#include "mpegvideoencdsp.h"
#include "dirac_dwt.h"
#include "dirac.h"
+#include "diractab.h"
#include "diracdsp.h"
#include "videodsp.h"
-/**
- * The spec limits the number of wavelet decompositions to 4 for both
- * level 1 (VC-2) and 128 (long-gop default).
- * 5 decompositions is the maximum before >16-bit buffers are needed.
- * Schroedinger allows this for DD 9,7 and 13,7 wavelets only, limiting
- * the others to 4 decompositions (or 3 for the fidelity filter).
- *
- * We use this instead of MAX_DECOMPOSITIONS to save some memory.
- */
-#define MAX_DWT_LEVELS 5
-
/**
* The spec limits this to 3 for frame coding, but in practice can be as high as 6
*/
} SubBand;
typedef struct Plane {
+ DWTPlane idwt;
+
int width;
int height;
ptrdiff_t stride;
- int idwt_width;
- int idwt_height;
- int idwt_stride;
- uint8_t *idwt_buf;
- uint8_t *idwt_buf_base;
- uint8_t *idwt_tmp;
-
/* block length */
uint8_t xblen;
uint8_t yblen;
subband_nb,
};
-static const uint8_t default_qmat[][4][4] = {
- { { 5, 3, 3, 0}, { 0, 4, 4, 1}, { 0, 5, 5, 2}, { 0, 6, 6, 3} },
- { { 4, 2, 2, 0}, { 0, 4, 4, 2}, { 0, 5, 5, 3}, { 0, 7, 7, 5} },
- { { 5, 3, 3, 0}, { 0, 4, 4, 1}, { 0, 5, 5, 2}, { 0, 6, 6, 3} },
- { { 8, 4, 4, 0}, { 0, 4, 4, 0}, { 0, 4, 4, 0}, { 0, 4, 4, 0} },
- { { 8, 4, 4, 0}, { 0, 4, 4, 0}, { 0, 4, 4, 0}, { 0, 4, 4, 0} },
- { { 0, 4, 4, 8}, { 0, 8, 8, 12}, { 0, 13, 13, 17}, { 0, 17, 17, 21} },
- { { 3, 1, 1, 0}, { 0, 4, 4, 2}, { 0, 6, 6, 5}, { 0, 9, 9, 7} },
-};
-
-static const int32_t qscale_tab[128] = {
- 4, 5, 6, 7, 8, 10, 11, 13,
- 16, 19, 23, 27, 32, 38, 45, 54,
- 64, 76, 91, 108, 128, 152, 181, 215,
- 256, 304, 362, 431, 512, 609, 724, 861,
- 1024, 1218, 1448, 1722, 2048, 2435, 2896, 3444,
- 4096, 4871, 5793, 6889, 8192, 9742, 11585, 13777,
- 16384, 19484, -13317, 27554, 32768, -1581, 9853, -10518,
- 65536, -3164, -16782, -21037, 131072, -6328, 2922, 23552,
- 262144, -12658, 5844, -18524, 524288, 15232, 11689, 28578,
- 1048576, -10085, -13110, -8471, 2097152, -20170, 10267, -16943,
- 4194304, 208, -15954, 31741, 8388608, 416, 4579, -2146,
- 16777216, 832, 9158, -4293, 33554432, 1663, -18172, -8587,
- 67108864, 3326, 143, -17175, 134217728, 6653, 285, 31276,
-268435456, 13306, 570, -3075, 536870912, -13938, 1140, -6152,
-1073741824, 12672, 2281, -12304, -2147483648, -15205, 4561, -24610,
- 0, 10138, 9122, 16407, 0, -20274, -18243, -32813,
-};
-
-static const int32_t qoffset_intra_tab[128] = {
- 1, 2, 3, 4, 4, 5, 6, 7,
- 8, 10, 12, 14, 16, 19, 23, 27,
- 32, 38, 46, 54, 64, 76, 91, 108,
- 128, 152, 181, 216, 256, 305, 362, 431,
- 512, 609, 724, 861, 1024, 1218, 1448, 1722,
- 2048, 2436, 2897, 3445, 4096, 4871, 5793, 6889,
- 8192, 9742, -6658, 13777, 16384, -790, 4927, -5258,
- 32768, -1581, -8390, -10518, 65536, -3163, 1461, 11776,
- 131072, -6328, 2922, -9261, 262144, 7616, 5845, 14289,
- 524288, -5042, -6554, -4235, 1048576, -10084, 5134, -8471,
- 2097152, 104, -7976, 15871, 4194304, 208, 2290, -1072,
- 8388608, 416, 4579, -2146, 16777216, 832, -9085, -4293,
- 33554432, 1663, 72, -8587, 67108864, 3327, 143, 15638,
-134217728, 6653, 285, -1537, 268435456, -6968, 570, -3075,
-536870912, 6336, 1141, -6151, -1073741823, -7602, 2281, -12304,
- 0, 5069, 4561, 8204, 0, -10136, -9121, -16406,
-};
-
-static const int qoffset_inter_tab[MAX_QUANT+1] = {
- 1, 2, 2, 3, 3, 4, 4, 5,
- 6, 7, 9, 10, 12, 14, 17, 20,
- 24, 29, 34, 41, 48, 57, 68, 81,
- 96, 114, 136, 162, 192, 228, 272, 323,
- 384, 457, 543, 646, 768, 913, 1086, 1292,
- 1536, 1827, 2172, 2583, 3072, 3653, 4344, 5166,
- 6144, 7307, 8689, 10333, 12288, 14613, 17378, 20666,
- 24576, 29226
-};
-
/* magic number division by 3 from schroedinger */
static inline int divide3(int x)
{
w = FFALIGN(CALC_PADDING(w, MAX_DWT_LEVELS), 8); /* FIXME: Should this be 16 for SSE??? */
h = top_padding + CALC_PADDING(h, MAX_DWT_LEVELS) + max_yblen/2;
- s->plane[i].idwt_buf_base = av_mallocz_array((w+max_xblen), h * (2 << s->pshift));
- s->plane[i].idwt_tmp = av_malloc_array((w+16), 2 << s->pshift);
- s->plane[i].idwt_buf = s->plane[i].idwt_buf_base + (top_padding*w)*(2 << s->pshift);
- if (!s->plane[i].idwt_buf_base || !s->plane[i].idwt_tmp)
+ s->plane[i].idwt.buf_base = av_mallocz_array((w+max_xblen), h * (2 << s->pshift));
+ s->plane[i].idwt.tmp = av_malloc_array((w+16), 2 << s->pshift);
+ s->plane[i].idwt.buf = s->plane[i].idwt.buf_base + (top_padding*w)*(2 << s->pshift);
+ if (!s->plane[i].idwt.buf_base || !s->plane[i].idwt.tmp)
return AVERROR(ENOMEM);
}
memset(s->delay_frames, 0, sizeof(s->delay_frames));
for (i = 0; i < 3; i++) {
- av_freep(&s->plane[i].idwt_buf_base);
- av_freep(&s->plane[i].idwt_tmp);
+ av_freep(&s->plane[i].idwt.buf_base);
+ av_freep(&s->plane[i].idwt.tmp);
}
s->buffer_stride = 0;
b->quant = quant;
}
- b->quant = FFMIN(b->quant, MAX_QUANT);
+ if (b->quant > 115) {
+ av_log(s->avctx, AV_LOG_ERROR, "Unsupported quant %d\n", b->quant);
+ b->quant = 0;
+ return;
+ }
- qfactor = qscale_tab[b->quant];
+ qfactor = ff_dirac_qscale_tab[b->quant];
/* TODO: context pointer? */
if (!s->num_refs)
- qoffset = qoffset_intra_tab[b->quant] + 2;
+ qoffset = ff_dirac_qoffset_intra_tab[b->quant] + 2;
else
- qoffset = qoffset_inter_tab[b->quant] + 2;
+ qoffset = ff_dirac_qoffset_inter_tab[b->quant] + 2;
buf = b->ibuf + top * b->stride;
if (is_arith) {
int top = b1->height * slice_y / s->num_y;
int bottom = b1->height *(slice_y+1) / s->num_y;
- int qfactor = qscale_tab[quant & 0x7f];
- int qoffset = qoffset_intra_tab[quant & 0x7f] + 2;
+ int qfactor, qoffset;
uint8_t *buf1 = b1->ibuf + top * b1->stride;
uint8_t *buf2 = b2 ? b2->ibuf + top * b2->stride: NULL;
int x, y;
+
+ if (quant > 115) {
+ av_log(s->avctx, AV_LOG_ERROR, "Unsupported quant %d\n", quant);
+ return;
+ }
+ qfactor = ff_dirac_qscale_tab[quant & 0x7f];
+ qoffset = ff_dirac_qoffset_intra_tab[quant & 0x7f] + 2;
/* we have to constantly check for overread since the spec explicitly
requires this, with the meaning that all remaining coeffs are set to 0 */
if (get_bits_count(gb) >= bits_end)
p->width = s->seq.width >> (i ? s->chroma_x_shift : 0);
p->height = s->seq.height >> (i ? s->chroma_y_shift : 0);
- p->idwt_width = w = CALC_PADDING(p->width , s->wavelet_depth);
- p->idwt_height = h = CALC_PADDING(p->height, s->wavelet_depth);
- p->idwt_stride = FFALIGN(p->idwt_width, 8) << (1 + s->pshift);
+ p->idwt.width = w = CALC_PADDING(p->width , s->wavelet_depth);
+ p->idwt.height = h = CALC_PADDING(p->height, s->wavelet_depth);
+ p->idwt.stride = FFALIGN(p->idwt.width, 8) << (1 + s->pshift);
for (level = s->wavelet_depth-1; level >= 0; level--) {
w = w>>1;
SubBand *b = &p->band[level][orientation];
b->pshift = s->pshift;
- b->ibuf = p->idwt_buf;
+ b->ibuf = p->idwt.buf;
b->level = level;
- b->stride = p->idwt_stride << (s->wavelet_depth - level);
+ b->stride = p->idwt.stride << (s->wavelet_depth - level);
b->width = w;
b->height = h;
b->orientation = orientation;
/* default quantization matrix */
for (level = 0; level < s->wavelet_depth; level++)
for (i = 0; i < 4; i++) {
- s->lowdelay.quant[level][i] = default_qmat[s->wavelet_idx][level][i];
+ s->lowdelay.quant[level][i] = ff_dirac_default_qmat[s->wavelet_idx][level][i];
/* haar with no shift differs for different depths */
if (s->wavelet_idx == 3)
s->lowdelay.quant[level][i] += 4*(s->wavelet_depth-1 - level);
/* [DIRAC_STD] 13.5.1 low_delay_transform_data() */
for (comp = 0; comp < 3; comp++) {
Plane *p = &s->plane[comp];
- memset(p->idwt_buf, 0, p->idwt_stride * p->idwt_height);
+ memset(p->idwt.buf, 0, p->idwt.stride * p->idwt.height);
}
if (!s->zero_res) {
if ((ret = decode_lowdelay(s)) < 0)
if (!s->zero_res && !s->low_delay)
{
- memset(p->idwt_buf, 0, p->idwt_stride * p->idwt_height);
+ memset(p->idwt.buf, 0, p->idwt.stride * p->idwt.height);
decode_component(s, comp); /* [DIRAC_STD] 13.4.1 core_transform_data() */
}
- ret = ff_spatial_idwt_init2(&d, p->idwt_buf, p->idwt_width, p->idwt_height, p->idwt_stride,
- s->wavelet_idx+2, s->wavelet_depth, p->idwt_tmp, s->bit_depth);
+ ret = ff_spatial_idwt_init(&d, &p->idwt, s->wavelet_idx+2,
+ s->wavelet_depth, s->bit_depth);
if (ret < 0)
return ret;
if (!s->num_refs) { /* intra */
for (y = 0; y < p->height; y += 16) {
+ int idx = (s->bit_depth - 8) >> 1;
ff_spatial_idwt_slice2(&d, y+16); /* decode */
- s->diracdsp.put_signed_rect_clamped[s->pshift](frame + y*p->stride, p->stride,
- p->idwt_buf + y*p->idwt_stride, p->idwt_stride, p->width, 16);
+ s->diracdsp.put_signed_rect_clamped[idx](frame + y*p->stride,
+ p->stride,
+ p->idwt.buf + y*p->idwt.stride,
+ p->idwt.stride, p->width, 16);
}
} else { /* inter */
int rowheight = p->ybsep*p->stride;
mctmp += (start - dsty)*p->stride + p->xoffset;
ff_spatial_idwt_slice2(&d, start + h); /* decode */
/* NOTE: add_rect_clamped hasn't been templated hence the shifts.
- * idwt_stride is passed as pixels, not in bytes as in the rest of the decoder */
+ * idwt.stride is passed as pixels, not in bytes as in the rest of the decoder */
s->diracdsp.add_rect_clamped(frame + start*p->stride, mctmp, p->stride,
- (int16_t*)(p->idwt_buf) + start*(p->idwt_stride >> 1), (p->idwt_stride >> 1), p->width, h);
+ (int16_t*)(p->idwt.buf) + start*(p->idwt.stride >> 1), (p->idwt.stride >> 1), p->width, h);
dsty += p->ybsep;
}
avctx->level = dsh->level;
avctx->framerate = dsh->framerate;
s->bit_depth = dsh->bit_depth;
+ s->version.major = dsh->version.major;
+ s->version.minor = dsh->version.minor;
s->seq = *dsh;
av_freep(&dsh);
pic->avframe->key_frame = s->num_refs == 0; /* [DIRAC_STD] is_intra() */
pic->avframe->pict_type = s->num_refs + 1; /* Definition of AVPictureType in avutil.h */
+ /* VC-2 Low Delay has a different parse code than the Dirac Low Delay */
if (s->version.minor == 2 && parse_code == 0x88)
s->ld_picture = 1;
projects / ffmpeg / blobdiff