profilelevel.tex

   1 \label{profilelevel}
   2
   3 A Dirac decoder shall support one or more different profiles and levels.
   4 Profiles and levels determine which tools, syntax elements and structures
   5 shall be supported, and what decoder resources (computational and memory) are required.
   6
   7 \subsection{Profiles}
   8 \label{profiles}
   9 A given profile requires that particular syntax/syntax elements shall
  10 be used and that decoder variables or functions shall be set to particular values.
  11
  12 Dirac defines four profiles, Main (Long GOP), Main (Intra), Simple and Low Delay,
  13 corresponding to different picture types. The Main (Intra), Simple and Low Delay
  14 profiles shall correspond to the Main, Simple and Low Delay profiles of VC-2.
  15
  16 The profiles shall satisfy the following conditions:
  17 \begin{itemize}
  18 \item A Low Delay profile Dirac sequence shall set $\Profile$ equal to a value of
  19 $0$ in the parse parameters (Section~\ref{parseparameters}) for each sequence header
  20 in the sequence.
  21 \item A Simple profile sequence shall set $\Profile$ equal to a value of $1$
  22 in the parse parameters for each sequence header in the sequence.
  23 \item A Main (Intra) profile sequence shall set $\Profile$ equal to a value of $2$
  24 in the parse parameters for each sequence header in the sequence.
  25 \item A Main (Long GOP) profile sequence shall set $\Profile$ equal to a value of $8$
  26 in the parse parameters for each sequence header in the sequence.
  27 \end{itemize}
  28
  29 Further VC-2 compatible profiles may be added in future revisions of this
  30 specification with profile number less than 8; other profiles may be added with
  31 profile number greater than 8.
  32
  33 A Dirac sequence shall comply with one of the supported profiles.
  34
  35 \subsubsection{Low Delay profile}
  36
  37 A Low Delay profile sequence shall contain only those data units whose
  38 parse codes are listed in Table \ref{table:ldprofile}.
  39
  40 \begin{table}[!ht]
  41 \centering
  42 \begin{tabular}{|c|c|l|c|}
  43 \hline
  44 \rowcolor[gray]{0.75}
  45 \ParseCode &  {\bf Bits} & {\bf Description} & \begin{tabular}{c} {\bf Number of}\\ {\bf Reference}\\{\bf Pictures}\end{tabular}\\
  46 \hline
  47 0x00 & 0000 0000 & Sequence header &--\\
  48 \hline
  49 0x10 & 0001 0000 & End of Sequence & -- \\
  50 \hline
  51 0x20 & 0010 0000 & Auxiliary data & -- \\
  52 \hline
  53 0x30 & 0011 0000 & Padding data & -- \\
  54 \hline
  55 0xC8 & 1100 1000 & Intra Non Reference Picture & 0\\
  56 \hline
  57 \end{tabular}
  58 \caption{Parse code values for Low Delay profile sequences}\label{table:ldprofile}
  59 \end{table}
  60
  61 \subsubsection{Simple profile}
  62
  63 A Simple profile sequence shall contain only those data units whose
  64 parse codes are listed in Table \ref{table:simpleprofile}.
  65
  66 \begin{table}[!ht]
  67 \centering
  68 \begin{tabular}{|c|c|l|c|}
  69 \hline
  70 \rowcolor[gray]{0.75}
  71 \ParseCode &  {\bf Bits} & {\bf Description} & \begin{tabular}{c} {\bf Number of}\\ {\bf Reference}\\{\bf Pictures}\end{tabular}\\
  72 \hline
  73 0x00 & 0000 0000 & Sequence header &--\\
  74 \hline
  75 0x10 & 0001 0000 & End of Sequence & -- \\
  76 \hline
  77 0x20 & 0010 0000 & Auxiliary data & -- \\
  78 \hline
  79 0x30 & 0011 0000 & Padding data & -- \\
  80 \hline
  81 0x48 & 0100 1000 & Intra Non Reference Picture (no arithmetic coding) & 0\\
  82 \hline
  83 \end{tabular}
  84 \caption{Parse code values for Simple profile sequences}\label{table:simpleprofile}
  85 \end{table}
  86
  87 \subsubsection{Main (Intra) profile}
  88
  89 A Main (Intra) profile sequence shall contain only those data units whose
  90 parse codes are listed in Table \ref{table:mainintraprofile}.
  91
  92 \begin{table}[!ht]
  93 \centering
  94 \begin{tabular}{|c|c|l|c|}
  95 \hline
  96 \rowcolor[gray]{0.75}
  97 \ParseCode &  {\bf Bits} & {\bf Description} & \begin{tabular}{c} {\bf Number of}\\ {\bf Reference}\\{\bf Pictures}\end{tabular}\\
  98 \hline
  99 0x00 & 0000 0000 & Sequence header &--\\
 100 \hline
 101 0x10 & 0001 0000 & End of Sequence & -- \\
 102 \hline
 103 0x20 & 0010 0000 & Auxiliary data & -- \\
 104 \hline
 105 0x30 & 0011 0000 & Padding data & -- \\
 106 \hline
 107 0x08 & 0000 1000 & Intra Non Reference Picture (arithmetic coding) & 0\\
 108 \hline
 109 \end{tabular}
 110 \caption{Parse code values for Main (Intra)  profile sequences}\label{table:mainintraprofile}
 111 \end{table}
 112
 113 \subsubsection{Main (Long GOP) profile [TBD]}
 114
 115 A Main (Long GOP) profile sequence shall contain only those data units whose
 116 parse codes are listed in Table \ref{table:mainlonggopprofile}.
 117
 118 \begin{table}[!ht]
 119 \centering
 120 \begin{tabular}{|c|c|l|c|}
 121 \hline
 122 \rowcolor[gray]{0.75}
 123 \ParseCode &  {\bf Bits} & {\bf Description} & \begin{tabular}{c} {\bf Number of}\\ {\bf Reference}\\{\bf Pictures}\end{tabular}\\
 124 \hline
 125 0x00 & 0000 0000 & Sequence header &--\\
 126 \hline
 127 0x10 & 0001 0000 & End of Sequence & -- \\
 128 \hline
 129 0x20 & 0010 0000 & Auxiliary data & -- \\
 130 \hline
 131 0x30 & 0011 0000 & Padding data & -- \\
 132 \hline
 133 0x0C & 0000 1100 & Intra Reference Picture (arithmetic coding) & 0\\
 134 \hline
 135 0x08 & 0000 1000 & Intra Non Reference Picture (arithmetic coding) & 0\\
 136 \hline
 137 0x0D & 0000 1101 & Inter Reference Picture (arithmetic coding) & 1\\
 138 \hline
 139 0x0E & 0000 1110 & Inter Reference Picture (arithmetic coding) & 2\\
 140 \hline
 141 0x09 & 0000 1001 & Inter Non Reference Picture (arithmetic coding)& 1\\
 142 \hline
 143 0x0A & 0000 1010 & Inter Non Reference Picture (arithmetic coding) & 2\\
 144 \hline
 145 \end{tabular}
 146 \caption{Parse code values for Main (Long GOP)  profile sequences}
 147 \label{table:mainlonggopprofile}
 148 \end{table}
 149
 150 Low delay syntax pictures shall not be present in a Main (Long GOP) profile
 151 sequence.
 152
 153 \subsection{Levels}
 154 \label{level}
 155
 156 A given value of level shall define constraints on the decoder
 157 resources required to decode a compliant sequence. The values that are
 158 constrained are defined individually for each level.
 159
 160 A particular application domain may impose additional constraints
 161 on a decoder, for example the presence or absence of a suitable video
 162 interface. Such additional constraints are not covered by this specification.
 163
 164 This specification defines levels 1 and 128. Other levels are application
 165 specific and will be defined in future revisions of this specification. VC-2
 166 compatible levels will have level number less than 128 and other levels
 167 will have level number greater than 128.
 168
 169 The value 0 shall be RESERVED and shall not be used for any defined level.
 170 Sequences may use the value 0 for streams that do not conform to any defined level.
 171
 172 For level 1, the video parameters shall correspond
 173 to one of the base video formats as defined in Annex~\ref{videoformatdefaults}
 174 and the video parameters
 175 of the base video format shall not be overridden. Level 1 shall be available
 176 for Low Delay, Simple and Main (Intra) (VC-2 compatible) profiles only.
 177
 178 For level 128, the video parameters shall correspond to one of the base video
 179 formats as defined in Annex~\ref{videoformatdefaults},
 180 except that the signal range is restricted
 181 to 8 bit ranges and the chroma sampling format and frame dimensions may be
 182 overridden by suitable values. Level 128 shall be available for Main (Long GOP)
 183 profile only.
 184
 185 A sequence compliant with a given level $n$ shall set $\Level$ equal to $n$.
 186
 187 \subsubsection{Decoder data buffers[DRAFT-TBC]}
 188
 189 A reference picture buffer $\RefBuffer$ is defined for storing
 190 decoded reference pictures (if any). In addition, levels may define
 191 two additional buffers, and applicable parameters for them:
 192
 193 \begin{itemize}
 194 \item a bit stream buffer $\StreamBuffer$ for buffering
 195 stream data prior to decoding
 196 \item a decoded picture buffer $\DecodedBuffer$ for storing
 197 decoded pictures (reference or non-reference) for the purposes
 198 of picture reordering
 199 \end{itemize}
 200
 201 For the purposes of this specification, the reference picture
 202 buffer shall be deemed to be additional to the
 203 decoded picture buffer (i.e. reference picture storage will be
 204 duplicated in the decoded picture buffer).
 205
 206 \paragraph{Bit stream buffer operation[TBC]}$\ $\newline
 207
 208 \paragraph{Picture reordering and decoded picture buffer[TBC]}$\ $\newline
 209 \label{picturereordering}
 210
 211 Two parameters shall be defined for constraining picture reordering.
 212 The first is the size, $\DPBSize$, of the decoded picture buffer.
 213
 214 The second is the reordering depth applicable to a Dirac sequence. This shall
 215 be defined as the maximum number of picture data units that may occur
 216 between a picture with picture number $N$ and a picture with picture number
 217 $N+1$ or $N-1$.
 218
 219 \subsubsection{Buffer models [TBC]}
 220
 221
 222 \subsubsection{Level 1: VC-2 default level}
 223
 224 This level is intended to provide minimal constraints on VC-2 compatible
 225 streams encoding one of the base video formats
 226 (Annex~\ref{videoformatdefaults}). Data rates in
 227 this level are not bounded. This level is not intended to provide
 228 guarantees of real time decoding.
 229
 230 This level shall only be available if the profile is set to Low Delay, Simple, or
 231 Main (Intra).
 232
 233 \paragraph{Sequence header parameters}$\ $\newline
 234
 235 The following constraints shall apply to the sequence header parameters
 236 (Section~\ref{sequenceheader}):
 237 \begin{itemize}
 238 \item$base\_video\_format$ shall be between 1 and 20 inclusive
 239 \item $custom\_dimensions\_flag$ shall be $\false$
 240 \item $custom\_chroma\_format\_flag$ shall be $\false$
 241 \item$custom\_scan\_format\_flag$ shall be $\false$
 242 \item $custom\_frame\_rate\_flag$ shall be $\false$
 243 \item $custom\_pixel\_aspect\_ratio\_flag$ be $\false$
 244 \item $custom\_clean\_area\_flag$ shall be $\false$
 245 \item $custom\_signal\_range\_flag$ shall be $\false$
 246 \item $custom\_color\_spec\_flag$ shall be $\false$
 247 \item $picture\_coding\_mode$ shall be as per Section~\ref{picturecodingmode}
 248 \end{itemize}
 249
 250 \paragraph{Picture header parameters}$\ $\newline
 251
 252 The following constraints shall apply to picture header parameters
 253 (Section~\ref{pictureheader}):
 254 \begin{itemize}
 255 \item $\WaveletIndex$ shall be as table \ref{wltfilterpresets}
 256 \item $\TransformDepth$ shall be between 0 and 4 inclusive
 257 \item for core syntax pictures:
 258     \begin{itemize}
 259     \item $\CodeblocksX$ and $\CodeblocksY$ shall be such that there is
 260     at least one coefficient in each codeblock.
 261     \end{itemize}
 262 \item for low delay syntax pictures:
 263     \begin{itemize}
 264     \item $\SlicesX$ and $\SlicesY$ shall be such that there is at least one
 265     DC (0-LL) coefficient per slice.
 266     \item values in the quantizer matrix shall lie between 0 and 127 inclusive
 267     \end{itemize}
 268 \end{itemize}
 269
 270 \paragraph{Transform data}$\ $\newline
 271
 272 The following constraints shall apply to transform data elements
 273 (Section~\ref{wavelettransform}):
 274 \begin{itemize}
 275 \item quantized and inverse quantized wavelet coefficients shall lie
 276 between $-2^{19}$ and $2^{19}$ inclusive
 277 \item for core syntax pictures:
 278     \begin{itemize}
 279     \item quantization indices for subbands and codeblocks shall lie between
 280     0 and 68 inclusive
 281     \item quantization offsets encoded in codeblocks shall lie between
 282     -68 and 68 inclusive
 283     \end{itemize}
 284 \end{itemize}
 285
 286 \subsubsection{Level 128: Long-GOP default level [DRAFT-TBD]}
 287
 288 This level is intended to provide minimal constraints on long GOP
 289 streams encoding simple variants of the base video formats
 290 (Annex~\ref{videoformatdefaults}).
 291 Data rates in this level are not bounded. This level is not intended to provide
 292 guarantees of real time decoding.
 293
 294 This level shall only be available if the profile is set to Main (Long GOP).
 295
 296 \paragraph{Sequence header parameters}$\ $\newline
 297
 298 The following constraints shall apply to the sequence header parameters
 299 (Section~\ref{sequenceheader}):
 300 \begin{itemize}
 301 \item$base\_video\_format$ shall be between 1 and 20 inclusive
 302 \item $custom\_dimensions\_flag$ may be $\false$ or $\true$; if $\true$ then
 303 the frame width and frame height shall be less than the values set in the
 304 base video format
 305 \item $custom\_chroma\_format\_flag$ may be $\false$ or $\true$
 306 \item$custom\_scan\_format\_flag$ may be $\false$ or $\true$
 307 \item $custom\_frame\_rate\_flag$ shall be $\false$
 308 \item $custom\_pixel\_aspect\_ratio\_flag$ be $\false$
 309 \item $custom\_clean\_area\_flag$ shall be $\false$
 310 \item $custom\_signal\_range\_flag$ shall be $\true$ and the signal range
 311 parameters set to 8 bit SDI ranges (index 2 in table \ref{table:signalrangevalues})
 312 \item $custom\_color\_spec\_flag$ shall be $\false$
 313 \item $picture\_coding\_mode$ shall be as per Section~\ref{picturecodingmode}
 314 \end{itemize}
 315
 316 \paragraph{Picture header parameters}$\ $\newline
 317
 318 The following constraints shall apply to picture header parameters
 319 (Section~\ref{pictureheader}):
 320 \begin{itemize}
 321 \item $\WaveletIndex$ shall be between 0 and 4 inclusive
 322 \item $\TransformDepth$ shall be between 0 and 4 inclusive
 323 \item $\CodeblocksX$ and $\CodeblocksY$ shall be such that there is
 324     at least one coefficient in each codeblock
 325 \end{itemize}
 326
 327 \paragraph{Transform data}$\ $\newline
 328
 329 The following constraints shall apply to transform data elements
 330 (Section~\ref{wavelettransform}):
 331 \begin{itemize}
 332 \item quantized and inverse quantized wavelet coefficients shall lie
 333 between $-2^{15}$ and $2^{15}$ inclusive
 334 \item quantization indices for subbands and codeblocks shall lie between
 335     0 and 63 inclusive
 336 \item quantization offsets encoded in codeblocks shall lie between
 337     -63 and 63 inclusive
 338 \end{itemize}
 339
 340 \paragraph{Reordering and reference buffers [DRAFT-TBD]}$\ $\newline
 341
 342
 343 A decoder must maintain a reference picture buffer for the
 344 purposes of inter picture prediction, and (separately) a decoded
 345 picture buffer for the purposes of picture reordering, in addition to
 346 storage provided for decoding the current picture. The decoded picture
 347 buffer shall contain non-reference pictures only.
 348
 349 A bit stream buffer is not required for this level.
 350
 351 The following constraints shall apply:
 352 \begin{itemize}
 353 \item The reference buffer size $\MaxRefBufferSize$ shall be 3.
 354 \item The decoded picture buffer size shall be 2 in the case
 355 of frame coding or 4 in the case of field coding.
 356 \item The picture reordering depth shall be 5 in the case of frame coding
 357 or 11 in the case of field coding
 358 \end{itemize}
 359
 360
 361
 362
 363
 364
 365
 366
 367
 368
 369
 370
 371
 372