mirror of
https://github.com/FFmpeg/FFmpeg.git
synced 2024-12-12 19:18:44 +02:00
d42b410e05
This commit moves some of the functionality from avfilter/colorspace into avutil/csp and exposes it as a public API so it can be used by libavcodec and/or libavformat. It also converts those structs from double values to AVRational to make regression testing easier and more consistent. Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
328 lines
11 KiB
C
328 lines
11 KiB
C
/*
|
|
* 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/csp.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,
|
|
};
|
|
|
|
typedef struct TonemapContext {
|
|
const AVClass *class;
|
|
|
|
enum TonemapAlgorithm tonemap;
|
|
double param;
|
|
double desat;
|
|
double peak;
|
|
|
|
const AVLumaCoefficients *coeffs;
|
|
} TonemapContext;
|
|
|
|
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)
|
|
{
|
|
int map[3] = { desc->comp[0].plane, desc->comp[1].plane, desc->comp[2].plane };
|
|
const float *r_in = (const float *)(in->data[map[0]] + x * desc->comp[map[0]].step + y * in->linesize[map[0]]);
|
|
const float *g_in = (const float *)(in->data[map[1]] + x * desc->comp[map[1]].step + y * in->linesize[map[1]]);
|
|
const float *b_in = (const float *)(in->data[map[2]] + x * desc->comp[map[2]].step + y * in->linesize[map[2]]);
|
|
float *r_out = (float *)(out->data[map[0]] + x * desc->comp[map[0]].step + y * out->linesize[map[0]]);
|
|
float *g_out = (float *)(out->data[map[1]] + x * desc->comp[map[1]].step + y * out->linesize[map[1]]);
|
|
float *b_out = (float *)(out->data[map[2]] + x * desc->comp[map[2]].step + y * out->linesize[map[2]]);
|
|
float sig, sig_orig;
|
|
|
|
/* load values */
|
|
*r_out = *r_in;
|
|
*g_out = *g_in;
|
|
*b_out = *b_in;
|
|
|
|
/* desaturate to prevent unnatural colors */
|
|
if (s->desat > 0) {
|
|
float luma = av_q2d(s->coeffs->cr) * *r_in + av_q2d(s->coeffs->cg) * *g_in + av_q2d(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 = av_csp_luma_coeffs_from_avcsp(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,
|
|
FILTER_INPUTS(tonemap_inputs),
|
|
FILTER_OUTPUTS(tonemap_outputs),
|
|
FILTER_PIXFMTS(AV_PIX_FMT_GBRPF32, AV_PIX_FMT_GBRAPF32),
|
|
.flags = AVFILTER_FLAG_SLICE_THREADS,
|
|
};
|