FFmpeg/libavfilter/vf_xbr.c
Andreas Rheinhardt b4f5201967 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>
2021-10-05 17:48:25 +02:00

434 lines
23 KiB
C

/*
* This file is part of FFmpeg.
*
* Copyright (c) 2011, 2012 Hyllian/Jararaca <sergiogdb@gmail.com>
* Copyright (c) 2014 Arwa Arif <arwaarif1994@gmail.com>
*
* 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
* XBR Filter is used for depixelization of image.
* This is based on Hyllian's xBR shader.
*
* @see https://forums.libretro.com/t/xbr-algorithm-tutorial/123
* @see https://github.com/yoyofr/iFBA/blob/master/fba_src/src/intf/video/scalers/xbr.cpp
*/
#include "libavutil/opt.h"
#include "libavutil/pixdesc.h"
#include "internal.h"
#define LB_MASK 0x00FEFEFE
#define RED_BLUE_MASK 0x00FF00FF
#define GREEN_MASK 0x0000FF00
#ifdef PI
#undef PI
#endif
typedef int (*xbrfunc_t)(AVFilterContext *ctx, void *arg, int jobnr, int nb_jobs);
typedef struct XBRContext {
const AVClass *class;
int n;
xbrfunc_t func;
uint32_t rgbtoyuv[1<<24];
} XBRContext;
typedef struct ThreadData {
AVFrame *in, *out;
const uint32_t *rgbtoyuv;
} ThreadData;
#define OFFSET(x) offsetof(XBRContext, x)
#define FLAGS AV_OPT_FLAG_FILTERING_PARAM|AV_OPT_FLAG_VIDEO_PARAM
static const AVOption xbr_options[] = {
{ "n", "set scale factor", OFFSET(n), AV_OPT_TYPE_INT, {.i64 = 3}, 2, 4, .flags = FLAGS },
{ NULL }
};
AVFILTER_DEFINE_CLASS(xbr);
static uint32_t pixel_diff(uint32_t x, uint32_t y, const uint32_t *r2y)
{
#define YMASK 0xff0000
#define UMASK 0x00ff00
#define VMASK 0x0000ff
#define ABSDIFF(a,b) (abs((int)(a)-(int)(b)))
uint32_t yuv1 = r2y[x & 0xffffff];
uint32_t yuv2 = r2y[y & 0xffffff];
return (ABSDIFF(yuv1 & YMASK, yuv2 & YMASK) >> 16) +
(ABSDIFF(yuv1 & UMASK, yuv2 & UMASK) >> 8) +
ABSDIFF(yuv1 & VMASK, yuv2 & VMASK);
}
#define ALPHA_BLEND_128_W(a, b) ((((a) & LB_MASK) >> 1) + (((b) & LB_MASK) >> 1))
#define ALPHA_BLEND_BASE(a, b, m, s) ( (RED_BLUE_MASK & (((a) & RED_BLUE_MASK) + (((((b) & RED_BLUE_MASK) - ((a) & RED_BLUE_MASK)) * (m)) >> (s)))) \
| (GREEN_MASK & (((a) & GREEN_MASK) + (((((b) & GREEN_MASK) - ((a) & GREEN_MASK)) * (m)) >> (s)))))
#define ALPHA_BLEND_32_W(a, b) ALPHA_BLEND_BASE(a, b, 1, 3)
#define ALPHA_BLEND_64_W(a, b) ALPHA_BLEND_BASE(a, b, 1, 2)
#define ALPHA_BLEND_192_W(a, b) ALPHA_BLEND_BASE(a, b, 3, 2)
#define ALPHA_BLEND_224_W(a, b) ALPHA_BLEND_BASE(a, b, 7, 3)
#define df(A, B) pixel_diff(A, B, r2y)
#define eq(A, B) (df(A, B) < 155)
#define FILT2(PE, PI, PH, PF, PG, PC, PD, PB, PA, G5, C4, G0, D0, C1, B1, F4, I4, H5, I5, A0, A1, \
N0, N1, N2, N3) do { \
if (PE != PH && PE != PF) { \
const unsigned e = df(PE,PC) + df(PE,PG) + df(PI,H5) + df(PI,F4) + (df(PH,PF)<<2); \
const unsigned i = df(PH,PD) + df(PH,I5) + df(PF,I4) + df(PF,PB) + (df(PE,PI)<<2); \
if (e <= i) { \
const unsigned px = df(PE,PF) <= df(PE,PH) ? PF : PH; \
if (e < i && (!eq(PF,PB) && !eq(PH,PD) || eq(PE,PI) \
&& (!eq(PF,I4) && !eq(PH,I5)) \
|| eq(PE,PG) || eq(PE,PC))) { \
const unsigned ke = df(PF,PG); \
const unsigned ki = df(PH,PC); \
const int left = ke<<1 <= ki && PE != PG && PD != PG; \
const int up = ke >= ki<<1 && PE != PC && PB != PC; \
if (left && up) { \
E[N3] = ALPHA_BLEND_224_W(E[N3], px); \
E[N2] = ALPHA_BLEND_64_W( E[N2], px); \
E[N1] = E[N2]; \
} else if (left) { \
E[N3] = ALPHA_BLEND_192_W(E[N3], px); \
E[N2] = ALPHA_BLEND_64_W( E[N2], px); \
} else if (up) { \
E[N3] = ALPHA_BLEND_192_W(E[N3], px); \
E[N1] = ALPHA_BLEND_64_W( E[N1], px); \
} else { /* diagonal */ \
E[N3] = ALPHA_BLEND_128_W(E[N3], px); \
} \
} else { \
E[N3] = ALPHA_BLEND_128_W(E[N3], px); \
} \
} \
} \
} while (0)
#define FILT3(PE, PI, PH, PF, PG, PC, PD, PB, PA, G5, C4, G0, D0, C1, B1, F4, I4, H5, I5, A0, A1, \
N0, N1, N2, N3, N4, N5, N6, N7, N8) do { \
if (PE != PH && PE != PF) { \
const unsigned e = df(PE,PC) + df(PE,PG) + df(PI,H5) + df(PI,F4) + (df(PH,PF)<<2); \
const unsigned i = df(PH,PD) + df(PH,I5) + df(PF,I4) + df(PF,PB) + (df(PE,PI)<<2); \
if (e <= i) { \
const unsigned px = df(PE,PF) <= df(PE,PH) ? PF : PH; \
if (e < i && (!eq(PF,PB) && !eq(PF,PC) || !eq(PH,PD) && !eq(PH,PG) || eq(PE,PI) \
&& (!eq(PF,F4) && !eq(PF,I4) || !eq(PH,H5) && !eq(PH,I5)) \
|| eq(PE,PG) || eq(PE,PC))) { \
const unsigned ke = df(PF,PG); \
const unsigned ki = df(PH,PC); \
const int left = ke<<1 <= ki && PE != PG && PD != PG; \
const int up = ke >= ki<<1 && PE != PC && PB != PC; \
if (left && up) { \
E[N7] = ALPHA_BLEND_192_W(E[N7], px); \
E[N6] = ALPHA_BLEND_64_W( E[N6], px); \
E[N5] = E[N7]; \
E[N2] = E[N6]; \
E[N8] = px; \
} else if (left) { \
E[N7] = ALPHA_BLEND_192_W(E[N7], px); \
E[N5] = ALPHA_BLEND_64_W( E[N5], px); \
E[N6] = ALPHA_BLEND_64_W( E[N6], px); \
E[N8] = px; \
} else if (up) { \
E[N5] = ALPHA_BLEND_192_W(E[N5], px); \
E[N7] = ALPHA_BLEND_64_W( E[N7], px); \
E[N2] = ALPHA_BLEND_64_W( E[N2], px); \
E[N8] = px; \
} else { /* diagonal */ \
E[N8] = ALPHA_BLEND_224_W(E[N8], px); \
E[N5] = ALPHA_BLEND_32_W( E[N5], px); \
E[N7] = ALPHA_BLEND_32_W( E[N7], px); \
} \
} else { \
E[N8] = ALPHA_BLEND_128_W(E[N8], px); \
} \
} \
} \
} while (0)
#define FILT4(PE, PI, PH, PF, PG, PC, PD, PB, PA, G5, C4, G0, D0, C1, B1, F4, I4, H5, I5, A0, A1, \
N15, N14, N11, N3, N7, N10, N13, N12, N9, N6, N2, N1, N5, N8, N4, N0) do { \
if (PE != PH && PE != PF) { \
const unsigned e = df(PE,PC) + df(PE,PG) + df(PI,H5) + df(PI,F4) + (df(PH,PF)<<2); \
const unsigned i = df(PH,PD) + df(PH,I5) + df(PF,I4) + df(PF,PB) + (df(PE,PI)<<2); \
if (e <= i) { \
const unsigned px = df(PE,PF) <= df(PE,PH) ? PF : PH; \
if (e < i && (!eq(PF,PB) && !eq(PH,PD) || eq(PE,PI) \
&& (!eq(PF,I4) && !eq(PH,I5)) \
|| eq(PE,PG) || eq(PE,PC))) { \
const unsigned ke = df(PF,PG); \
const unsigned ki = df(PH,PC); \
const int left = ke<<1 <= ki && PE != PG && PD != PG; \
const int up = ke >= ki<<1 && PE != PC && PB != PC; \
if (left && up) { \
E[N13] = ALPHA_BLEND_192_W(E[N13], px); \
E[N12] = ALPHA_BLEND_64_W( E[N12], px); \
E[N15] = E[N14] = E[N11] = px; \
E[N10] = E[N3] = E[N12]; \
E[N7] = E[N13]; \
} else if (left) { \
E[N11] = ALPHA_BLEND_192_W(E[N11], px); \
E[N13] = ALPHA_BLEND_192_W(E[N13], px); \
E[N10] = ALPHA_BLEND_64_W( E[N10], px); \
E[N12] = ALPHA_BLEND_64_W( E[N12], px); \
E[N14] = px; \
E[N15] = px; \
} else if (up) { \
E[N14] = ALPHA_BLEND_192_W(E[N14], px); \
E[N7 ] = ALPHA_BLEND_192_W(E[N7 ], px); \
E[N10] = ALPHA_BLEND_64_W( E[N10], px); \
E[N3 ] = ALPHA_BLEND_64_W( E[N3 ], px); \
E[N11] = px; \
E[N15] = px; \
} else { /* diagonal */ \
E[N11] = ALPHA_BLEND_128_W(E[N11], px); \
E[N14] = ALPHA_BLEND_128_W(E[N14], px); \
E[N15] = px; \
} \
} else { \
E[N15] = ALPHA_BLEND_128_W(E[N15], px); \
} \
} \
} \
} while (0)
static av_always_inline void xbr_filter(const ThreadData *td, int jobnr, int nb_jobs, int n)
{
int x, y;
const AVFrame *input = td->in;
AVFrame *output = td->out;
const uint32_t *r2y = td->rgbtoyuv;
const int slice_start = (input->height * jobnr ) / nb_jobs;
const int slice_end = (input->height * (jobnr+1)) / nb_jobs;
const int nl = output->linesize[0] >> 2;
const int nl1 = nl + nl;
const int nl2 = nl1 + nl;
for (y = slice_start; y < slice_end; y++) {
uint32_t *E = (uint32_t *)(output->data[0] + y * output->linesize[0] * n);
const uint32_t *sa2 = (uint32_t *)(input->data[0] + y * input->linesize[0] - 8); /* center */
const uint32_t *sa1 = sa2 - (input->linesize[0]>>2); /* up x1 */
const uint32_t *sa0 = sa1 - (input->linesize[0]>>2); /* up x2 */
const uint32_t *sa3 = sa2 + (input->linesize[0]>>2); /* down x1 */
const uint32_t *sa4 = sa3 + (input->linesize[0]>>2); /* down x2 */
if (y <= 1) {
sa0 = sa1;
if (y == 0) {
sa0 = sa1 = sa2;
}
}
if (y >= input->height - 2) {
sa4 = sa3;
if (y == input->height - 1) {
sa4 = sa3 = sa2;
}
}
for (x = 0; x < input->width; x++) {
const uint32_t B1 = sa0[2];
const uint32_t PB = sa1[2];
const uint32_t PE = sa2[2];
const uint32_t PH = sa3[2];
const uint32_t H5 = sa4[2];
const int pprev = 2 - (x > 0);
const uint32_t A1 = sa0[pprev];
const uint32_t PA = sa1[pprev];
const uint32_t PD = sa2[pprev];
const uint32_t PG = sa3[pprev];
const uint32_t G5 = sa4[pprev];
const int pprev2 = pprev - (x > 1);
const uint32_t A0 = sa1[pprev2];
const uint32_t D0 = sa2[pprev2];
const uint32_t G0 = sa3[pprev2];
const int pnext = 3 - (x == input->width - 1);
const uint32_t C1 = sa0[pnext];
const uint32_t PC = sa1[pnext];
const uint32_t PF = sa2[pnext];
const uint32_t PI = sa3[pnext];
const uint32_t I5 = sa4[pnext];
const int pnext2 = pnext + 1 - (x >= input->width - 2);
const uint32_t C4 = sa1[pnext2];
const uint32_t F4 = sa2[pnext2];
const uint32_t I4 = sa3[pnext2];
if (n == 2) {
E[0] = E[1] = // 0, 1
E[nl] = E[nl + 1] = PE; // 2, 3
FILT2(PE, PI, PH, PF, PG, PC, PD, PB, PA, G5, C4, G0, D0, C1, B1, F4, I4, H5, I5, A0, A1, 0, 1, nl, nl+1);
FILT2(PE, PC, PF, PB, PI, PA, PH, PD, PG, I4, A1, I5, H5, A0, D0, B1, C1, F4, C4, G5, G0, nl, 0, nl+1, 1);
FILT2(PE, PA, PB, PD, PC, PG, PF, PH, PI, C1, G0, C4, F4, G5, H5, D0, A0, B1, A1, I4, I5, nl+1, nl, 1, 0);
FILT2(PE, PG, PD, PH, PA, PI, PB, PF, PC, A0, I5, A1, B1, I4, F4, H5, G5, D0, G0, C1, C4, 1, nl+1, 0, nl);
} else if (n == 3) {
E[0] = E[1] = E[2] = // 0, 1, 2
E[nl] = E[nl+1] = E[nl+2] = // 3, 4, 5
E[nl1] = E[nl1+1] = E[nl1+2] = PE; // 6, 7, 8
FILT3(PE, PI, PH, PF, PG, PC, PD, PB, PA, G5, C4, G0, D0, C1, B1, F4, I4, H5, I5, A0, A1, 0, 1, 2, nl, nl+1, nl+2, nl1, nl1+1, nl1+2);
FILT3(PE, PC, PF, PB, PI, PA, PH, PD, PG, I4, A1, I5, H5, A0, D0, B1, C1, F4, C4, G5, G0, nl1, nl, 0, nl1+1, nl+1, 1, nl1+2, nl+2, 2);
FILT3(PE, PA, PB, PD, PC, PG, PF, PH, PI, C1, G0, C4, F4, G5, H5, D0, A0, B1, A1, I4, I5, nl1+2, nl1+1, nl1, nl+2, nl+1, nl, 2, 1, 0);
FILT3(PE, PG, PD, PH, PA, PI, PB, PF, PC, A0, I5, A1, B1, I4, F4, H5, G5, D0, G0, C1, C4, 2, nl+2, nl1+2, 1, nl+1, nl1+1, 0, nl, nl1);
} else if (n == 4) {
E[0] = E[1] = E[2] = E[3] = // 0, 1, 2, 3
E[nl] = E[nl+1] = E[nl+2] = E[nl+3] = // 4, 5, 6, 7
E[nl1] = E[nl1+1] = E[nl1+2] = E[nl1+3] = // 8, 9, 10, 11
E[nl2] = E[nl2+1] = E[nl2+2] = E[nl2+3] = PE; // 12, 13, 14, 15
FILT4(PE, PI, PH, PF, PG, PC, PD, PB, PA, G5, C4, G0, D0, C1, B1, F4, I4, H5, I5, A0, A1, nl2+3, nl2+2, nl1+3, 3, nl+3, nl1+2, nl2+1, nl2, nl1+1, nl+2, 2, 1, nl+1, nl1, nl, 0);
FILT4(PE, PC, PF, PB, PI, PA, PH, PD, PG, I4, A1, I5, H5, A0, D0, B1, C1, F4, C4, G5, G0, 3, nl+3, 2, 0, 1, nl+2, nl1+3, nl2+3, nl1+2, nl+1, nl, nl1, nl1+1, nl2+2, nl2+1, nl2);
FILT4(PE, PA, PB, PD, PC, PG, PF, PH, PI, C1, G0, C4, F4, G5, H5, D0, A0, B1, A1, I4, I5, 0, 1, nl, nl2, nl1, nl+1, 2, 3, nl+2, nl1+1, nl2+1, nl2+2, nl1+2, nl+3, nl1+3, nl2+3);
FILT4(PE, PG, PD, PH, PA, PI, PB, PF, PC, A0, I5, A1, B1, I4, F4, H5, G5, D0, G0, C1, C4, nl2, nl1, nl2+1, nl2+3, nl2+2, nl1+1, nl, 0, nl+1, nl1+2, nl1+3, nl+3, nl+2, 1, 2, 3);
}
sa0 += 1;
sa1 += 1;
sa2 += 1;
sa3 += 1;
sa4 += 1;
E += n;
}
}
}
#define XBR_FUNC(size) \
static int xbr##size##x(AVFilterContext *ctx, void *arg, int jobnr, int nb_jobs) \
{ \
xbr_filter(arg, jobnr, nb_jobs, size); \
return 0; \
}
XBR_FUNC(2)
XBR_FUNC(3)
XBR_FUNC(4)
static int config_output(AVFilterLink *outlink)
{
AVFilterContext *ctx = outlink->src;
XBRContext *s = ctx->priv;
AVFilterLink *inlink = ctx->inputs[0];
outlink->w = inlink->w * s->n;
outlink->h = inlink->h * s->n;
return 0;
}
static int query_formats(AVFilterContext *ctx)
{
static const enum AVPixelFormat pix_fmts[] = {
AV_PIX_FMT_0RGB32, AV_PIX_FMT_NONE,
};
return ff_set_common_formats_from_list(ctx, pix_fmts);
}
static int filter_frame(AVFilterLink *inlink, AVFrame *in)
{
AVFilterContext *ctx = inlink->dst;
AVFilterLink *outlink = ctx->outputs[0];
XBRContext *s = ctx->priv;
ThreadData td;
AVFrame *out = ff_get_video_buffer(outlink, outlink->w, outlink->h);
if (!out) {
av_frame_free(&in);
return AVERROR(ENOMEM);
}
av_frame_copy_props(out, in);
td.in = in;
td.out = out;
td.rgbtoyuv = s->rgbtoyuv;
ff_filter_execute(ctx, s->func, &td, NULL,
FFMIN(inlink->h, ff_filter_get_nb_threads(ctx)));
out->width = outlink->w;
out->height = outlink->h;
av_frame_free(&in);
return ff_filter_frame(outlink, out);
}
static av_cold int init(AVFilterContext *ctx)
{
XBRContext *s = ctx->priv;
static const xbrfunc_t xbrfuncs[] = {xbr2x, xbr3x, xbr4x};
uint32_t c;
int bg, rg, g;
for (bg = -255; bg < 256; bg++) {
for (rg = -255; rg < 256; rg++) {
const uint32_t u = (uint32_t)((-169*rg + 500*bg)/1000) + 128;
const uint32_t v = (uint32_t)(( 500*rg - 81*bg)/1000) + 128;
int startg = FFMAX3(-bg, -rg, 0);
int endg = FFMIN3(255-bg, 255-rg, 255);
uint32_t y = (uint32_t)(( 299*rg + 1000*startg + 114*bg)/1000);
c = bg + rg * (1 << 16) + 0x010101 * startg;
for (g = startg; g <= endg; g++) {
s->rgbtoyuv[c] = ((y++) << 16) + (u << 8) + v;
c+= 0x010101;
}
}
}
s->func = xbrfuncs[s->n - 2];
return 0;
}
static const AVFilterPad xbr_inputs[] = {
{
.name = "default",
.type = AVMEDIA_TYPE_VIDEO,
.filter_frame = filter_frame,
},
};
static const AVFilterPad xbr_outputs[] = {
{
.name = "default",
.type = AVMEDIA_TYPE_VIDEO,
.config_props = config_output,
},
};
const AVFilter ff_vf_xbr = {
.name = "xbr",
.description = NULL_IF_CONFIG_SMALL("Scale the input using xBR algorithm."),
FILTER_INPUTS(xbr_inputs),
FILTER_OUTPUTS(xbr_outputs),
FILTER_QUERY_FUNC(query_formats),
.priv_size = sizeof(XBRContext),
.priv_class = &xbr_class,
.init = init,
.flags = AVFILTER_FLAG_SLICE_THREADS,
};