 ++++++++++++++++++++++++++----------------------- 2 files changed, 47 insertions(+), 45 deletions(-) diff --git a/configure b/configure index 646d16e3c9..d0da55fbd0 100755 --- a/configure +++ b/configure @@ -3542,8 +3542,6 @@ libzmq_protocol_deps="libzmq" libzmq_protocol_select="network" # filters -afftdn_filter_deps="avcodec" -afftdn_filter_select="fft" afftfilt_filter_deps="avcodec" afftfilt_filter_select="fft" afir_filter_deps="avcodec" @@ -7261,7 +7259,6 @@ done enabled zlib && add_cppflags -DZLIB_CONST # conditional library dependencies, in any order -enabled afftdn_filter && prepend avfilter_deps "avcodec" enabled afftfilt_filter && prepend avfilter_deps "avcodec" enabled afir_filter && prepend avfilter_deps "avcodec" enabled amovie_filter && prepend avfilter_deps "avformat avcodec" diff --git a/libavfilter/af_afftdn.c b/libavfilter/af_afftdn.c index 064009ffd5..693585439b 100644 --- a/libavfilter/af_afftdn.c +++ b/libavfilter/af_afftdn.c @@ -24,7 +24,7 @@ #include "libavutil/avstring.h" #include "libavutil/channel_layout.h" #include "libavutil/opt.h" -#include "libavcodec/avfft.h" +#include "libavutil/tx.h" #include "avfilter.h" #include "audio.h" #include "formats.h" @@ -66,8 +66,10 @@ typedef struct DeNoiseChannel { double *abs_var; double *rel_var; double *min_abs_var; - FFTComplex *fft_data; - FFTContext *fft, *ifft; + AVComplexFloat *fft_in; + AVComplexFloat *fft_out; + AVTXContext *fft, *ifft; + av_tx_fn tx_fn, itx_fn; double noise_band_norm[15]; double noise_band_avr[15]; @@ -290,7 +292,7 @@ static double limit_gain(double a, double b) } static void process_frame(AudioFFTDeNoiseContext *s, DeNoiseChannel *dnch, - FFTComplex *fft_data, + AVComplexFloat *fft_data, double *prior, double *prior_band_excit, int track_noise) { double d1, d2, d3, gain; @@ -612,7 +614,7 @@ static int config_input(AVFilterLink *inlink) AVFilterContext *ctx = inlink->dst; AudioFFTDeNoiseContext *s = ctx->priv; double wscale, sar, sum, sdiv; - int i, j, k, m, n; + int i, j, k, m, n, ret; s->dnch = av_calloc(inlink->channels, sizeof(*s->dnch)); if (!s->dnch) @@ -680,6 +682,7 @@ static int config_input(AVFilterLink *inlink) for (int ch = 0; ch < inlink->channels; ch++) { DeNoiseChannel *dnch = &s->dnch[ch]; + float scale; switch (s->noise_type) { case WHITE_NOISE: @@ -727,9 +730,14 @@ static int config_input(AVFilterLink *inlink) dnch->abs_var = av_calloc(s->bin_count, sizeof(*dnch->abs_var)); dnch->rel_var = av_calloc(s->bin_count, sizeof(*dnch->rel_var)); dnch->min_abs_var = av_calloc(s->bin_count, sizeof(*dnch->min_abs_var)); - dnch->fft_data = av_calloc(s->fft_length2 + 1, sizeof(*dnch->fft_data)); - dnch->fft = av_fft_init(av_log2(s->fft_length2), 0); - dnch->ifft = av_fft_init(av_log2(s->fft_length2), 1); + dnch->fft_in = av_calloc(s->fft_length2 + 1, sizeof(*dnch->fft_in)); + dnch->fft_out = av_calloc(s->fft_length2 + 1, sizeof(*dnch->fft_out)); + ret = av_tx_init(&dnch->fft, &dnch->tx_fn, AV_TX_FLOAT_FFT, 0, s->fft_length2, &scale, 0); + if (ret < 0) + return ret; + ret = av_tx_init(&dnch->ifft, &dnch->itx_fn, AV_TX_FLOAT_FFT, 1, s->fft_length2, &scale, 0); + if (ret < 0) + return ret; dnch->spread_function = av_calloc(s->number_of_bands * s->number_of_bands, sizeof(*dnch->spread_function)); @@ -742,7 +750,8 @@ static int config_input(AVFilterLink *inlink) !dnch->clean_data || !dnch->noisy_data || !dnch->out_samples || - !dnch->fft_data || + !dnch->fft_in || + !dnch->fft_out || !dnch->abs_var || !dnch->rel_var || !dnch->min_abs_var || @@ -855,7 +864,7 @@ static int config_input(AVFilterLink *inlink) return 0; } -static void preprocess(FFTComplex *in, int len) +static void preprocess(AVComplexFloat *in, int len) { double d1, d2, d3, d4, d5, d6, d7, d8, d9, d10; int n, i, k; @@ -888,7 +897,7 @@ static void preprocess(FFTComplex *in, int len) in[0].im = d2 - in[0].im; } -static void postprocess(FFTComplex *in, int len) +static void postprocess(AVComplexFloat *in, int len) { double d1, d2, d3, d4, d5, d6, d7, d8, d9, d10; int n, i, k; @@ -938,28 +947,27 @@ static void sample_noise_block(AudioFFTDeNoiseContext *s, int edge, j, k, n, edgemax; for (int i = 0; i < s->window_length; i++) { - dnch->fft_data[i].re = s->window[i] * src[i] * (1LL << 24); - dnch->fft_data[i].im = 0.0; + dnch->fft_in[i].re = s->window[i] * src[i] * (1LL << 24); + dnch->fft_in[i].im = 0.0; } for (int i = s->window_length; i < s->fft_length2; i++) { - dnch->fft_data[i].re = 0.0; - dnch->fft_data[i].im = 0.0; + dnch->fft_in[i].re = 0.0; + dnch->fft_in[i].im = 0.0; } - av_fft_permute(dnch->fft, dnch->fft_data); - av_fft_calc(dnch->fft, dnch->fft_data); + dnch->tx_fn(dnch->fft, dnch->fft_out, dnch->fft_in, sizeof(float)); - preprocess(dnch->fft_data, s->fft_length); + preprocess(dnch->fft_out, s->fft_length); edge = s->noise_band_edge[0]; j = edge; k = 0; n = j; edgemax = fmin(s->fft_length2, s->noise_band_edge[15]); - dnch->fft_data[s->fft_length2].re = dnch->fft_data[0].im; - dnch->fft_data[0].im = 0.0; - dnch->fft_data[s->fft_length2].im = 0.0; + dnch->fft_out[s->fft_length2].re = dnch->fft_out[0].im; + dnch->fft_out[0].im = 0.0; + dnch->fft_out[s->fft_length2].im = 0.0; for (int i = j; i <= edgemax; i++) { if ((i == j) && (i < edgemax)) { @@ -979,10 +987,10 @@ static void sample_noise_block(AudioFFTDeNoiseContext *s, avr = 0.0; avi = 0.0; } - avr += dnch->fft_data[n].re; - avi += dnch->fft_data[n].im; - mag2 = dnch->fft_data[n].re * dnch->fft_data[n].re + - dnch->fft_data[n].im * dnch->fft_data[n].im; + avr += dnch->fft_out[n].re; + avi += dnch->fft_out[n].im; + mag2 = dnch->fft_out[n].re * dnch->fft_out[n].re + + dnch->fft_out[n].im * dnch->fft_out[n].im; mag2 = fmax(mag2, s->sample_floor); @@ -1107,30 +1115,28 @@ static int filter_channel(AVFilterContext *ctx, void *arg, int jobnr, int nb_job } for (int m = 0; m < s->window_length; m++) { - dnch->fft_data[m].re = s->window[m] * src[m] * (1LL << 24); - dnch->fft_data[m].im = 0; + dnch->fft_in[m].re = s->window[m] * src[m] * (1LL << 24); + dnch->fft_in[m].im = 0; } for (int m = s->window_length; m < s->fft_length2; m++) { - dnch->fft_data[m].re = 0; - dnch->fft_data[m].im = 0; + dnch->fft_in[m].re = 0; + dnch->fft_in[m].im = 0; } - av_fft_permute(dnch->fft, dnch->fft_data); - av_fft_calc(dnch->fft, dnch->fft_data); + dnch->tx_fn(dnch->fft, dnch->fft_out, dnch->fft_in, sizeof(float)); - preprocess(dnch->fft_data, s->fft_length); - process_frame(s, dnch, dnch->fft_data, + preprocess(dnch->fft_out, s->fft_length); + process_frame(s, dnch, dnch->fft_out, dnch->prior, dnch->prior_band_excit, s->track_noise); - postprocess(dnch->fft_data, s->fft_length); + postprocess(dnch->fft_out, s->fft_length); - av_fft_permute(dnch->ifft, dnch->fft_data); - av_fft_calc(dnch->ifft, dnch->fft_data); + dnch->itx_fn(dnch->ifft, dnch->fft_in, dnch->fft_out, sizeof(float)); for (int m = 0; m < s->window_length; m++) - dst[m] += s->window[m] * dnch->fft_data[m].re / (1LL << 24); + dst[m] += s->window[m] * dnch->fft_in[m].re / (1LL << 24); } return 0; @@ -1330,11 +1336,10 @@ static av_cold void uninit(AVFilterContext *ctx) av_freep(&dnch->abs_var); av_freep(&dnch->rel_var); av_freep(&dnch->min_abs_var); - av_freep(&dnch->fft_data); - av_fft_end(dnch->fft); - dnch->fft = NULL; - av_fft_end(dnch->ifft); - dnch->ifft = NULL; + av_freep(&dnch->fft_in); + av_freep(&dnch->fft_out); + av_tx_uninit(&dnch->fft); + av_tx_uninit(&dnch->ifft); } av_freep(&s->dnch); }

1 0

avfilter/avf_concat: fix EOF timestamp
by Paul B Mahol 27 Jul '21

27 Jul '21

ffmpeg | branch: master | Paul B Mahol <onemda(a)gmail.com> | Sun Jul 25 20:09:44 2021 +0200| [51c22f6621e57f2836048ec1779e2fe195786a9a] | committer: Paul B Mahol avfilter/avf_concat: fix EOF timestamp > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=51c22f6621e57f283… --- libavfilter/avf_concat.c | 9 +++++++-- 1 file changed, 7 insertions(+), 2 deletions(-) diff --git a/libavfilter/avf_concat.c b/libavfilter/avf_concat.c index cb46f52a04..333a0b090c 100644 --- a/libavfilter/avf_concat.c +++ b/libavfilter/avf_concat.c @@ -397,12 +397,17 @@ static int activate(AVFilterContext *ctx) /* Forward status change */ if (cat->cur_idx < ctx->nb_inputs) { for (i = 0; i < ctx->nb_outputs; i++) { - ret = ff_inlink_acknowledge_status(ctx->inputs[cat->cur_idx + i], &status, &pts); + AVFilterLink *inlink = ctx->inputs[cat->cur_idx + i]; + + ret = ff_inlink_acknowledge_status(inlink, &status, &pts); /* TODO use pts */ if (ret > 0) { close_input(ctx, cat->cur_idx + i); if (cat->cur_idx + ctx->nb_outputs >= ctx->nb_inputs) { - ff_outlink_set_status(ctx->outputs[i], status, pts); + int64_t eof_pts = cat->delta_ts; + + eof_pts += av_rescale_q(pts, inlink->time_base, ctx->outputs[i]->time_base); + ff_outlink_set_status(ctx->outputs[i], status, eof_pts); } if (!cat->nb_in_active) { ret = flush_segment(ctx);

1 0

qsvenc: Value '0' is not a valid value for parameter GopOptFlag
by Haihao Xiang 27 Jul '21

27 Jul '21

ffmpeg | branch: master | Haihao Xiang <haihao.xiang(a)intel.com> | Thu Nov 5 15:08:23 2020 +0800| [f2feb31b78be9f862867d567f12758b635b32b74] | committer: Zhong Li qsvenc: Value '0' is not a valid value for parameter GopOptFlag The accepted values for GopOptFlag are MFX_GOP_CLOSED (1) and MFX_GOP_STRICT (2). Signed-off-by: Haihao Xiang <haihao.xiang(a)intel.com> Signed-off-by: Zhong Li <zhongli_dev(a)126.com> > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=f2feb31b78be9f862… --- libavcodec/qsvenc.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/libavcodec/qsvenc.c b/libavcodec/qsvenc.c index b9a922d6b9..f6fa759447 100644 --- a/libavcodec/qsvenc.c +++ b/libavcodec/qsvenc.c @@ -531,7 +531,7 @@ static int init_video_param(AVCodecContext *avctx, QSVEncContext *q) q->param.mfx.GopPicSize = FFMAX(0, avctx->gop_size); q->param.mfx.GopRefDist = FFMAX(-1, avctx->max_b_frames) + 1; q->param.mfx.GopOptFlag = avctx->flags & AV_CODEC_FLAG_CLOSED_GOP ? - MFX_GOP_CLOSED : 0; + MFX_GOP_CLOSED : MFX_GOP_STRICT; q->param.mfx.IdrInterval = q->idr_interval; q->param.mfx.NumSlice = avctx->slices; q->param.mfx.NumRefFrame = FFMAX(0, avctx->refs);

1 0

avformat/concat: check the terminating character returned by av_get_token()
by James Almer 27 Jul '21

27 Jul '21

ffmpeg | branch: master | James Almer <jamrial(a)gmail.com> | Tue Jul 27 12:00:03 2021 -0300| [487dfa9f04bb825669441476ffd683a7783d5ce9] | committer: James Almer avformat/concat: check the terminating character returned by av_get_token() Don't attempt to increase the cursor pointer if it was \0. Fixes invalid reads. Reviewed-by: Marton Balint <cus(a)passwd.hu> Reviewed-by: Nicolas George <george(a)nsup.org> Signed-off-by: James Almer <jamrial(a)gmail.com> > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=487dfa9f04bb82566… --- libavformat/concat.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/libavformat/concat.c b/libavformat/concat.c index aec1f52d8e..2560811ff9 100644 --- a/libavformat/concat.c +++ b/libavformat/concat.c @@ -251,7 +251,8 @@ static av_cold int concatf_open(URLContext *h, const char *uri, int flags) err = AVERROR(ENOMEM); break; } - cursor++; + if (*cursor) + cursor++; if (++len == SIZE_MAX / sizeof(*nodes)) { av_free(node_uri);

1 0

avfilter/vf_pseudocolor: add three more presets
by Paul B Mahol 26 Jul '21

26 Jul '21

ffmpeg | branch: master | Paul B Mahol <onemda(a)gmail.com> | Mon Jul 26 22:49:02 2021 +0200| [bc5faacc3f38484e3e9ee2f0143d0ad6884ec747] | committer: Paul B Mahol avfilter/vf_pseudocolor: add three more presets > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=bc5faacc3f38484e3… --- doc/filters.texi | 3 ++ libavfilter/vf_pseudocolor.c | 111 +++++++++++++++++++++++-------------------- 2 files changed, 62 insertions(+), 52 deletions(-) diff --git a/doc/filters.texi b/doc/filters.texi index 60b8b2886c..66c0f87e47 100644 --- a/doc/filters.texi +++ b/doc/filters.texi @@ -16838,6 +16838,9 @@ Available LUTs: @item shadows @item highlights @item solar +@item nominal +@item preferred +@item total @end table @item opacity diff --git a/libavfilter/vf_pseudocolor.c b/libavfilter/vf_pseudocolor.c index 5440dce536..ec14a0c363 100644 --- a/libavfilter/vf_pseudocolor.c +++ b/libavfilter/vf_pseudocolor.c @@ -82,6 +82,9 @@ enum Presets { PRESET_SHADOWS, PRESET_HIGHLIGHTS, PRESET_SOLAR, + PRESET_NOMINAL, + PRESET_PREFERRED, + PRESET_TOTAL, NB_PRESETS, }; @@ -98,7 +101,7 @@ typedef struct Fill { } Fill; typedef struct Range { - int start, end; + AVRational start, end; } Range; typedef struct Preset { @@ -108,11 +111,14 @@ typedef struct Preset { const Fill *fills; } Preset; -static const Range full_range = {0, 256}; -static const Range spec1_range[] = {{0, 16}, {16, 236}, {236, 256}}; -static const Range spec2_range[] = {{0, 16}, {16, 22}, {22, 226}, {226, 236}, {236, 256}}; -static const Range shadows_range[] = {{0, 32}, {32, 256}}; -static const Range highlights_range[] = {{0, 214}, {214, 224}, {224, 256}}; +static const Range full_range = {{0, 1}, {1, 1}}; +static const Range nominal_range[] = {{{0, 1}, {4096, 65536}}, {{4096, 65536}, {60161, 65536}}, {{60161, 65536}, {1, 1}}}; +static const Range preferred_range[] = {{{0, 1}, {1280, 65536}}, {{1280, 65536}, {62977, 65536}}, {{62977, 65536}, {1, 1}}}; +static const Range total_range[] = {{{0, 1}, {256, 65536}}, {{256, 65536}, {65280, 65536}}, {{65280, 65536}, {1, 1}}}; +static const Range spec1_range[] = {{{0, 1}, {16, 256}}, {{16, 256}, {236, 256}}, {{236, 256}, {256, 256}}}; +static const Range spec2_range[] = {{{0, 1}, {16, 256}}, {{16, 256}, {22, 256}}, {{22, 256}, {226, 256}}, {{226, 256}, {236, 256}}, {{236, 256}, {256, 256}}}; +static const Range shadows_range[] = {{{0, 1}, {32, 256}}, {{32, 256}, {256, 256}}}; +static const Range highlights_range[] = {{{0,1}, {214,256}}, {{214, 256}, {224, 256}}, {{224, 256}, {256, 256}}}; static const Fill spec1_fills[] = {{{0.5f, 0.f, .5f, 1.f}}, {{-1.f, -1.f, -1.f, 1.f}}, {{1.f, 0.f, 0.f, 1.f}}}; static const Fill spec2_fills[] = {{{0.5f, 0.f, .5f, 1.f}}, {{0.f, 1.f, 1.f, 1.f}}, {{-1.f, -1.f, -1.f, 1.f}}, {{1.f, 1.f, 0.f, 1.f}}, {{1.f, 0.f, 0.f, 1.f}}}; @@ -179,6 +185,9 @@ static const Preset presets[] = [PRESET_TURBO] = { 1, &full_range, &curves[TURBO], NULL }, [PRESET_CIVIDIS] = { 1, &full_range, &curves[CIVIDIS], NULL }, [PRESET_RANGE1] = { 3, spec1_range, NULL, spec1_fills }, + [PRESET_NOMINAL] = { 3, nominal_range, NULL, spec1_fills }, + [PRESET_PREFERRED]={ 3, preferred_range, NULL, spec1_fills }, + [PRESET_TOTAL] = { 3, total_range, NULL, spec1_fills }, [PRESET_RANGE2] = { 5, spec2_range, NULL, spec2_fills }, [PRESET_SHADOWS] = { 2, shadows_range, NULL, shadows_fills }, [PRESET_HIGHLIGHTS] = { 3, highlights_range, NULL, highlights_fills }, @@ -234,6 +243,9 @@ static const AVOption pseudocolor_options[] = { { "shadows", NULL, 0, AV_OPT_TYPE_CONST, {.i64=PRESET_SHADOWS}, .flags = FLAGS, "preset" }, { "highlights", NULL, 0, AV_OPT_TYPE_CONST, {.i64=PRESET_HIGHLIGHTS},.flags=FLAGS, "preset" }, { "solar", NULL, 0, AV_OPT_TYPE_CONST, {.i64=PRESET_SOLAR}, .flags=FLAGS, "preset" }, + { "nominal", NULL, 0, AV_OPT_TYPE_CONST, {.i64=PRESET_NOMINAL}, .flags=FLAGS, "preset" }, + { "preferred", NULL, 0, AV_OPT_TYPE_CONST, {.i64=PRESET_PREFERRED},.flags=FLAGS,"preset" }, + { "total", NULL, 0, AV_OPT_TYPE_CONST, {.i64=PRESET_TOTAL}, .flags=FLAGS, "preset" }, { "opacity", "set pseudocolor opacity",OFFSET(opacity), AV_OPT_TYPE_FLOAT, {.dbl=1}, 0, 1, .flags = FLAGS }, { NULL } }; @@ -574,11 +586,10 @@ static int config_input(AVFilterLink *inlink) AVFilterContext *ctx = inlink->dst; PseudoColorContext *s = ctx->priv; const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(inlink->format); - int depth, ret, hsub, vsub, color, factor, rgb; + int depth, ret, hsub, vsub, color, rgb; rgb = desc->flags & AV_PIX_FMT_FLAG_RGB; depth = desc->comp[0].depth; - factor = 1 << (depth - 8); s->max = (1 << depth) - 1; s->nb_planes = av_pix_fmt_count_planes(inlink->format); @@ -644,52 +655,25 @@ static int config_input(AVFilterLink *inlink) int nb_segments = presets[s->preset].nb_segments; for (int seg = 0; seg < nb_segments; seg++) { - int start = presets[s->preset].ranges[seg].start; - int end = presets[s->preset].ranges[seg].end; + AVRational rstart = presets[s->preset].ranges[seg].start; + AVRational rend = presets[s->preset].ranges[seg].end; + int start = av_rescale_rnd(s->max + 1, rstart.num, rstart.den, AV_ROUND_UP); + int end = av_rescale_rnd(s->max + 1, rend.num, rend.den, AV_ROUND_UP); for (int i = start; i < end; i++) { if (!presets[s->preset].curves) { const Fill fill = presets[s->preset].fills[seg]; + double r, g, b, a; - for (int j = 0; j < factor; j++) { - double r, g, b, a; + g = fill.fill[1]; + b = fill.fill[2]; + r = fill.fill[0]; + a = fill.fill[3]; - g = fill.fill[1]; - b = fill.fill[2]; - r = fill.fill[0]; - a = fill.fill[3]; - - if (g >= 0.f && b >= 0.f && r >= 0.f) { - g *= s->max; - b *= s->max; - r *= s->max; - - if (!rgb) { - double y = RGB_TO_Y_BT709(r, g, b); - double u = RGB_TO_U_BT709(r, g, b, s->max); - double v = RGB_TO_V_BT709(r, g, b, s->max); - - r = v; - g = y; - b = u; - } - } - - s->lut[0][i*factor+j] = g; - s->lut[1][i*factor+j] = b; - s->lut[2][i*factor+j] = r; - s->lut[3][i*factor+j] = a * s->max; - } - } else { - const Curve curve = presets[s->preset].curves[seg]; - - for (int j = 0; j < factor; j++) { - const double lf = j / (double)factor; - double r, g, b; - - g = poly_eval(curve.coef[1], i + lf + curve.offset[1], curve.fun[1]) * s->max; - b = poly_eval(curve.coef[2], i + lf + curve.offset[2], curve.fun[2]) * s->max; - r = poly_eval(curve.coef[0], i + lf + curve.offset[0], curve.fun[0]) * s->max; + if (g >= 0.f && b >= 0.f && r >= 0.f) { + g *= s->max; + b *= s->max; + r *= s->max; if (!rgb) { double y = RGB_TO_Y_BT709(r, g, b); @@ -700,12 +684,35 @@ static int config_input(AVFilterLink *inlink) g = y; b = u; } + } + + s->lut[0][i] = g; + s->lut[1][i] = b; + s->lut[2][i] = r; + s->lut[3][i] = a * s->max; + } else { + const Curve curve = presets[s->preset].curves[seg]; + const double lf = i / (double)s->max * 256.; + double r, g, b; + + g = poly_eval(curve.coef[1], lf + curve.offset[1], curve.fun[1]) * s->max; + b = poly_eval(curve.coef[2], lf + curve.offset[2], curve.fun[2]) * s->max; + r = poly_eval(curve.coef[0], lf + curve.offset[0], curve.fun[0]) * s->max; - s->lut[0][i*factor+j] = g; - s->lut[1][i*factor+j] = b; - s->lut[2][i*factor+j] = r; - s->lut[3][i*factor+j] = 1.f * s->max; + if (!rgb) { + double y = RGB_TO_Y_BT709(r, g, b); + double u = RGB_TO_U_BT709(r, g, b, s->max); + double v = RGB_TO_V_BT709(r, g, b, s->max); + + r = v; + g = y; + b = u; } + + s->lut[0][i] = g; + s->lut[1][i] = b; + s->lut[2][i] = r; + s->lut[3][i] = 1.f * s->max; } } }

1 0

avcodec/h264_sei: honor the Film Grain Characteristics cancel flag
by James Almer 26 Jul '21

26 Jul '21

ffmpeg | branch: master | James Almer <jamrial(a)gmail.com> | Sun Jul 25 23:09:00 2021 -0300| [f7958d0883286479a6c9cd91e68e506dfa02ed7c] | committer: James Almer avcodec/h264_sei: honor the Film Grain Characteristics cancel flag Film Grain persistence is cancelled when it's true. Signed-off-by: James Almer <jamrial(a)gmail.com> > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=f7958d0883286479a… --- libavcodec/h264_sei.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/libavcodec/h264_sei.c b/libavcodec/h264_sei.c index b1785c6198..7797f24650 100644 --- a/libavcodec/h264_sei.c +++ b/libavcodec/h264_sei.c @@ -420,9 +420,9 @@ static int decode_alternative_transfer(H264SEIAlternativeTransfer *h, static int decode_film_grain_characteristics(H264SEIFilmGrainCharacteristics *h, GetBitContext *gb) { - int film_grain_characteristics_cancel_flag = get_bits1(gb); + h->present = !get_bits1(gb); // film_grain_characteristics_cancel_flag - if (!film_grain_characteristics_cancel_flag) { + if (h->present) { memset(h, 0, sizeof(*h)); h->model_id = get_bits(gb, 8); h->separate_colour_description_present_flag = get_bits1(gb);

1 0

avfilter/af_silenceremove: make window also depends on input sample format
by Paul B Mahol 25 Jul '21

25 Jul '21

ffmpeg | branch: master | Paul B Mahol <onemda(a)gmail.com> | Sun Jul 25 21:50:31 2021 +0200| [da6975f0222a2db4cfbc1c17fe6e32c37f6a88fe] | committer: Paul B Mahol avfilter/af_silenceremove: make window also depends on input sample format > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=da6975f0222a2db4c… --- libavfilter/af_silenceremove.c | 203 ++++++++++++++++++++++------------------- 1 file changed, 108 insertions(+), 95 deletions(-) diff --git a/libavfilter/af_silenceremove.c b/libavfilter/af_silenceremove.c index 2a6f659e2d..9d6d683688 100644 --- a/libavfilter/af_silenceremove.c +++ b/libavfilter/af_silenceremove.c @@ -86,9 +86,8 @@ typedef struct SilenceRemoveContext { int stop_found_periods; double window_ratio; - double *window; - double *window_current; - double *window_end; + AVFrame *window; + int window_offset; int window_size; double sum; @@ -140,9 +139,9 @@ static void copy_double(SilenceRemoveContext *s, AVFrame *out, AVFrame *in, static void copy_doublep(SilenceRemoveContext *s, AVFrame *out, AVFrame *in, int ch, int out_offset, int in_offset) { - const double *srcp = (const double *)in->data[ch]; + const double *srcp = (const double *)in->extended_data[ch]; const double src = srcp[in_offset]; - double *dstp = (double *)out->data[ch]; + double *dstp = (double *)out->extended_data[ch]; dstp[out_offset] = src; } @@ -160,9 +159,9 @@ static void copy_float(SilenceRemoveContext *s, AVFrame *out, AVFrame *in, static void copy_floatp(SilenceRemoveContext *s, AVFrame *out, AVFrame *in, int ch, int out_offset, int in_offset) { - const float *srcp = (const float *)in->data[ch]; + const float *srcp = (const float *)in->extended_data[ch]; const float src = srcp[in_offset]; - float *dstp = (float *)out->data[ch]; + float *dstp = (float *)out->extended_data[ch]; dstp[out_offset] = src; } @@ -170,11 +169,13 @@ static void copy_floatp(SilenceRemoveContext *s, AVFrame *out, AVFrame *in, static double compute_peak_double(SilenceRemoveContext *s, AVFrame *frame, int ch, int offset) { const double *samples = (const double *)frame->data[0]; + const double *wsamples = (const double *)s->window->data[0]; double sample = samples[frame->channels * offset + ch]; + double wsample = wsamples[frame->channels * s->window_offset + ch]; double new_sum; new_sum = s->sum; - new_sum -= *s->window_current; + new_sum -= wsample; new_sum += fabs(sample); return new_sum / s->window_size; @@ -183,26 +184,26 @@ static double compute_peak_double(SilenceRemoveContext *s, AVFrame *frame, int c static void update_peak_double(SilenceRemoveContext *s, AVFrame *frame, int ch, int offset) { const double *samples = (const double *)frame->data[0]; + double *wsamples = (double *)s->window->data[0]; double sample = samples[frame->channels * offset + ch]; + double *wsample = &wsamples[frame->channels * s->window_offset + ch]; - s->sum -= *s->window_current; - *s->window_current = fabs(sample); - s->sum += *s->window_current; - - s->window_current++; - if (s->window_current >= s->window_end) - s->window_current = s->window; + s->sum -= *wsample; + *wsample = fabs(sample); + s->sum += *wsample; } static double compute_peak_float(SilenceRemoveContext *s, AVFrame *frame, int ch, int offset) { const float *samples = (const float *)frame->data[0]; + const float *wsamples = (const float *)s->window->data[0]; float sample = samples[frame->channels * offset + ch]; + float wsample = wsamples[frame->channels * s->window_offset + ch]; float new_sum; new_sum = s->sum; - new_sum -= *s->window_current; - new_sum += fabs(sample); + new_sum -= wsample; + new_sum += fabsf(sample); return new_sum / s->window_size; } @@ -210,25 +211,25 @@ static double compute_peak_float(SilenceRemoveContext *s, AVFrame *frame, int ch static void update_peak_float(SilenceRemoveContext *s, AVFrame *frame, int ch, int offset) { const float *samples = (const float *)frame->data[0]; + float *wsamples = (float *)s->window->data[0]; float sample = samples[frame->channels * offset + ch]; + float *wsample = &wsamples[frame->channels * s->window_offset + ch]; - s->sum -= *s->window_current; - *s->window_current = fabs(sample); - s->sum += *s->window_current; - - s->window_current++; - if (s->window_current >= s->window_end) - s->window_current = s->window; + s->sum -= *wsample; + *wsample = fabsf(sample); + s->sum += *wsample; } static double compute_rms_double(SilenceRemoveContext *s, AVFrame *frame, int ch, int offset) { const double *samples = (const double *)frame->data[0]; + const double *wsamples = (const double *)s->window->data[0]; double sample = samples[frame->channels * offset + ch]; + double wsample = wsamples[frame->channels * s->window_offset + ch]; double new_sum; new_sum = s->sum; - new_sum -= *s->window_current; + new_sum -= wsample; new_sum += sample * sample; return sqrt(new_sum / s->window_size); @@ -237,25 +238,25 @@ static double compute_rms_double(SilenceRemoveContext *s, AVFrame *frame, int ch static void update_rms_double(SilenceRemoveContext *s, AVFrame *frame, int ch, int offset) { const double *samples = (const double *)frame->data[0]; + double *wsamples = (double *)s->window->data[0]; double sample = samples[frame->channels * offset + ch]; + double *wsample = &wsamples[frame->channels * s->window_offset + ch]; - s->sum -= *s->window_current; - *s->window_current = sample * sample; - s->sum += *s->window_current; - - s->window_current++; - if (s->window_current >= s->window_end) - s->window_current = s->window; + s->sum -= *wsample; + *wsample = sample * sample; + s->sum += *wsample; } static double compute_rms_float(SilenceRemoveContext *s, AVFrame *frame, int ch, int offset) { const float *samples = (const float *)frame->data[0]; + const float *wsamples = (const float *)s->window->data[0]; float sample = samples[frame->channels * offset + ch]; + float wsample = wsamples[frame->channels * s->window_offset + ch]; float new_sum; new_sum = s->sum; - new_sum -= *s->window_current; + new_sum -= wsample; new_sum += sample * sample; return sqrtf(new_sum / s->window_size); @@ -265,24 +266,24 @@ static void update_rms_float(SilenceRemoveContext *s, AVFrame *frame, int ch, in { const float *samples = (const float *)frame->data[0]; float sample = samples[frame->channels * offset + ch]; + float *wsamples = (float *)s->window->data[0]; + float *wsample = &wsamples[frame->channels * s->window_offset + ch]; - s->sum -= *s->window_current; - *s->window_current = sample * sample; - s->sum += *s->window_current; - - s->window_current++; - if (s->window_current >= s->window_end) - s->window_current = s->window; + s->sum -= *wsample; + *wsample = sample * sample; + s->sum += *wsample; } static double compute_peak_doublep(SilenceRemoveContext *s, AVFrame *frame, int ch, int offset) { - const double *samples = (const double *)frame->data[ch]; + const double *samples = (const double *)frame->extended_data[ch]; + const double *wsamples = (const double *)s->window->extended_data[ch]; double sample = samples[offset]; + double wsample = wsamples[s->window_offset]; double new_sum; new_sum = s->sum; - new_sum -= *s->window_current; + new_sum -= wsample; new_sum += fabs(sample); return new_sum / s->window_size; @@ -290,53 +291,53 @@ static double compute_peak_doublep(SilenceRemoveContext *s, AVFrame *frame, int static void update_peak_doublep(SilenceRemoveContext *s, AVFrame *frame, int ch, int offset) { - const double *samples = (const double *)frame->data[ch]; + const double *samples = (const double *)frame->extended_data[ch]; + double *wsamples = (double *)s->window->extended_data[ch]; double sample = samples[offset]; + double *wsample = &wsamples[s->window_offset]; - s->sum -= *s->window_current; - *s->window_current = fabs(sample); - s->sum += *s->window_current; - - s->window_current++; - if (s->window_current >= s->window_end) - s->window_current = s->window; + s->sum -= *wsample; + *wsample = fabs(sample); + s->sum += *wsample; } static double compute_peak_floatp(SilenceRemoveContext *s, AVFrame *frame, int ch, int offset) { - const float *samples = (const float *)frame->data[ch]; + const float *samples = (const float *)frame->extended_data[ch]; + const float *wsamples = (const float *)s->window->extended_data[ch]; float sample = samples[offset]; + float wsample = wsamples[s->window_offset]; float new_sum; new_sum = s->sum; - new_sum -= *s->window_current; - new_sum += fabs(sample); + new_sum -= wsample; + new_sum += fabsf(sample); return new_sum / s->window_size; } static void update_peak_floatp(SilenceRemoveContext *s, AVFrame *frame, int ch, int offset) { - const float *samples = (const float *)frame->data[ch]; + const float *samples = (const float *)frame->extended_data[ch]; + float *wsamples = (float *)s->window->extended_data[ch]; float sample = samples[offset]; + float *wsample = &wsamples[s->window_offset]; - s->sum -= *s->window_current; - *s->window_current = fabs(sample); - s->sum += *s->window_current; - - s->window_current++; - if (s->window_current >= s->window_end) - s->window_current = s->window; + s->sum -= *wsample; + *wsample = fabsf(sample); + s->sum += *wsample; } static double compute_rms_doublep(SilenceRemoveContext *s, AVFrame *frame, int ch, int offset) { - const double *samples = (const double *)frame->data[ch]; + const double *samples = (const double *)frame->extended_data[ch]; + const double *wsamples = (const double *)s->window->extended_data[ch]; double sample = samples[offset]; + double wsample = wsamples[s->window_offset]; double new_sum; new_sum = s->sum; - new_sum -= *s->window_current; + new_sum -= wsample; new_sum += sample * sample; return sqrt(new_sum / s->window_size); @@ -344,26 +345,26 @@ static double compute_rms_doublep(SilenceRemoveContext *s, AVFrame *frame, int c static void update_rms_doublep(SilenceRemoveContext *s, AVFrame *frame, int ch, int offset) { - const double *samples = (const double *)frame->data[ch]; + const double *samples = (const double *)frame->extended_data[ch]; + double *wsamples = (double *)s->window->extended_data[ch]; double sample = samples[offset]; + double *wsample = &wsamples[s->window_offset]; - s->sum -= *s->window_current; - *s->window_current = sample * sample; - s->sum += *s->window_current; - - s->window_current++; - if (s->window_current >= s->window_end) - s->window_current = s->window; + s->sum -= *wsample; + *wsample = sample * sample; + s->sum += *wsample; } static double compute_rms_floatp(SilenceRemoveContext *s, AVFrame *frame, int ch, int offset) { - const float *samples = (const float *)frame->data[ch]; + const float *samples = (const float *)frame->extended_data[ch]; + const float *wsamples = (const float *)s->window->extended_data[ch]; float sample = samples[offset]; + float wsample = wsamples[s->window_offset]; float new_sum; new_sum = s->sum; - new_sum -= *s->window_current; + new_sum -= wsample; new_sum += sample * sample; return sqrtf(new_sum / s->window_size); @@ -371,16 +372,14 @@ static double compute_rms_floatp(SilenceRemoveContext *s, AVFrame *frame, int ch static void update_rms_floatp(SilenceRemoveContext *s, AVFrame *frame, int ch, int offset) { - const float *samples = (const float *)frame->data[ch]; + const float *samples = (const float *)frame->extended_data[ch]; + float *wsamples = (float *)s->window->extended_data[ch]; float sample = samples[offset]; + float *wsample = &wsamples[s->window_offset]; - s->sum -= *s->window_current; - *s->window_current = sample * sample; - s->sum += *s->window_current; - - s->window_current++; - if (s->window_current >= s->window_end) - s->window_current = s->window; + s->sum -= *wsample; + *wsample = sample * sample; + s->sum += *wsample; } static av_cold int init(AVFilterContext *ctx) @@ -397,10 +396,10 @@ static av_cold int init(AVFilterContext *ctx) static void clear_window(SilenceRemoveContext *s) { - memset(s->window, 0, s->window_size * sizeof(*s->window)); + av_samples_set_silence(s->window->extended_data, 0, s->window_size, + s->window->channels, s->window->format); - s->window_current = s->window; - s->window_end = s->window + s->window_size; + s->window_offset = 0; s->sum = 0; } @@ -410,8 +409,8 @@ static int config_input(AVFilterLink *inlink) SilenceRemoveContext *s = ctx->priv; s->next_pts = AV_NOPTS_VALUE; - s->window_size = FFMAX((inlink->sample_rate * s->window_ratio), 1) * inlink->channels; - s->window = av_malloc_array(s->window_size, sizeof(*s->window)); + s->window_size = FFMAX((inlink->sample_rate * s->window_ratio), 1); + s->window = ff_get_audio_buffer(ctx->outputs[0], s->window_size); if (!s->window) return AVERROR(ENOMEM); @@ -549,14 +548,14 @@ static void flush(SilenceRemoveContext *s, } if (s->stop_silence_offset < s->stop_silence_end) { - av_samples_copy(silence->data, s->stop_silence_hold->data, 0, + av_samples_copy(silence->extended_data, s->stop_silence_hold->extended_data, 0, s->stop_silence_offset, s->stop_silence_end - s->stop_silence_offset, outlink->channels, outlink->format); } if (s->stop_silence_offset > 0) { - av_samples_copy(silence->data, s->stop_silence_hold->data, + av_samples_copy(silence->extended_data, s->stop_silence_hold->extended_data, s->stop_silence_end - s->stop_silence_offset, 0, s->stop_silence_offset, outlink->channels, outlink->format); @@ -612,6 +611,10 @@ silence_trim: s->update(s, in, j, nb_samples_read); s->copy(s, s->start_holdoff, in, j, s->start_holdoff_end, nb_samples_read); } + + s->window_offset++; + if (s->window_offset >= s->window_size) + s->window_offset = 0; s->start_holdoff_end++; nb_samples_read++; @@ -635,6 +638,9 @@ silence_trim: s->copy(s, s->start_silence_hold, in, j, s->start_silence_offset, nb_samples_read); } + s->window_offset++; + if (s->window_offset >= s->window_size) + s->window_offset = 0; nb_samples_read++; s->start_silence_offset++; @@ -661,21 +667,21 @@ silence_trim_flush: if (s->start_silence_end > 0) { if (s->start_silence_offset < s->start_silence_end) { - av_samples_copy(out->data, s->start_silence_hold->data, 0, + av_samples_copy(out->extended_data, s->start_silence_hold->extended_data, 0, s->start_silence_offset, s->start_silence_end - s->start_silence_offset, outlink->channels, outlink->format); } if (s->start_silence_offset > 0) { - av_samples_copy(out->data, s->start_silence_hold->data, + av_samples_copy(out->extended_data, s->start_silence_hold->extended_data, s->start_silence_end - s->start_silence_offset, 0, s->start_silence_offset, outlink->channels, outlink->format); } } - av_samples_copy(out->data, s->start_holdoff->data, + av_samples_copy(out->extended_data, s->start_holdoff->extended_data, s->start_silence_end, s->start_holdoff_offset, nbs, outlink->channels, outlink->format); @@ -735,6 +741,9 @@ silence_copy: s->copy(s, out, in, j, nb_samples_written, nb_samples_read); } + s->window_offset++; + if (s->window_offset >= s->window_size) + s->window_offset = 0; nb_samples_read++; nb_samples_written++; } else if (!threshold) { @@ -750,6 +759,10 @@ silence_copy: s->copy(s, s->stop_holdoff, in, j, s->stop_holdoff_end, nb_samples_read); } + + s->window_offset++; + if (s->window_offset >= s->window_size) + s->window_offset = 0; nb_samples_read++; s->stop_holdoff_end++; @@ -783,7 +796,7 @@ silence_copy: } flush(s, out, outlink, &nb_samples_written, &ret, 0); } else { - av_samples_copy(out->data, in->data, + av_samples_copy(out->extended_data, in->extended_data, nb_samples_written, nb_samples_read, nbs, outlink->channels, outlink->format); @@ -809,7 +822,7 @@ silence_copy_flush: return AVERROR(ENOMEM); } - av_samples_copy(out->data, s->stop_holdoff->data, 0, + av_samples_copy(out->extended_data, s->stop_holdoff->extended_data, 0, s->stop_holdoff_offset, nbs, outlink->channels, outlink->format); @@ -858,7 +871,7 @@ static int request_frame(AVFilterLink *outlink) if (!frame) return AVERROR(ENOMEM); - av_samples_copy(frame->data, s->stop_holdoff->data, 0, + av_samples_copy(frame->extended_data, s->stop_holdoff->extended_data, 0, s->stop_holdoff_offset, nbs, outlink->channels, outlink->format); @@ -913,7 +926,7 @@ static av_cold void uninit(AVFilterContext *ctx) av_frame_free(&s->start_silence_hold); av_frame_free(&s->stop_holdoff); av_frame_free(&s->stop_silence_hold); - av_freep(&s->window); + av_frame_free(&s->window); } static const AVFilterPad silenceremove_inputs[] = {

1 0

ffmpeg: fix order between field order autodetection and override
by Jan Ekström 25 Jul '21

25 Jul '21

ffmpeg | branch: master | Jan Ekström <jeebjp(a)gmail.com> | Mon Jul 19 23:25:28 2021 +0300| [4c694093be68d401c60819e5171817c62afef8b2] | committer: Jan Ekström ffmpeg: fix order between field order autodetection and override Having the override before autodetection meant that the overridden value got overwritten by the autodetected result each time, effectively disabling the ability to utilize the `-top` option for override purposes. Somehow I missed this in fbb44bc51a647862eb05ae3f9d7d49a0be9bed57 , even though the lines were within the context. Probably the code originally being after this logic had something to do with it, but previously it only touched the avformat context's codecpar, which did not affect the encoder codec context whatsoever. Fixes #9320 Fixes #9339 > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=4c694093be68d401c… --- fftools/ffmpeg.c | 14 ++++++++------ 1 file changed, 8 insertions(+), 6 deletions(-) diff --git a/fftools/ffmpeg.c b/fftools/ffmpeg.c index 40e8010096..b0ce7c7c32 100644 --- a/fftools/ffmpeg.c +++ b/fftools/ffmpeg.c @@ -3483,12 +3483,7 @@ static int init_output_stream_encode(OutputStream *ost, AVFrame *frame) enc_ctx->bits_per_raw_sample = frame_bits_per_raw_sample; } - if (ost->top_field_first == 0) { - enc_ctx->field_order = AV_FIELD_BB; - } else if (ost->top_field_first == 1) { - enc_ctx->field_order = AV_FIELD_TT; - } - + // Field order: autodetection if (frame) { if (enc_ctx->flags & (AV_CODEC_FLAG_INTERLACED_DCT | AV_CODEC_FLAG_INTERLACED_ME) && ost->top_field_first >= 0) @@ -3503,6 +3498,13 @@ static int init_output_stream_encode(OutputStream *ost, AVFrame *frame) enc_ctx->field_order = AV_FIELD_PROGRESSIVE; } + // Field order: override + if (ost->top_field_first == 0) { + enc_ctx->field_order = AV_FIELD_BB; + } else if (ost->top_field_first == 1) { + enc_ctx->field_order = AV_FIELD_TT; + } + if (ost->forced_keyframes) { if (!strncmp(ost->forced_keyframes, "expr:", 5)) { ret = av_expr_parse(&ost->forced_keyframes_pexpr, ost->forced_keyframes+5,

1 0