1 =============================================
2 Snow Video Codec Specification Draft 20080110
3 =============================================
7 This specification describes the Snow bitstream syntax and semantics as
8 well as the formal Snow decoding process.
10 The decoding process is described precisely and any compliant decoder
11 MUST produce the exact same output for a spec-conformant Snow stream.
12 For encoding, though, any process which generates a stream compliant to
13 the syntactical and semantic requirements and which is decodable by
14 the process described in this spec shall be considered a conformant
20 MUST the specific part must be done to conform to this standard
21 SHOULD it is recommended to be done that way, but not strictly required
23 ilog2(x) is the rounded down logarithm of x with basis 2
30 u unsigned scalar value range coded
31 s signed scalar value range coded
44 if(keyframe || always_reset)
47 version u header_state
48 always_reset b header_state
49 temporal_decomposition_type u header_state
50 temporal_decomposition_count u header_state
51 spatial_decomposition_count u header_state
52 colorspace_type u header_state
53 chroma_h_shift u header_state
54 chroma_v_shift u header_state
55 spatial_scalability b header_state
56 max_ref_frames-1 u header_state
60 update_mc b header_state
62 for(plane=0; plane<2; plane++){
63 diag_mc b header_state
64 htaps/2-1 u header_state
65 for(i= p->htaps/2; i; i--)
66 |hcoeff[i]| u header_state
69 update_qlogs b header_state
71 spatial_decomposition_count u header_state
76 spatial_decomposition_type s header_state
78 mv_scale s header_state
80 block_max_depth s header_state
83 for(plane=0; plane<2; plane++){
84 quant_table[plane][0][0] s header_state
85 for(level=0; level < spatial_decomposition_count; level++){
86 quant_table[plane][level][1]s header_state
87 quant_table[plane][level][3]s header_state
95 for(y=0; y<block_count_vertical; y++)
96 for(x=0; x<block_count_horizontal; x++)
100 mvx_diff=mvy_diff=y_diff=cb_diff=cr_diff=0
104 if(level!=max_block_depth){
105 s_context= 2*left->level + 2*top->level + topleft->level + topright->level
106 leaf b block_state[4 + s_context]
108 if(level==max_block_depth || leaf){
109 intra b block_state[1 + left->intra + top->intra]
111 y_diff s block_state[32]
112 cb_diff s block_state[64]
113 cr_diff s block_state[96]
115 ref_context= ilog2(2*left->ref) + ilog2(2*top->ref)
117 ref u block_state[128 + 1024 + 32*ref_context]
118 mx_context= ilog2(2*abs(left->mx - top->mx))
119 my_context= ilog2(2*abs(left->my - top->my))
120 mvx_diff s block_state[128 + 32*(mx_context + 16*!!ref)]
121 mvy_diff s block_state[128 + 32*(my_context + 16*!!ref)]
138 for(level=0; level<spatial_decomposition_count; level++){
156 this MUST NOT change within a bitstream
159 if 1 then the range coder contexts will be reset after each frame
161 temporal_decomposition_type
164 temporal_decomposition_count
167 spatial_decomposition_count
172 this MUST NOT change within a bitstream
175 log2(luma.width / chroma.width)
176 this MUST NOT change within a bitstream
179 log2(luma.height / chroma.height)
180 this MUST NOT change within a bitstream
186 maximum number of reference frames
187 this MUST NOT change within a bitstream
190 indicates that motion compensation filter parameters are stored in the
194 flag to enable faster diagonal interpolation
195 this SHOULD be 1 unless it turns out to be covered by a valid patent
198 number of half pel interpolation filter taps, MUST be even, >0 and <10
201 half pel interpolation filter coefficients, hcoeff[0] are the 2 middle
202 coefficients [1] are the next outer ones and so on, resulting in a filter
203 like: ...eff[2], hcoeff[1], hcoeff[0], hcoeff[0], hcoeff[1], hcoeff[2] ...
204 the sign of the coefficients is not explicitly stored but alternates
205 after each coeff and coeff[0] is positive, so ...,+,-,+,-,+,+,-,+,-,+,...
206 hcoeff[0] is not explicitly stored but found by subtracting the sum
207 of all stored coefficients with signs from 32
208 hcoeff[0]= 32 - hcoeff[1] - hcoeff[2] - ...
209 a good choice for hcoeff and htaps is
212 an alternative which requires more computations at both encoder and
213 decoder side and may or may not be better is
219 minimum of the number of available reference frames and max_ref_frames
220 for example the first frame after a key frame always has ref_frames=1
222 spatial_decomposition_type
224 0 is a 9/7 symmetric compact integer wavelet
225 1 is a 5/3 symmetric compact integer wavelet
227 stored as delta from last, last is reset to 0 if always_reset || keyframe
230 quality (logarthmic quantizer scale)
231 stored as delta from last, last is reset to 0 if always_reset || keyframe
234 stored as delta from last, last is reset to 0 if always_reset || keyframe
235 FIXME check that everything works fine if this changes between frames
239 stored as delta from last, last is reset to 0 if always_reset || keyframe
242 maximum depth of the block tree
243 stored as delta from last, last is reset to 0 if always_reset || keyframe
249 Highlevel bitstream structure:
250 =============================
251 --------------------------------------------
253 --------------------------------------------
254 | ------------------------------------ |
258 | | ......... intra? | |
259 | | : Block01 : yes no | |
260 | | : Block02 : ....... .......... | |
261 | | : Block03 : : y DC : : ref index: | |
262 | | : Block04 : : cb DC : : motion x : | |
263 | | ......... : cr DC : : motion y : | |
264 | | ....... .......... | |
265 | ------------------------------------ |
266 | ------------------------------------ |
269 --------------------------------------------
270 | ------------ ------------ ------------ |
271 || Y subbands | | Cb subbands| | Cr subbands||
272 || --- --- | | --- --- | | --- --- ||
273 || |LL0||HL0| | | |LL0||HL0| | | |LL0||HL0| ||
274 || --- --- | | --- --- | | --- --- ||
275 || --- --- | | --- --- | | --- --- ||
276 || |LH0||HH0| | | |LH0||HH0| | | |LH0||HH0| ||
277 || --- --- | | --- --- | | --- --- ||
278 || --- --- | | --- --- | | --- --- ||
279 || |HL1||LH1| | | |HL1||LH1| | | |HL1||LH1| ||
280 || --- --- | | --- --- | | --- --- ||
281 || --- --- | | --- --- | | --- --- ||
282 || |HH1||HL2| | | |HH1||HL2| | | |HH1||HL2| ||
283 || ... | | ... | | ... ||
284 | ------------ ------------ ------------ |
285 --------------------------------------------
296 | | LL0 subband prediction
299 ------------------- \ |
300 | Reference frames | \ IDWT
301 | ------- ------- | Motion \ |
302 ||Frame 0| |Frame 1|| Compensation . OBMC v -------
303 | ------- ------- | --------------. \------> + --->|Frame n|-->output
304 | ------- ------- | -------
305 ||Frame 2| |Frame 3||<----------------------------------/
315 The implemented range coder is an adapted version based upon "Range encoding:
316 an algorithm for removing redundancy from a digitised message." by G. N. N.
318 The symbols encoded by the Snow range coder are bits (0|1). The
319 associated probabilities are not fix but change depending on the symbol mix
324 ---------+-----------------------------------------------
325 0 | 256 - state_transition_table[256 - old_state];
326 1 | state_transition_table[ old_state];
328 state_transition_table = {
329 0, 0, 0, 0, 0, 0, 0, 0, 20, 21, 22, 23, 24, 25, 26, 27,
330 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 37, 38, 39, 40, 41, 42,
331 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 56, 57,
332 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73,
333 74, 75, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88,
334 89, 90, 91, 92, 93, 94, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103,
335 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 114, 115, 116, 117, 118,
336 119, 120, 121, 122, 123, 124, 125, 126, 127, 128, 129, 130, 131, 132, 133, 133,
337 134, 135, 136, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 147, 148, 149,
338 150, 151, 152, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164,
339 165, 166, 167, 168, 169, 170, 171, 171, 172, 173, 174, 175, 176, 177, 178, 179,
340 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 190, 191, 192, 194, 194,
341 195, 196, 197, 198, 199, 200, 201, 202, 202, 204, 205, 206, 207, 208, 209, 209,
342 210, 211, 212, 213, 215, 215, 216, 217, 218, 219, 220, 220, 222, 223, 224, 225,
343 226, 227, 227, 229, 229, 230, 231, 232, 234, 234, 235, 236, 237, 238, 239, 240,
344 241, 242, 243, 244, 245, 246, 247, 248, 248, 0, 0, 0, 0, 0, 0, 0};
349 Range Coding of integers:
350 -------------------------
356 left and top are set to the respective blocks unless they are outside of
357 the image in which case they are set to the Null block
359 top-left is set to the top left block unless it is outside of the image in
360 which case it is set to the left block
362 if this block has no larger parent block or it is at the left side of its
363 parent block and the top right block is not outside of the image then the
364 top right block is used for top-right else the top-left block is used
368 level, ref, mx and my are 0
371 Motion Vector Prediction:
372 =========================
373 1. the motion vectors of all the neighboring blocks are scaled to
374 compensate for the difference of reference frames
376 scaled_mv= (mv * (256 * (current_reference+1) / (mv.reference+1)) + 128)>>8
378 2. the median of the scaled left, top and top-right vectors is used as
379 motion vector prediction
381 3. the used motion vector is the sum of the predictor and
382 (mvx_diff, mvy_diff)*mv_scale
386 ======================
387 the luma and chroma values of the left block are used as predictors
389 the used luma and chroma is the sum of the predictor and y_diff, cb_diff, cr_diff
390 to reverse this in the decoder apply the following:
391 block[y][x].dc[0] = block[y][x-1].dc[0] + y_diff;
392 block[y][x].dc[1] = block[y][x-1].dc[1] + cb_diff;
393 block[y][x].dc[2] = block[y][x-1].dc[2] + cr_diff;
394 block[*][-1].dc[*]= 128;
400 Halfpel interpolation:
401 ----------------------
402 halfpel interpolation is done by convolution with the halfpel filter stored
405 horizontal halfpel samples are found by
406 H1[y][x] = hcoeff[0]*(F[y][x ] + F[y][x+1])
407 + hcoeff[1]*(F[y][x-1] + F[y][x+2])
408 + hcoeff[2]*(F[y][x-2] + F[y][x+3])
410 h1[y][x] = (H1[y][x] + 32)>>6;
412 vertical halfpel samples are found by
413 H2[y][x] = hcoeff[0]*(F[y ][x] + F[y+1][x])
414 + hcoeff[1]*(F[y-1][x] + F[y+2][x])
416 h2[y][x] = (H2[y][x] + 32)>>6;
418 vertical+horizontal halfpel samples are found by
419 H3[y][x] = hcoeff[0]*(H2[y][x ] + H2[y][x+1])
420 + hcoeff[1]*(H2[y][x-1] + H2[y][x+2])
422 H3[y][x] = hcoeff[0]*(H1[y ][x] + H1[y+1][x])
423 + hcoeff[1]*(H1[y+1][x] + H1[y+2][x])
425 h3[y][x] = (H3[y][x] + 2048)>>12;
436 F-------F-------F-> H1<-F-------F-------F
440 F-------F-------F-> H1<-F-------F-------F
451 unavailable fullpel samples (outside the picture for example) shall be equal
452 to the closest available fullpel sample
455 Smaller pel interpolation:
456 --------------------------
457 if diag_mc is set then points which lie on a line between 2 vertically,
458 horiziontally or diagonally adjacent halfpel points shall be interpolated
459 linearls with rounding to nearest and halfway values rounded up.
460 points which lie on 2 diagonals at the same time should only use the one
461 diagonal not containing the fullpel point
465 F-->O---q---O<--h1->O---q---O<--F
473 h2-->O---q---O<--h3->O---q---O<--h2
481 F-->O---q---O<--h1->O---q---O<--F
485 the remaining points shall be bilinearly interpolated from the
486 up to 4 surrounding halfpel and fullpel points, again rounding should be to
487 nearest and halfway values rounded up
489 compliant Snow decoders MUST support 1-1/8 pel luma and 1/2-1/16 pel chroma
490 interpolation at least
493 Overlapped block motion compensation:
494 -------------------------------------
499 Each sample in the LL0 subband is predicted by the median of the left, top and
500 left+top-topleft samples, samples outside the subband shall be considered to
501 be 0. To reverse this prediction in the decoder apply the following.
502 for(y=0; y<height; y++){
503 for(x=0; x<width; x++){
504 sample[y][x] += median(sample[y-1][x],
506 sample[y-1][x]+sample[y][x-1]-sample[y-1][x-1]);
509 sample[-1][*]=sample[*][-1]= 0;
510 width,height here are the width and height of the LL0 subband not of the final
521 Snow supports 2 wavelet transforms, the symmetric biorthogonal 5/3 integer
522 transform and a integer approximation of the symmetric biorthogonal 9/7
525 2D IDWT (inverse discrete wavelet transform)
526 --------------------------------------------
527 The 2D IDWT applies a 2D filter recursively, each time combining the
528 4 lowest frequency subbands into a single subband until only 1 subband
530 The 2D filter is done by first applying a 1D filter in the vertical direction
531 and then applying it in the horizontal one.
532 --------------- --------------- --------------- ---------------
533 |LL0|HL0| | | | | | | | | | | |
534 |---+---| HL1 | | L0|H0 | HL1 | | LL1 | HL1 | | | |
535 |LH0|HH0| | | | | | | | | | | |
536 |-------+-------|->|-------+-------|->|-------+-------|->| L1 | H1 |->...
537 | | | | | | | | | | | |
538 | LH1 | HH1 | | LH1 | HH1 | | LH1 | HH1 | | | |
539 | | | | | | | | | | | |
540 --------------- --------------- --------------- ---------------
545 1. interleave the samples of the low and high frequency subbands like
546 s={L0, H0, L1, H1, L2, H2, L3, H3, ... }
547 note, this can end with a L or a H, the number of elements shall be w
548 s[-1] shall be considered equivalent to s[1 ]
549 s[w ] shall be considered equivalent to s[w-2]
551 2. perform the lifting steps in order as described below
554 1. s[i] -= (s[i-1] + s[i+1] + 2)>>2; for all even i < w
555 2. s[i] += (s[i-1] + s[i+1] )>>1; for all odd i < w
557 \ | /|\ | /|\ | /|\ | /|\
558 \|/ | \|/ | \|/ | \|/ |
560 /|\ | /|\ | /|\ | /|\ |
561 / | \|/ | \|/ | \|/ | \|/
565 Snow's 9/7 Integer filter:
566 1. s[i] -= (3*(s[i-1] + s[i+1]) + 4)>>3; for all even i < w
567 2. s[i] -= s[i-1] + s[i+1] ; for all odd i < w
568 3. s[i] += ( s[i-1] + s[i+1] + 4*s[i] + 8)>>4; for all even i < w
569 4. s[i] += (3*(s[i-1] + s[i+1]) )>>1; for all odd i < w
571 \ | /|\ | /|\ | /|\ | /|\
572 \|/ | \|/ | \|/ | \|/ |
574 /|\ | /|\ | /|\ | /|\ |
575 / | \|/ | \|/ | \|/ | \|/
576 (| + (| + (| + (| + -1
577 \ + /|\ + /|\ + /|\ + /|\ +1/4
578 \|/ | \|/ | \|/ | \|/ |
579 + | + | + | + | +1/16
580 /|\ | /|\ | /|\ | /|\ |
581 / | \|/ | \|/ | \|/ | \|/
585 following are exactly identical
586 (3a)>>1 == a + (a>>1)
587 (a + 4b + 8)>>4 == ((a>>2) + b + 2)>>2
589 16bit implementation note:
590 The IDWT can be implemented with 16bits, but this requires some care to
591 prevent overflows, the following list, lists the minimum number of bits needed
594 A= s[i-1] + s[i+1] 16bit
599 s[i-1] + s[i+1] 17bit
602 3*(s[i-1] + s[i+1]) 17bit
608 finetune initial contexts
610 try to use the wavelet transformed predicted image (motion compensated image) as context for coding the residual coefficients
611 try the MV length as context for coding the residual coefficients
612 use extradata for stuff which is in the keyframes now?
613 the MV median predictor is patented IIRC
614 implement per picture halfpel interpolation
615 try different range coder state transition tables for different contexts
618 compare the 6 tap and 8 tap hpel filters (psnr/bitrate and subjective quality)
619 spatial_scalability b vs u (!= 0 breaks syntax anyway so we can add a u later)
630 GPL + GFDL + whatever is needed to make this a RFC