Andreas Rheinhardt <andreas.rheinha...@outlook.com>:
> 发送时间: 2025年4月25日 15:11
> 收件人: ffmpeg-devel@ffmpeg.org <ffmpeg-devel@ffmpeg.org>
> 主题: Re: [FFmpeg-devel] [PATCH v1 03/19] avcodec/vvc/sei: add
> decode_film_grain_characteristics
>
> toq...@gmail.com:
>> From: Wu Jianhua <toq...@outlook.com>
>>
>> Signed-off-by: Wu Jianhua <toq...@outlook.com>
>> ---
>> libavcodec/vvc/sei.c | 66 ++++++++++++++++++++++++++++++++++++++++++++
>> 1 file changed, 66 insertions(+)
>>
>> diff --git a/libavcodec/vvc/sei.c b/libavcodec/vvc/sei.c
>> index 2842862a36..365f815950 100644
>> --- a/libavcodec/vvc/sei.c
>> +++ b/libavcodec/vvc/sei.c
>> @@ -24,15 +24,81 @@
>> #include "dec.h"
>> #include "libavutil/refstruct.h"
>>
>> +static int
>> decode_film_grain_characteristics(H2645SEIFilmGrainCharacteristics *h, const
>> SEIRawFilmGrainCharacteristics *s, const VVCFrameContext *fc)
>> +{
>> + const VVCSPS *sps = fc->ps.sps;
>> +
>> + h->present = !s->fg_characteristics_cancel_flag;
>> + if (h->present) {
>> + h->model_id = s->fg_model_id;
>> + h->separate_colour_description_present_flag =
>> s->fg_separate_colour_description_present_flag;
>> + if (h->separate_colour_description_present_flag) {
>> + h->bit_depth_luma = s->fg_bit_depth_luma_minus8 + 8;
>> + h->bit_depth_chroma = s->fg_bit_depth_chroma_minus8 +
>> 8;
>> + h->full_range = s->fg_full_range_flag;
>> + h->color_primaries = s->fg_colour_primaries;
>> + h->transfer_characteristics = s->fg_transfer_characteristics;
>> + h->matrix_coeffs = s->fg_matrix_coeffs;
>> + } else {
>> + if (!sps) {
>> + av_log(fc->log_ctx, AV_LOG_ERROR,
>> + "No active SPS for film_grain_characteristics.\n");
>> + return AVERROR_INVALIDDATA;
>> + }
>> + h->bit_depth_luma = sps->bit_depth;
>> + h->bit_depth_chroma = sps->bit_depth;
>> + h->full_range = sps->r->vui.vui_full_range_flag;
>> + h->color_primaries = sps->r->vui.vui_colour_primaries;
>> + h->transfer_characteristics =
>> sps->r->vui.vui_transfer_characteristics;
>> + h->matrix_coeffs = sps->r->vui.vui_matrix_coeffs ;
>> + }
>> +
>> + h->blending_mode_id = s->fg_blending_mode_id;
>> + h->log2_scale_factor = s->fg_log2_scale_factor;
>> +
>> + for (int c = 0; c < 3; c++) {
>> + h->comp_model_present_flag[c] =
>> s->fg_comp_model_present_flag[c];
>> + if (h->comp_model_present_flag[c]) {
>> + h->num_intensity_intervals[c] =
>> s->fg_num_intensity_intervals_minus1[c] + 1;
>> + h->num_model_values[c] =
>> s->fg_num_model_values_minus1[c] + 1;
>> +
>> + if (h->num_model_values[c] > 6)
>> + return AVERROR_INVALIDDATA;
>> +
>> + for (int i = 0; i < h->num_intensity_intervals[c]; i++) {
>> + h->intensity_interval_lower_bound[c][i] =
>> s->fg_intensity_interval_lower_bound[c][i];
>> + h->intensity_interval_upper_bound[c][i] =
>> s->fg_intensity_interval_upper_bound[c][i];
>> + for (int j = 0; j < h->num_model_values[c]; j++)
>> + h->comp_model_value[c][i][j] =
>> s->fg_comp_model_value[c][i][j];
>> + }
>> + }
>> + }
>> +
>> + h->persistence_flag = s->fg_characteristics_persistence_flag;
>> + }
>> +
>> + return 0;
>> +}
>> +
>
> This is a gigantic translation from CBS to h2645_sei. Several of the
> latter patches do likewise. Is it really beneficial to use h2645_sei,
> given that you do not use it for parsing?
>
Hi Andreas,
Yeah. The VVC decoder used the cbs to parse the sei message. The reason why
we need to translate the cbs to h2645 sei is that we want to use
ff_h2645_sei_to_frame
to export the side data so the user can use the film grain sei or other sei,
the same
as hevc/h264.
Thanks,
Jianhua
_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
To unsubscribe, visit link above, or email
ffmpeg-devel-requ...@ffmpeg.org with subject "unsubscribe".
