FFmpeg/libavfilter/vf_tonemap.c

347 lines
11 KiB
C
Raw Normal View History

/*
* Copyright (c) 2017 Vittorio Giovara <vittorio.giovara@gmail.com>
*
* This file is part of FFmpeg.
*
* FFmpeg is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2.1 of the License, or (at your option) any later version.
*
* FFmpeg is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with FFmpeg; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
/**
* @file
* tonemap algorithms
*/
#include <float.h>
#include <stdio.h>
#include <string.h>
#include "libavutil/imgutils.h"
#include "libavutil/internal.h"
#include "libavutil/intreadwrite.h"
#include "libavutil/opt.h"
#include "libavutil/pixdesc.h"
#include "avfilter.h"
#include "colorspace.h"
#include "formats.h"
#include "internal.h"
#include "video.h"
enum TonemapAlgorithm {
TONEMAP_NONE,
TONEMAP_LINEAR,
TONEMAP_GAMMA,
TONEMAP_CLIP,
TONEMAP_REINHARD,
TONEMAP_HABLE,
TONEMAP_MOBIUS,
TONEMAP_MAX,
};
static const struct LumaCoefficients luma_coefficients[AVCOL_SPC_NB] = {
[AVCOL_SPC_FCC] = { 0.30, 0.59, 0.11 },
[AVCOL_SPC_BT470BG] = { 0.299, 0.587, 0.114 },
[AVCOL_SPC_SMPTE170M] = { 0.299, 0.587, 0.114 },
[AVCOL_SPC_BT709] = { 0.2126, 0.7152, 0.0722 },
[AVCOL_SPC_SMPTE240M] = { 0.212, 0.701, 0.087 },
[AVCOL_SPC_BT2020_NCL] = { 0.2627, 0.6780, 0.0593 },
[AVCOL_SPC_BT2020_CL] = { 0.2627, 0.6780, 0.0593 },
};
typedef struct TonemapContext {
const AVClass *class;
enum TonemapAlgorithm tonemap;
double param;
double desat;
double peak;
const struct LumaCoefficients *coeffs;
} TonemapContext;
static const enum AVPixelFormat pix_fmts[] = {
AV_PIX_FMT_GBRPF32,
AV_PIX_FMT_GBRAPF32,
AV_PIX_FMT_NONE,
};
static int query_formats(AVFilterContext *ctx)
{
return ff_set_common_formats_from_list(ctx, pix_fmts);
}
static av_cold int init(AVFilterContext *ctx)
{
TonemapContext *s = ctx->priv;
switch(s->tonemap) {
case TONEMAP_GAMMA:
if (isnan(s->param))
s->param = 1.8f;
break;
case TONEMAP_REINHARD:
if (!isnan(s->param))
s->param = (1.0f - s->param) / s->param;
break;
case TONEMAP_MOBIUS:
if (isnan(s->param))
s->param = 0.3f;
break;
}
if (isnan(s->param))
s->param = 1.0f;
return 0;
}
static float hable(float in)
{
float a = 0.15f, b = 0.50f, c = 0.10f, d = 0.20f, e = 0.02f, f = 0.30f;
return (in * (in * a + b * c) + d * e) / (in * (in * a + b) + d * f) - e / f;
}
static float mobius(float in, float j, double peak)
{
float a, b;
if (in <= j)
return in;
a = -j * j * (peak - 1.0f) / (j * j - 2.0f * j + peak);
b = (j * j - 2.0f * j * peak + peak) / FFMAX(peak - 1.0f, 1e-6);
return (b * b + 2.0f * b * j + j * j) / (b - a) * (in + a) / (in + b);
}
#define MIX(x,y,a) (x) * (1 - (a)) + (y) * (a)
static void tonemap(TonemapContext *s, AVFrame *out, const AVFrame *in,
const AVPixFmtDescriptor *desc, int x, int y, double peak)
{
const float *r_in = (const float *)(in->data[0] + x * desc->comp[0].step + y * in->linesize[0]);
const float *b_in = (const float *)(in->data[1] + x * desc->comp[1].step + y * in->linesize[1]);
const float *g_in = (const float *)(in->data[2] + x * desc->comp[2].step + y * in->linesize[2]);
float *r_out = (float *)(out->data[0] + x * desc->comp[0].step + y * out->linesize[0]);
float *b_out = (float *)(out->data[1] + x * desc->comp[1].step + y * out->linesize[1]);
float *g_out = (float *)(out->data[2] + x * desc->comp[2].step + y * out->linesize[2]);
float sig, sig_orig;
/* load values */
*r_out = *r_in;
*b_out = *b_in;
*g_out = *g_in;
/* desaturate to prevent unnatural colors */
if (s->desat > 0) {
float luma = s->coeffs->cr * *r_in + s->coeffs->cg * *g_in + s->coeffs->cb * *b_in;
float overbright = FFMAX(luma - s->desat, 1e-6) / FFMAX(luma, 1e-6);
*r_out = MIX(*r_in, luma, overbright);
*g_out = MIX(*g_in, luma, overbright);
*b_out = MIX(*b_in, luma, overbright);
}
/* pick the brightest component, reducing the value range as necessary
* to keep the entire signal in range and preventing discoloration due to
* out-of-bounds clipping */
sig = FFMAX(FFMAX3(*r_out, *g_out, *b_out), 1e-6);
sig_orig = sig;
switch(s->tonemap) {
default:
case TONEMAP_NONE:
// do nothing
break;
case TONEMAP_LINEAR:
sig = sig * s->param / peak;
break;
case TONEMAP_GAMMA:
sig = sig > 0.05f ? pow(sig / peak, 1.0f / s->param)
: sig * pow(0.05f / peak, 1.0f / s->param) / 0.05f;
break;
case TONEMAP_CLIP:
sig = av_clipf(sig * s->param, 0, 1.0f);
break;
case TONEMAP_HABLE:
sig = hable(sig) / hable(peak);
break;
case TONEMAP_REINHARD:
sig = sig / (sig + s->param) * (peak + s->param) / peak;
break;
case TONEMAP_MOBIUS:
sig = mobius(sig, s->param, peak);
break;
}
/* apply the computed scale factor to the color,
* linearly to prevent discoloration */
*r_out *= sig / sig_orig;
*g_out *= sig / sig_orig;
*b_out *= sig / sig_orig;
}
typedef struct ThreadData {
AVFrame *in, *out;
const AVPixFmtDescriptor *desc;
double peak;
} ThreadData;
static int tonemap_slice(AVFilterContext *ctx, void *arg, int jobnr, int nb_jobs)
{
TonemapContext *s = ctx->priv;
ThreadData *td = arg;
AVFrame *in = td->in;
AVFrame *out = td->out;
const AVPixFmtDescriptor *desc = td->desc;
const int slice_start = (in->height * jobnr) / nb_jobs;
const int slice_end = (in->height * (jobnr+1)) / nb_jobs;
double peak = td->peak;
for (int y = slice_start; y < slice_end; y++)
for (int x = 0; x < out->width; x++)
tonemap(s, out, in, desc, x, y, peak);
return 0;
}
static int filter_frame(AVFilterLink *link, AVFrame *in)
{
AVFilterContext *ctx = link->dst;
TonemapContext *s = ctx->priv;
AVFilterLink *outlink = ctx->outputs[0];
ThreadData td;
AVFrame *out;
const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(link->format);
const AVPixFmtDescriptor *odesc = av_pix_fmt_desc_get(outlink->format);
int ret, x, y;
double peak = s->peak;
if (!desc || !odesc) {
av_frame_free(&in);
return AVERROR_BUG;
}
out = ff_get_video_buffer(outlink, outlink->w, outlink->h);
if (!out) {
av_frame_free(&in);
return AVERROR(ENOMEM);
}
ret = av_frame_copy_props(out, in);
if (ret < 0) {
av_frame_free(&in);
av_frame_free(&out);
return ret;
}
/* input and output transfer will be linear */
if (in->color_trc == AVCOL_TRC_UNSPECIFIED) {
av_log(s, AV_LOG_WARNING, "Untagged transfer, assuming linear light\n");
out->color_trc = AVCOL_TRC_LINEAR;
} else if (in->color_trc != AVCOL_TRC_LINEAR)
av_log(s, AV_LOG_WARNING, "Tonemapping works on linear light only\n");
/* read peak from side data if not passed in */
if (!peak) {
peak = ff_determine_signal_peak(in);
av_log(s, AV_LOG_DEBUG, "Computed signal peak: %f\n", peak);
}
/* load original color space even if pixel format is RGB to compute overbrights */
s->coeffs = &luma_coefficients[in->colorspace];
if (s->desat > 0 && (in->colorspace == AVCOL_SPC_UNSPECIFIED || !s->coeffs)) {
if (in->colorspace == AVCOL_SPC_UNSPECIFIED)
av_log(s, AV_LOG_WARNING, "Missing color space information, ");
else if (!s->coeffs)
av_log(s, AV_LOG_WARNING, "Unsupported color space '%s', ",
av_color_space_name(in->colorspace));
av_log(s, AV_LOG_WARNING, "desaturation is disabled\n");
s->desat = 0;
}
/* do the tone map */
td.out = out;
td.in = in;
td.desc = desc;
td.peak = peak;
ff_filter_execute(ctx, tonemap_slice, &td, NULL,
FFMIN(in->height, ff_filter_get_nb_threads(ctx)));
/* copy/generate alpha if needed */
if (desc->flags & AV_PIX_FMT_FLAG_ALPHA && odesc->flags & AV_PIX_FMT_FLAG_ALPHA) {
av_image_copy_plane(out->data[3], out->linesize[3],
in->data[3], in->linesize[3],
out->linesize[3], outlink->h);
} else if (odesc->flags & AV_PIX_FMT_FLAG_ALPHA) {
for (y = 0; y < out->height; y++) {
for (x = 0; x < out->width; x++) {
AV_WN32(out->data[3] + x * odesc->comp[3].step + y * out->linesize[3],
av_float2int(1.0f));
}
}
}
av_frame_free(&in);
ff_update_hdr_metadata(out, peak);
return ff_filter_frame(outlink, out);
}
#define OFFSET(x) offsetof(TonemapContext, x)
#define FLAGS AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_FILTERING_PARAM
static const AVOption tonemap_options[] = {
{ "tonemap", "tonemap algorithm selection", OFFSET(tonemap), AV_OPT_TYPE_INT, {.i64 = TONEMAP_NONE}, TONEMAP_NONE, TONEMAP_MAX - 1, FLAGS, "tonemap" },
{ "none", 0, 0, AV_OPT_TYPE_CONST, {.i64 = TONEMAP_NONE}, 0, 0, FLAGS, "tonemap" },
{ "linear", 0, 0, AV_OPT_TYPE_CONST, {.i64 = TONEMAP_LINEAR}, 0, 0, FLAGS, "tonemap" },
{ "gamma", 0, 0, AV_OPT_TYPE_CONST, {.i64 = TONEMAP_GAMMA}, 0, 0, FLAGS, "tonemap" },
{ "clip", 0, 0, AV_OPT_TYPE_CONST, {.i64 = TONEMAP_CLIP}, 0, 0, FLAGS, "tonemap" },
{ "reinhard", 0, 0, AV_OPT_TYPE_CONST, {.i64 = TONEMAP_REINHARD}, 0, 0, FLAGS, "tonemap" },
{ "hable", 0, 0, AV_OPT_TYPE_CONST, {.i64 = TONEMAP_HABLE}, 0, 0, FLAGS, "tonemap" },
{ "mobius", 0, 0, AV_OPT_TYPE_CONST, {.i64 = TONEMAP_MOBIUS}, 0, 0, FLAGS, "tonemap" },
{ "param", "tonemap parameter", OFFSET(param), AV_OPT_TYPE_DOUBLE, {.dbl = NAN}, DBL_MIN, DBL_MAX, FLAGS },
{ "desat", "desaturation strength", OFFSET(desat), AV_OPT_TYPE_DOUBLE, {.dbl = 2}, 0, DBL_MAX, FLAGS },
{ "peak", "signal peak override", OFFSET(peak), AV_OPT_TYPE_DOUBLE, {.dbl = 0}, 0, DBL_MAX, FLAGS },
{ NULL }
};
AVFILTER_DEFINE_CLASS(tonemap);
static const AVFilterPad tonemap_inputs[] = {
{
.name = "default",
.type = AVMEDIA_TYPE_VIDEO,
.filter_frame = filter_frame,
},
};
static const AVFilterPad tonemap_outputs[] = {
{
.name = "default",
.type = AVMEDIA_TYPE_VIDEO,
},
};
const AVFilter ff_vf_tonemap = {
.name = "tonemap",
.description = NULL_IF_CONFIG_SMALL("Conversion to/from different dynamic ranges."),
.init = init,
.priv_size = sizeof(TonemapContext),
.priv_class = &tonemap_class,
2021-08-12 11:05:31 +00:00
FILTER_INPUTS(tonemap_inputs),
FILTER_OUTPUTS(tonemap_outputs),
avfilter: Replace query_formats callback with union of list and callback If one looks at the many query_formats callbacks in existence, one will immediately recognize that there is one type of default callback for video and a slightly different default callback for audio: It is "return ff_set_common_formats_from_list(ctx, pix_fmts);" for video with a filter-specific pix_fmts list. For audio, it is the same with a filter-specific sample_fmts list together with ff_set_common_all_samplerates() and ff_set_common_all_channel_counts(). This commit allows to remove the boilerplate query_formats callbacks by replacing said callback with a union consisting the old callback and pointers for pixel and sample format arrays. For the not uncommon case in which these lists only contain a single entry (besides the sentinel) enum AVPixelFormat and enum AVSampleFormat fields are also added to the union to store them directly in the AVFilter, thereby avoiding a relocation. The state of said union will be contained in a new, dedicated AVFilter field (the nb_inputs and nb_outputs fields have been shrunk to uint8_t in order to create a hole for this new field; this is no problem, as the maximum of all the nb_inputs is four; for nb_outputs it is only two). The state's default value coincides with the earlier default of query_formats being unset, namely that the filter accepts all formats (and also sample rates and channel counts/layouts for audio) provided that these properties agree coincide for all inputs and outputs. By using different union members for audio and video filters the type-unsafety of using the same functions for audio and video lists will furthermore be more confined to formats.c than before. When the new fields are used, they will also avoid allocations: Currently something nearly equivalent to ff_default_query_formats() is called after every successful call to a query_formats callback; yet in the common case that the newly allocated AVFilterFormats are not used at all (namely if there are no free links) these newly allocated AVFilterFormats are freed again without ever being used. Filters no longer using the callback will not exhibit this any more. Reviewed-by: Paul B Mahol <onemda@gmail.com> Reviewed-by: Nicolas George <george@nsup.org> Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
2021-09-27 10:07:35 +00:00
FILTER_QUERY_FUNC(query_formats),
.flags = AVFILTER_FLAG_SLICE_THREADS,
};