summaryrefslogtreecommitdiffstats
path: root/video/filter/vf_gradfun.c
blob: ea3b615caa8d18575e81f0f4a1ab611849531cd4 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
/*
 * Copyright (C) 2009 Loren Merritt <lorenm@u.washignton.edu>
 *
 * This file is part of MPlayer.
 *
 * MPlayer is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 2 of the License, or
 * (at your option) any later version.
 *
 * MPlayer 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 General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License along
 * with MPlayer; if not, write to the Free Software Foundation, Inc.,
 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
 */

/*
 * Debanding algorithm (from gradfun2db by prunedtree):
 * Boxblur.
 * Foreach pixel, if it's within threshold of the blurred value, make it closer.
 * So now we have a smoothed and higher bitdepth version of all the shallow
 * gradients, while leaving detailed areas untouched.
 * Dither it back to 8bit.
 */

#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <inttypes.h>
#include <math.h>

#include <libavutil/common.h>

#include "config.h"
#include "mpvcore/cpudetect.h"
#include "video/img_format.h"
#include "video/mp_image.h"
#include "vf.h"
#include "video/memcpy_pic.h"
#include "compat/x86_cpu.h"

#include "options/m_option.h"

#include "vf_lavfi.h"

struct vf_priv_s {
    float cfg_thresh;
    int cfg_radius;
    float cfg_size;
    int thresh;
    int radius;
    uint16_t *buf;
    void (*filter_line)(uint8_t *dst, uint8_t *src, uint16_t *dc,
                        int width, int thresh, const uint16_t *dithers);
    void (*blur_line)(uint16_t *dc, uint16_t *buf, uint16_t *buf1,
                      uint8_t *src, int sstride, int width);
    struct vf_lw_opts *lw_opts;
} const vf_priv_dflt = {
  .cfg_thresh = 1.5,
  .cfg_radius = -1,
  .cfg_size = -1,
};

static const uint16_t __attribute__((aligned(16))) pw_7f[8] = {127,127,127,127,127,127,127,127};
static const uint16_t __attribute__((aligned(16))) pw_ff[8] = {255,255,255,255,255,255,255,255};
static const uint16_t __attribute__((aligned(16))) dither[8][8] = {
    {  0, 96, 24,120,  6,102, 30,126 },
    { 64, 32, 88, 56, 70, 38, 94, 62 },
    { 16,112,  8,104, 22,118, 14,110 },
    { 80, 48, 72, 40, 86, 54, 78, 46 },
    {  4,100, 28,124,  2, 98, 26,122 },
    { 68, 36, 92, 60, 66, 34, 90, 58 },
    { 20,116, 12,108, 18,114, 10,106 },
    { 84, 52, 76, 44, 82, 50, 74, 42 },
};

static void filter_line_c(uint8_t *dst, uint8_t *src, uint16_t *dc,
                          int width, int thresh, const uint16_t *dithers)
{
    int x;
    for (x=0; x<width; x++, dc+=x&1) {
        int pix = src[x]<<7;
        int delta = dc[0] - pix;
        int m = abs(delta) * thresh >> 16;
        m = FFMAX(0, 127-m);
        m = m*m*delta >> 14;
        pix += m + dithers[x&7];
        dst[x] = av_clip_uint8(pix>>7);
    }
}

static void blur_line_c(uint16_t *dc, uint16_t *buf, uint16_t *buf1,
                        uint8_t *src, int sstride, int width)
{
    int x, v, old;
    for (x=0; x<width; x++) {
        v = buf1[x] + src[2*x] + src[2*x+1] + src[2*x+sstride] + src[2*x+1+sstride];
        old = buf[x];
        buf[x] = v;
        dc[x] = v - old;
    }
}

#if HAVE_MMX2
static void filter_line_mmx2(uint8_t *dst, uint8_t *src, uint16_t *dc,
                             int width, int thresh, const uint16_t *dithers)
{
    intptr_t x;
    if (width&3) {
        x = width&~3;
        filter_line_c(dst+x, src+x, dc+x/2, width-x, thresh, dithers);
        width = x;
    }
    x = -width;
    __asm__ volatile(
        "movd          %4, %%mm5 \n"
        "pxor       %%mm7, %%mm7 \n"
        "pshufw $0, %%mm5, %%mm5 \n"
        "movq          %6, %%mm6 \n"
        "movq          %5, %%mm4 \n"
        "1: \n"
        "movd     (%2,%0), %%mm0 \n"
        "movd     (%3,%0), %%mm1 \n"
        "punpcklbw  %%mm7, %%mm0 \n"
        "punpcklwd  %%mm1, %%mm1 \n"
        "psllw         $7, %%mm0 \n"
        "pxor       %%mm2, %%mm2 \n"
        "psubw      %%mm0, %%mm1 \n" // delta = dc - pix
        "psubw      %%mm1, %%mm2 \n"
        "pmaxsw     %%mm1, %%mm2 \n"
        "pmulhuw    %%mm5, %%mm2 \n" // m = abs(delta) * thresh >> 16
        "psubw      %%mm6, %%mm2 \n"
        "pminsw     %%mm7, %%mm2 \n" // m = -max(0, 127-m)
        "pmullw     %%mm2, %%mm2 \n"
        "paddw      %%mm4, %%mm0 \n" // pix += dither
        "pmulhw     %%mm2, %%mm1 \n"
        "psllw         $2, %%mm1 \n" // m = m*m*delta >> 14
        "paddw      %%mm1, %%mm0 \n" // pix += m
        "psraw         $7, %%mm0 \n"
        "packuswb   %%mm0, %%mm0 \n"
        "movd       %%mm0, (%1,%0) \n" // dst = clip(pix>>7)
        "add           $4, %0 \n"
        "jl 1b \n"
        "emms \n"
        :"+r"(x)
        :"r"(dst+width), "r"(src+width), "r"(dc+width/2),
         "rm"(thresh), "m"(*dithers), "m"(*pw_7f)
        :"memory"
    );
}
#endif

#if HAVE_SSSE3
static void filter_line_ssse3(uint8_t *dst, uint8_t *src, uint16_t *dc,
                              int width, int thresh, const uint16_t *dithers)
{
    intptr_t x;
    if (width&7) {
        // could be 10% faster if I somehow eliminated this
        x = width&~7;
        filter_line_c(dst+x, src+x, dc+x/2, width-x, thresh, dithers);
        width = x;
    }
    x = -width;
    __asm__ volatile(
        "movd           %4, %%xmm5 \n"
        "pxor       %%xmm7, %%xmm7 \n"
        "pshuflw $0,%%xmm5, %%xmm5 \n"
        "movdqa         %6, %%xmm6 \n"
        "punpcklqdq %%xmm5, %%xmm5 \n"
        "movdqa         %5, %%xmm4 \n"
        "1: \n"
        "movq      (%2,%0), %%xmm0 \n"
        "movq      (%3,%0), %%xmm1 \n"
        "punpcklbw  %%xmm7, %%xmm0 \n"
        "punpcklwd  %%xmm1, %%xmm1 \n"
        "psllw          $7, %%xmm0 \n"
        "psubw      %%xmm0, %%xmm1 \n" // delta = dc - pix
        "pabsw      %%xmm1, %%xmm2 \n"
        "pmulhuw    %%xmm5, %%xmm2 \n" // m = abs(delta) * thresh >> 16
        "psubw      %%xmm6, %%xmm2 \n"
        "pminsw     %%xmm7, %%xmm2 \n" // m = -max(0, 127-m)
        "pmullw     %%xmm2, %%xmm2 \n"
        "psllw          $1, %%xmm2 \n"
        "paddw      %%xmm4, %%xmm0 \n" // pix += dither
        "pmulhrsw   %%xmm2, %%xmm1 \n" // m = m*m*delta >> 14
        "paddw      %%xmm1, %%xmm0 \n" // pix += m
        "psraw          $7, %%xmm0 \n"
        "packuswb   %%xmm0, %%xmm0 \n"
        "movq       %%xmm0, (%1,%0) \n" // dst = clip(pix>>7)
        "add            $8, %0 \n"
        "jl 1b \n"
        :"+&r"(x)
        :"r"(dst+width), "r"(src+width), "r"(dc+width/2),
         "rm"(thresh), "m"(*dithers), "m"(*pw_7f)
        :"memory"
    );
}
#endif // HAVE_SSSE3

#if HAVE_SSE2 && HAVE_6REGS
#define BLURV(load)\
    intptr_t x = -2*width;\
    __asm__ volatile(\
        "movdqa %6, %%xmm7 \n"\
        "1: \n"\
        load"   (%4,%0), %%xmm0 \n"\
        load"   (%5,%0), %%xmm1 \n"\
        "movdqa  %%xmm0, %%xmm2 \n"\
        "movdqa  %%xmm1, %%xmm3 \n"\
        "psrlw       $8, %%xmm0 \n"\
        "psrlw       $8, %%xmm1 \n"\
        "pand    %%xmm7, %%xmm2 \n"\
        "pand    %%xmm7, %%xmm3 \n"\
        "paddw   %%xmm1, %%xmm0 \n"\
        "paddw   %%xmm3, %%xmm2 \n"\
        "paddw   %%xmm2, %%xmm0 \n"\
        "paddw  (%2,%0), %%xmm0 \n"\
        "movdqa (%1,%0), %%xmm1 \n"\
        "movdqa  %%xmm0, (%1,%0) \n"\
        "psubw   %%xmm1, %%xmm0 \n"\
        "movdqa  %%xmm0, (%3,%0) \n"\
        "add        $16, %0 \n"\
        "jl 1b \n"\
        :"+&r"(x)\
        :"r"(buf+width),\
         "r"(buf1+width),\
         "r"(dc+width),\
         "r"(src+width*2),\
         "r"(src+width*2+sstride),\
         "m"(*pw_ff)\
        :"memory"\
    );

static void blur_line_sse2(uint16_t *dc, uint16_t *buf, uint16_t *buf1,
                           uint8_t *src, int sstride, int width)
{
    if (((intptr_t)src|sstride)&15) {
        BLURV("movdqu");
    } else {
        BLURV("movdqa");
    }
}
#endif // HAVE_6REGS && HAVE_SSE2

static void filter_plane(struct vf_priv_s *ctx, uint8_t *dst, uint8_t *src,
                         int width, int height, int dstride, int sstride, int r)
{
    int bstride = ((width+15)&~15)/2;
    int y;
    uint32_t dc_factor = (1<<21)/(r*r);
    uint16_t *dc = ctx->buf+16;
    uint16_t *buf = ctx->buf+bstride+32;
    int thresh = ctx->thresh;

    memset(dc, 0, (bstride+16)*sizeof(*buf));
    for (y=0; y<r; y++)
        ctx->blur_line(dc, buf+y*bstride, buf+(y-1)*bstride, src+2*y*sstride, sstride, width/2);
    for (;;) {
        if (y < height-r) {
            int mod = ((y+r)/2)%r;
            uint16_t *buf0 = buf+mod*bstride;
            uint16_t *buf1 = buf+(mod?mod-1:r-1)*bstride;
            int x, v;
            ctx->blur_line(dc, buf0, buf1, src+(y+r)*sstride, sstride, width/2);
            for (x=v=0; x<r; x++)
                v += dc[x];
            for (; x<width/2; x++) {
                v += dc[x] - dc[x-r];
                dc[x-r] = v * dc_factor >> 16;
            }
            for (; x<(width+r+1)/2; x++)
                dc[x-r] = v * dc_factor >> 16;
            for (x=-r/2; x<0; x++)
                dc[x] = dc[0];
        }
        if (y == r) {
            for (y=0; y<r; y++)
                ctx->filter_line(dst+y*dstride, src+y*sstride, dc-r/2, width, thresh, dither[y&7]);
        }
        ctx->filter_line(dst+y*dstride, src+y*sstride, dc-r/2, width, thresh, dither[y&7]);
        if (++y >= height) break;
        ctx->filter_line(dst+y*dstride, src+y*sstride, dc-r/2, width, thresh, dither[y&7]);
        if (++y >= height) break;
    }
}

static struct mp_image *filter(struct vf_instance *vf, struct mp_image *mpi)
{
    struct mp_image *dmpi = mpi;
    if (!mp_image_is_writeable(mpi)) {
        dmpi = vf_alloc_out_image(vf);
        mp_image_copy_attributes(dmpi, mpi);
    }

    for (int p=0; p < mpi->num_planes; p++) {
        int w = mpi->w;
        int h = mpi->h;
        int r = vf->priv->radius;
        if (p) {
            w >>= mpi->chroma_x_shift;
            h >>= mpi->chroma_y_shift;
            r = ((r>>mpi->chroma_x_shift) + (r>>mpi->chroma_y_shift)) / 2;
            r = av_clip((r+1)&~1,4,32);
        }
        if (FFMIN(w,h) > 2*r)
            filter_plane(vf->priv, dmpi->planes[p], mpi->planes[p], w, h,
                                   dmpi->stride[p], mpi->stride[p], r);
        else if (dmpi->planes[p] != mpi->planes[p])
            memcpy_pic(dmpi->planes[p], mpi->planes[p], w, h,
                       dmpi->stride[p], mpi->stride[p]);
    }

    if (dmpi != mpi)
        talloc_free(mpi);
    return dmpi;
}

static int query_format(struct vf_instance *vf, unsigned int fmt)
{
    switch (fmt){
    case IMGFMT_Y8:
    case IMGFMT_444P:
    case IMGFMT_422P:
    case IMGFMT_440P:
    case IMGFMT_420P:
    case IMGFMT_411P:
    case IMGFMT_410P:
        return vf_next_query_format(vf,fmt);
    }
    return 0;
}

static int config(struct vf_instance *vf,
                  int width, int height, int d_width, int d_height,
                  unsigned int flags, unsigned int outfmt)
{
    free(vf->priv->buf);
    vf->priv->radius = vf->priv->cfg_radius;
    if (vf->priv->cfg_size > -1) {
        vf->priv->radius = (vf->priv->cfg_size / 100.0f)
                           * sqrtf(width * width + height * height);
    }
    vf->priv->radius = av_clip((vf->priv->radius+1)&~1, 4, 32);
    vf->priv->buf = av_mallocz((((width+15)&~15)*(vf->priv->radius+1)/2+32)*sizeof(uint16_t));
    return vf_next_config(vf,width,height,d_width,d_height,flags,outfmt);
}

static void uninit(struct vf_instance *vf)
{
    if (!vf->priv) return;
    av_free(vf->priv->buf);
}

static void lavfi_recreate(struct vf_instance *vf)
{
    struct vf_priv_s *p = vf_lw_old_priv(vf);
    int w = vf->fmt_in.w;
    int h = vf->fmt_in.h;
    p->radius = p->cfg_radius;
    if (p->cfg_size > -1)
        p->radius = (p->cfg_size / 100.0f) * sqrtf(w * w + h * h);
    p->radius = av_clip((p->radius+1)&~1, 4, 32);
    vf_lw_update_graph(vf, "gradfun", "%f:%d", p->cfg_thresh, p->radius);
}

static int vf_open(vf_instance_t *vf)
{
    vf->filter=filter;
    vf->query_format=query_format;
    vf->config=config;
    vf->uninit=uninit;

    bool have_radius = vf->priv->cfg_radius > -1;
    bool have_size = vf->priv->cfg_size > -1;

    if (have_radius && have_size) {
        mp_msg(MSGT_VFILTER, MSGL_ERR, "scale: gradfun: only one of "
              "radius/size parameters allowed at the same time!\n");
        return 0;
    }

    if (!have_radius && !have_size)
        vf->priv->cfg_size = 1.0;

    if (vf_lw_set_graph(vf, vf->priv->lw_opts, "gradfun", "%f:4",
                        vf->priv->cfg_thresh) >= 0)
    {
        vf_lw_set_recreate_cb(vf, lavfi_recreate);
        return 1;
    }

    vf->priv->thresh = (1<<15)/av_clipf(vf->priv->cfg_thresh,0.51,255);

    vf->priv->blur_line = blur_line_c;
    vf->priv->filter_line = filter_line_c;
#if HAVE_SSE2 && HAVE_6REGS
    if (gCpuCaps.hasSSE2)
        vf->priv->blur_line = blur_line_sse2;
#endif
#if HAVE_MMX2
    if (gCpuCaps.hasMMX2)
        vf->priv->filter_line = filter_line_mmx2;
#endif
#if HAVE_SSSE3
    if (gCpuCaps.hasSSSE3)
        vf->priv->filter_line = filter_line_ssse3;
#endif

    return 1;
}

#define OPT_BASE_STRUCT struct vf_priv_s
static const m_option_t vf_opts_fields[] = {
    OPT_FLOATRANGE("strength", cfg_thresh, 0, 0.51, 255),
    OPT_INTRANGE("radius", cfg_radius, 0, 4, 32),
    OPT_FLOATRANGE("size", cfg_size, 0, 0.1, 5.0),
    OPT_SUBSTRUCT("", lw_opts, vf_lw_conf, 0),
    {0}
};

const vf_info_t vf_info_gradfun = {
    .description = "gradient deband",
    .name = "gradfun",
    .open = vf_open,
    .priv_size = sizeof(struct vf_priv_s),
    .priv_defaults = &vf_priv_dflt,
    .options = vf_opts_fields,
};