This spec focuses on the different points of extension where the existing D3D12 Video Encode API needs new structures to support AV1 Encode. The rest of the D3D12 Encode API will remain unmodified for this feature unless explicited in this spec.
The next sections detail the API and DDI for video encoding. In many cases, the DDI is extremely similar to the API. The structures and enumerations which are basically the same (differing solely in name convention) are not repeated in the this specification. We include just the DDI structures/enumerations and functions that differ substantially from the API.
typedef enum D3D12_VIDEO_ENCODER_SUPPORT_FLAGS
{
...
D3D12_VIDEO_ENCODER_SUPPORT_FLAG_RATE_CONTROL_EXTENSION1_SUPPORT = 0x2000,
D3D12_VIDEO_ENCODER_SUPPORT_FLAG_RATE_CONTROL_QUALITY_VS_SPEED_AVAILABLE = 0x4000,
} D3D12_VIDEO_ENCODER_SUPPORT_FLAGS;D3D12_VIDEO_ENCODER_SUPPORT_FLAG_RATE_CONTROL_EXTENSION1_SUPPORT
When enabled, indicates the use of D3D12_VIDEO_ENCODER_RATE_CONTROL_FLAG_ENABLE_EXTENSION1_SUPPORT is available.
D3D12_VIDEO_ENCODER_SUPPORT_FLAG_RATE_CONTROL_QUALITY_VS_SPEED_AVAILABLE
Requires D3D12_VIDEO_ENCODER_SUPPORT_FLAG_RATE_CONTROL_EXTENSION1_SUPPORT. When enabled, indicates the use of D3D12_VIDEO_ENCODER_RATE_CONTROL_FLAG_ENABLE_QUALITY_VS_SPEED is available.
typedef enum D3D12_VIDEO_ENCODER_RATE_CONTROL_FLAGS
{
...
D3D12_VIDEO_ENCODER_RATE_CONTROL_FLAG_ENABLE_EXTENSION1_SUPPORT = 0x40,
D3D12_VIDEO_ENCODER_RATE_CONTROL_FLAG_ENABLE_QUALITY_VS_SPEED = 0x80,
} D3D12_VIDEO_ENCODER_RATE_CONTROL_FLAGS;D3D12_VIDEO_ENCODER_RATE_CONTROL_FLAG_ENABLE_EXTENSION1_SUPPORT
Requires D3D12_VIDEO_ENCODER_SUPPORT_FLAG_RATE_CONTROL_EXTENSION1_SUPPORT. Indicates when enabled that the extended rate control structures will be used in *D3D12_VIDEO_ENCODER_RATE_CONTROL_CONFIGURATION_PARAMS.pConfiguration_* or legacy when disabled, as per table below.
When DISABLED:
| Rate control mode | D3D12_VIDEO_ENCODER_RATE_CONTROL_CONFIGURATION_PARAMS type | D3D12_VIDEO_ENCODER_RATE_CONTROL_CONFIGURATION_PARAMS DataSize |
|---|---|---|
| D3D12_VIDEO_ENCODER_RATE_CONTROL_MODE_ABSOLUTE_QP_MAP | NULL | 0 |
| D3D12_VIDEO_ENCODER_RATE_CONTROL_MODE_CQP | D3D12_VIDEO_ENCODER_RATE_CONTROL_CQP | sizeof(D3D12_VIDEO_ENCODER_RATE_CONTROL_CQP) |
| D3D12_VIDEO_ENCODER_RATE_CONTROL_MODE_CBR | D3D12_VIDEO_ENCODER_RATE_CONTROL_CBR | sizeof(D3D12_VIDEO_ENCODER_RATE_CONTROL_CBR) |
| D3D12_VIDEO_ENCODER_RATE_CONTROL_MODE_VBR | D3D12_VIDEO_ENCODER_RATE_CONTROL_VBR | sizeof(D3D12_VIDEO_ENCODER_RATE_CONTROL_VBR) |
| D3D12_VIDEO_ENCODER_RATE_CONTROL_MODE_QVBR | D3D12_VIDEO_ENCODER_RATE_CONTROL_QVBR | sizeof(D3D12_VIDEO_ENCODER_RATE_CONTROL_QVBR) |
When ENABLED:
| Rate control mode | D3D12_VIDEO_ENCODER_RATE_CONTROL_CONFIGURATION_PARAMS type | D3D12_VIDEO_ENCODER_RATE_CONTROL_CONFIGURATION_PARAMS DataSize |
|---|---|---|
| D3D12_VIDEO_ENCODER_RATE_CONTROL_MODE_ABSOLUTE_QP_MAP | D3D12_VIDEO_ENCODER_RATE_CONTROL_ABSOLUTE_QP_MAP | sizeof(D3D12_VIDEO_ENCODER_RATE_CONTROL_ABSOLUTE_QP_MAP) |
| D3D12_VIDEO_ENCODER_RATE_CONTROL_MODE_CQP | D3D12_VIDEO_ENCODER_RATE_CONTROL_CQP1 | sizeof(D3D12_VIDEO_ENCODER_RATE_CONTROL_CQP1) |
| D3D12_VIDEO_ENCODER_RATE_CONTROL_MODE_CBR | D3D12_VIDEO_ENCODER_RATE_CONTROL_CBR1 | sizeof(D3D12_VIDEO_ENCODER_RATE_CONTROL_CBR1) |
| D3D12_VIDEO_ENCODER_RATE_CONTROL_MODE_VBR | D3D12_VIDEO_ENCODER_RATE_CONTROL_VBR1 | sizeof(D3D12_VIDEO_ENCODER_RATE_CONTROL_VBR1) |
| D3D12_VIDEO_ENCODER_RATE_CONTROL_MODE_QVBR | D3D12_VIDEO_ENCODER_RATE_CONTROL_QVBR1 | sizeof(D3D12_VIDEO_ENCODER_RATE_CONTROL_QVBR1) |
D3D12_VIDEO_ENCODER_RATE_CONTROL_FLAG_ENABLE_QUALITY_VS_SPEED
Requires D3D12_VIDEO_ENCODER_SUPPORT_FLAG_RATE_CONTROL_EXTENSION1_SUPPORT and D3D12_VIDEO_ENCODER_SUPPORT_FLAG_RATE_CONTROL_QUALITY_VS_SPEED_AVAILABLE. When enabled, indicates the use of QualityVsSpeed in the rate control structure.
The following D3D12_VIDEO_ENCODER_RATE_CONTROL_* structures are added. QualityVsSpeed is added to all modes and VBVCapacity, InitialVBVFullness are added to QVBR1 in addition.
QualityVsSpeed must be in the range [0, D3D12_FEATURE_DATA_VIDEO_ENCODER_SUPPORT1.MaxQualityVsSpeed]. The lower the value, the fastest the encode operation, similar as defined in the MF API.
The settings associated to each of the levels exposed by QualityVsSpeed must only refer to hardware/driver implementation optimizations and heuristics that are not related to specific codec configurations or encoding tools selection, which are already independently exposed in the D3D12 API to the user individually. Please note that other codec configurations and codec encoding tools exposed through this API may also affect quality and speed.
typedef struct D3D12_VIDEO_ENCODER_RATE_CONTROL_CQP1
{
UINT ConstantQP_FullIntracodedFrame;
UINT ConstantQP_InterPredictedFrame_PrevRefOnly;
UINT ConstantQP_InterPredictedFrame_BiDirectionalRef;
UINT QualityVsSpeed;
} D3D12_VIDEO_ENCODER_RATE_CONTROL_CQP1;typedef struct D3D12_VIDEO_ENCODER_RATE_CONTROL_CBR1
{
UINT InitialQP;
UINT MinQP;
UINT MaxQP;
UINT64 MaxFrameBitSize;
UINT64 TargetBitRate;
UINT64 VBVCapacity;
UINT64 InitialVBVFullness;
UINT QualityVsSpeed;
} D3D12_VIDEO_ENCODER_RATE_CONTROL_CBR1;typedef struct D3D12_VIDEO_ENCODER_RATE_CONTROL_VBR1
{
UINT InitialQP;
UINT MinQP;
UINT MaxQP;
UINT64 MaxFrameBitSize;
UINT64 TargetAvgBitRate;
UINT64 PeakBitRate;
UINT64 VBVCapacity;
UINT64 InitialVBVFullness;
UINT QualityVsSpeed;
} D3D12_VIDEO_ENCODER_RATE_CONTROL_VBR1;typedef struct D3D12_VIDEO_ENCODER_RATE_CONTROL_QVBR1
{
UINT InitialQP;
UINT MinQP;
UINT MaxQP;
UINT64 MaxFrameBitSize;
UINT64 TargetAvgBitRate;
UINT64 PeakBitRate;
UINT64 VBVCapacity;
UINT64 InitialVBVFullness;
UINT ConstantQualityTarget;
UINT QualityVsSpeed;
} D3D12_VIDEO_ENCODER_RATE_CONTROL_QVBR1;typedef struct D3D12_VIDEO_ENCODER_RATE_CONTROL_ABSOLUTE_QP_MAP {
UINT QualityVsSpeed;
} D3D12_VIDEO_ENCODER_RATE_CONTROL_ABSOLUTE_QP_MAP;typedef struct D3D12_VIDEO_ENCODER_RATE_CONTROL_CONFIGURATION_PARAMS
{
UINT DataSize;
union
{
...
const D3D12_VIDEO_ENCODER_RATE_CONTROL_CQP *pConfiguration_CQP;
const D3D12_VIDEO_ENCODER_RATE_CONTROL_CBR *pConfiguration_CBR;
const D3D12_VIDEO_ENCODER_RATE_CONTROL_VBR *pConfiguration_VBR;
const D3D12_VIDEO_ENCODER_RATE_CONTROL_QVBR *pConfiguration_QVBR;
// Extension 1 structs
const D3D12_VIDEO_ENCODER_RATE_CONTROL_CQP1 *pConfiguration_CQP1;
const D3D12_VIDEO_ENCODER_RATE_CONTROL_CBR1 *pConfiguration_CBR1;
const D3D12_VIDEO_ENCODER_RATE_CONTROL_VBR1 *pConfiguration_VBR1;
const D3D12_VIDEO_ENCODER_RATE_CONTROL_QVBR1 *pConfiguration_QVBR1;
const D3D12_VIDEO_ENCODER_RATE_CONTROL_ABSOLUTE_QP_MAP* pConfiguration_AbsoluteQPMap;
};
} D3D12_VIDEO_ENCODER_RATE_CONTROL_CONFIGURATION_PARAMS;typedef enum D3D12_FEATURE_VIDEO
{
…
D3D12_FEATURE_VIDEO_ENCODER_FRAME_SUBREGION_LAYOUT_CONFIG = 46,
D3D12_FEATURE_VIDEO_ENCODER_SUPPORT1 = 47,
} D3D12_FEATURE_VIDEO;New members
D3D12_FEATURE_VIDEO_ENCODER_FRAME_SUBREGION_LAYOUT_CONFIG
Used with struct D3D12_FEATURE_DATA_VIDEO_ENCODER_FRAME_SUBREGION_LAYOUT_CONFIG.
D3D12_FEATURE_VIDEO_ENCODER_SUPPORT1
Used with struct D3D12_FEATURE_DATA_VIDEO_ENCODER_SUPPORT1.
typedef enum D3D12_VIDEO_ENCODER_CODEC
{
...
D3D12_VIDEO_ENCODER_CODEC_AV1 = 3,
} D3D12_VIDEO_ENCODER_CODEC;typedef enum D3D12_VIDEO_ENCODER_AV1_PROFILE
{
D3D12_VIDEO_ENCODER_AV1_PROFILE_MAIN = 0,
D3D12_VIDEO_ENCODER_AV1_PROFILE_HIGH = 1,
D3D12_VIDEO_ENCODER_AV1_PROFILE_PROFESSIONAL = 2,
} D3D12_VIDEO_ENCODER_AV1_PROFILE;Remarks
Driver uses D3D12_FEATURE_DATA_VIDEO_ENCODER_INPUT_FORMAT to report optionally supported formats for a given input D3D12_VIDEO_ENCODER_AV1_PROFILE to the query.
When a format is supported, the API Client sets accordingly the related AV1 syntax elements based on the DXGI_FORMAT used: color_config(), high_bitdepth, twelve_bit, mono_chrome, subsampling_x, subsampling_y, etc.
The DXGI_FORMATs are consistent with the AV1 DXVA decode counterpart spec.
D3D12_VIDEO_ENCODER_AV1_PROFILE_MAINallows driver reporting support for:DXGI_FORMAT_NV12for 8-bit YUV 4:2:0DXGI_FORMAT_P010for 10-bit YUV 4:2:0DXGI_FORMAT_R8_UNORMfor 8-bit monochromeDXGI_FORMAT_R16_UNORMfor 10-bit monochrome
D3D12_VIDEO_ENCODER_AV1_PROFILE_HIGHalso allows:DXGI_FORMAT_AYUVfor 8-bit YUV 4:4:4DXGI_FORMAT_Y410for 10-bit YUV 4:4:4
D3D12_VIDEO_ENCODER_AV1_PROFILE_PROFESSIONALalso allows:DXGI_FORMAT_YUY2for 8-bit YUV 4:2:2DXGI_FORMAT_Y210for 10-bit YUV 4:2:2DXGI_FORMAT_Y216for 12-bit YUV 4:2:2DXGI_FORMAT_Y416for 12-bit YUV 4:4:4DXGI_FORMAT_R16_UNORMfor 12-bit monochromeDXGI_FORMAT_R16_UNORMfor 12-bit monochromeDXGI_FORMAT_P016for 12-bit YUV 4:2:0
typedef struct D3D12_VIDEO_ENCODER_PROFILE_DESC
{
UINT DataSize;
union
{
...
D3D12_VIDEO_ENCODER_AV1_PROFILE* pAV1Profile;
};
} D3D12_VIDEO_ENCODER_PROFILE_DESC;Generic structure for codec profiles.
typedef enum D3D12_VIDEO_ENCODER_AV1_LEVELS
{
D3D12_VIDEO_ENCODER_AV1_LEVELS_2_0 = 0,
D3D12_VIDEO_ENCODER_AV1_LEVELS_2_1 = 1,
D3D12_VIDEO_ENCODER_AV1_LEVELS_2_2 = 2,
D3D12_VIDEO_ENCODER_AV1_LEVELS_2_3 = 3,
D3D12_VIDEO_ENCODER_AV1_LEVELS_3_0 = 4,
D3D12_VIDEO_ENCODER_AV1_LEVELS_3_1 = 5,
D3D12_VIDEO_ENCODER_AV1_LEVELS_3_2 = 6,
D3D12_VIDEO_ENCODER_AV1_LEVELS_3_3 = 7,
D3D12_VIDEO_ENCODER_AV1_LEVELS_4_0 = 8,
D3D12_VIDEO_ENCODER_AV1_LEVELS_4_1 = 9,
D3D12_VIDEO_ENCODER_AV1_LEVELS_4_2 = 10,
D3D12_VIDEO_ENCODER_AV1_LEVELS_4_3 = 11,
D3D12_VIDEO_ENCODER_AV1_LEVELS_5_0 = 12,
D3D12_VIDEO_ENCODER_AV1_LEVELS_5_1 = 13,
D3D12_VIDEO_ENCODER_AV1_LEVELS_5_2 = 14,
D3D12_VIDEO_ENCODER_AV1_LEVELS_5_3 = 15,
D3D12_VIDEO_ENCODER_AV1_LEVELS_6_0 = 16,
D3D12_VIDEO_ENCODER_AV1_LEVELS_6_1 = 17,
D3D12_VIDEO_ENCODER_AV1_LEVELS_6_2 = 18,
D3D12_VIDEO_ENCODER_AV1_LEVELS_6_3 = 19,
D3D12_VIDEO_ENCODER_AV1_LEVELS_7_0 = 20,
D3D12_VIDEO_ENCODER_AV1_LEVELS_7_1 = 21,
D3D12_VIDEO_ENCODER_AV1_LEVELS_7_2 = 22,
D3D12_VIDEO_ENCODER_AV1_LEVELS_7_3 = 23,
} D3D12_VIDEO_ENCODER_AV1_LEVELS;typedef enum D3D12_VIDEO_ENCODER_AV1_TIER
{
D3D12_VIDEO_ENCODER_AV1_TIER_MAIN = 0,
D3D12_VIDEO_ENCODER_AV1_TIER_HIGH = 1,
} D3D12_VIDEO_ENCODER_AV1_TIER;typedef struct D3D12_VIDEO_ENCODER_AV1_LEVEL_TIER_CONSTRAINTS
{
D3D12_VIDEO_ENCODER_AV1_LEVELS Level;
D3D12_VIDEO_ENCODER_AV1_TIER Tier;
} D3D12_VIDEO_ENCODER_AV1_LEVEL_TIER_CONSTRAINTS;typedef struct D3D12_VIDEO_ENCODER_LEVEL_SETTING
{
UINT DataSize;
union
{
...
D3D12_VIDEO_ENCODER_AV1_LEVEL_TIER_CONSTRAINTS* pAV1LevelSetting;
};
} D3D12_VIDEO_ENCODER_LEVEL_SETTING;typedef enum D3D12_VIDEO_ENCODER_FRAME_SUBREGION_LAYOUT_MODE
{
D3D12_VIDEO_ENCODER_FRAME_SUBREGION_LAYOUT_MODE_UNIFORM_GRID_PARTITION = 5,
D3D12_VIDEO_ENCODER_FRAME_SUBREGION_LAYOUT_MODE_CONFIGURABLE_GRID_PARTITION = 6,
} D3D12_VIDEO_ENCODER_FRAME_SUBREGION_LAYOUT_MODE;Constants
D3D12_VIDEO_ENCODER_FRAME_SUBREGION_LAYOUT_MODE_UNIFORM_GRID_PARTITION
Allows the driver to uniformly partition the frame into a grid with only input from the API Client being number of rows and columns. Driver will return the heights and widths of each cell in the partitioned grid after the execution of the EncodeFrame command in resolved metadata buffer.
For the AV1 codec, This corresponds to the AV1 spec syntax uniform_tile_spacing_flag equal to 1, when using D3D12_VIDEO_ENCODER_FRAME_SUBREGION_LAYOUT_MODE_UNIFORM_GRID_PARTITION, tiles are uniformly partitioned except the right and bottom edges. When the dimensions cannot be partitioned exactly, the last tile can have smaller size.
D3D12_VIDEO_ENCODER_FRAME_SUBREGION_LAYOUT_MODE_CONFIGURABLE_GRID_PARTITION
Allows the API Client to fully customize a grid partition of the frame. API Client will pass a list of rows and columns along with the heights and widths of each cell in the partitioned grid in the EncodeFrame command and they have to be honored exactly.
For the AV1 codec, this corresponds to uniform_tile_spacing_flag equal to 0 means that the tile sizes are coded.
Remarks
For further tile support details, please check the associated tile details support cap D3D12_FEATURE_DATA_VIDEO_ENCODER_FRAME_SUBREGION_LAYOUT_CONFIG.
typedef
enum D3D12_VIDEO_ENCODER_MOTION_ESTIMATION_PRECISION_MODE
{
...
D3D12_VIDEO_ENCODER_MOTION_ESTIMATION_PRECISION_MODE_EIGHTH_PIXEL = 4;
} D3D12_VIDEO_ENCODER_MOTION_ESTIMATION_PRECISION_MODE;typedef enum D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAGS
{
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_NONE = 0x0,
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_128x128_SUPERBLOCK = 0x1,
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_FILTER_INTRA = 0x2,
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_INTRA_EDGE_FILTER = 0x4,
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_INTERINTRA_COMPOUND = 0x8,
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_MASKED_COMPOUND = 0x10,
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_WARPED_MOTION = 0x20,
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_DUAL_FILTER = 0x40,
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_JNT_COMP = 0x80,
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_FORCED_INTEGER_MOTION_VECTORS = 0x100,
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_SUPER_RESOLUTION = 0x200,
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_LOOP_RESTORATION_FILTER = 0x400,
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_PALETTE_ENCODING = 0x800,
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_CDEF_FILTERING = 0x1000,
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_INTRA_BLOCK_COPY = 0x2000,
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_FRAME_REFERENCE_MOTION_VECTORS = 0x4000,
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_ORDER_HINT_TOOLS = 0x8000,
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_AUTO_SEGMENTATION = 0x10000,
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_CUSTOM_SEGMENTATION = 0x20000,
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_LOOP_FILTER_DELTAS = 0x40000,
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_QUANTIZATION_DELTAS = 0x80000,
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_QUANTIZATION_MATRIX = 0x100000,
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_REDUCED_TX_SET = 0x200000,
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_MOTION_MODE_SWITCHABLE = 0x400000,
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_ALLOW_HIGH_PRECISION_MV = 0x800000,
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_SKIP_MODE_PRESENT = 0x1000000,
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_DELTA_LF_PARAMS = 0x2000000,
} D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAGS;Reports encoding capabilities for AV1.
Constants
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_128x128_SUPERBLOCK
Indicates if support is available for 128x128 Superblocks.
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_FILTER_INTRA
Indicates if support is available for Intra prediction filter.
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_INTRA_EDGE_FILTER
Indicates if support is available for intra edge filtering process.
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_INTERINTRA_COMPOUND
Indicates if support is available for interintra, where the mode info for inter blocks may contain the syntax element interintra. Equal to 0 specifies that the syntax element interintra will not be present.
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_MASKED_COMPOUND
Indicates if support is available for masked compound, where the mode info for inter blocks may contain the syntax element compound_type. Equal to 0 specifies that the syntax element compound_type will not be present.
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_WARPED_MOTION
Equal to 1 indicates that the syntax element motion_mode may be present. If equal to 0 indicates that the syntax element motion_mode will not be present (this means that LOCALWARP cannot be signaled if this flag is equal to 0).
Related to AV1 syntax enable_warped_motion in the sequence header.
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_DUAL_FILTER
Indicates if support is available for dual filter mode, where the inter prediction filter type may be specified independently in the horizontal and vertical directions. If the flag is equal to 0, only one filter type may be specified, which is then used in both directions.
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_JNT_COMP
Indicates if support is available for the scenario where distance weights process may be used for inter prediction.
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_FORCED_INTEGER_MOTION_VECTORS
Indicates if support is available for using the syntax element force_integer_mv.
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_SUPER_RESOLUTION
Indicates if support is available for super resolution.
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_LOOP_RESTORATION_FILTER
Indicates if support is available for loop restoration filtering.
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_PALETTE_ENCODING
Indicates if support is available for frame level control on palette encoding; Equal to 0 indicates that palette encoding is never used.
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_CDEF_FILTERING
Indicates if support is available for constrained directional enhancement filtering.
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_INTRA_BLOCK_COPY
Indicates if intra block copy is supported or not at frame level. Same syntax as AV1 spec.
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_FRAME_REFERENCE_MOTION_VECTORS
Indicates if support is available for use_ref_frame_mvs to be configured on a per frame basis. Equal to 0 specifies that use_ref_frame_mvs syntax element will not be used.
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_ORDER_HINT_TOOLS
Indicates if support is available for usage of tools based on the values of order hints. Equal to 0 indicates that tools based on order hints are not supported and can't be enabled.
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_AUTO_SEGMENTATION
Indicates if the driver can perform segmentation without API Client input and return segmentation_params() information in D3D12_VIDEO_ENCODER_AV1_POST_ENCODE_VALUES. Driver will write the segment map in the compressed bitstream.
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_CUSTOM_SEGMENTATION
Indicates if the driver supports the API Client passing customized segmentation segmentation_params() as well as the segment map and driver will honor exactly.
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_LOOP_FILTER_DELTAS
Indicates if the driver supports use of loop filter deltas. Related to loop_filter_delta_enabled AV1 syntax in loop_filter_params().
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_QUANTIZATION_DELTAS
Indicates if the driver supports use of quantization delta syntax.
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_QUANTIZATION_MATRIX
Indicates if the driver supports use of quantization matrix syntax.
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_REDUCED_TX_SET
Indicates if driver supports setting reduced_tx_set in the frame header or must be always set to zero.
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_MOTION_MODE_SWITCHABLE
Indicates if driver supports setting is_motion_mode_switchable in the frame header or must be always set to zero.
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_ALLOW_HIGH_PRECISION_MV
Indicates if driver supports setting allow_high_precision_mv in the frame header or must be always set to zero.
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_SKIP_MODE_PRESENT
Indicates if driver supports setting skip_mode_present in the frame header or must be always set to zero.
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_DELTA_LF_PARAMS
Indicates if the driver supports use of loop filter delta params syntax. Related to delta_lf_params() AV1 syntax.
typedef enum D3D12_VIDEO_ENCODER_AV1_TX_MODE {
D3D12_VIDEO_ENCODER_AV1_TX_MODE_ONLY4x4 = 0,
D3D12_VIDEO_ENCODER_AV1_TX_MODE_LARGEST = 1,
D3D12_VIDEO_ENCODER_AV1_TX_MODE_SELECT = 2,
} D3D12_VIDEO_ENCODER_AV1_TX_MODE;typedef enum D3D12_VIDEO_ENCODER_AV1_TX_MODE_FLAGS {
D3D12_VIDEO_ENCODER_AV1_TX_MODE_FLAG_NONE = 0x0,
D3D12_VIDEO_ENCODER_AV1_TX_MODE_FLAG_ONLY4x4 = (1 << D3D12_VIDEO_ENCODER_AV1_TX_MODE_ONLY4x4),
D3D12_VIDEO_ENCODER_AV1_TX_MODE_FLAG_LARGEST = (1 << D3D12_VIDEO_ENCODER_AV1_TX_MODE_LARGEST),
D3D12_VIDEO_ENCODER_AV1_TX_MODE_FLAG_SELECT = (1 << D3D12_VIDEO_ENCODER_AV1_TX_MODE_SELECT),
} D3D12_VIDEO_ENCODER_AV1_TX_MODE_FLAGS;typedef enum D3D12_VIDEO_ENCODER_AV1_INTERPOLATION_FILTERS {
D3D12_VIDEO_ENCODER_AV1_INTERPOLATION_FILTERS_EIGHTTAP = 0,
D3D12_VIDEO_ENCODER_AV1_INTERPOLATION_FILTERS_EIGHTTAP_SMOOTH = 1,
D3D12_VIDEO_ENCODER_AV1_INTERPOLATION_FILTERS_EIGHTTAP_SHARP = 2,
D3D12_VIDEO_ENCODER_AV1_INTERPOLATION_FILTERS_BILINEAR = 3,
D3D12_VIDEO_ENCODER_AV1_INTERPOLATION_FILTERS_SWITCHABLE = 4,
} D3D12_VIDEO_ENCODER_AV1_INTERPOLATION_FILTERS;typedef enum D3D12_VIDEO_ENCODER_AV1_INTERPOLATION_FILTERS_FLAGS {
D3D12_VIDEO_ENCODER_AV1_INTERPOLATION_FILTERS_FLAG_NONE = 0x0,
D3D12_VIDEO_ENCODER_AV1_INTERPOLATION_FILTERS_FLAG_EIGHTTAP = (1 << D3D12_VIDEO_ENCODER_AV1_INTERPOLATION_FILTERS_EIGHTTAP),
D3D12_VIDEO_ENCODER_AV1_INTERPOLATION_FILTERS_FLAG_EIGHTTAP_SMOOTH = (1 << D3D12_VIDEO_ENCODER_AV1_INTERPOLATION_FILTERS_EIGHTTAP_SMOOTH),
D3D12_VIDEO_ENCODER_AV1_INTERPOLATION_FILTERS_FLAG_EIGHTTAP_SHARP = (1 << D3D12_VIDEO_ENCODER_AV1_INTERPOLATION_FILTERS_EIGHTTAP_SHARP),
D3D12_VIDEO_ENCODER_AV1_INTERPOLATION_FILTERS_FLAG_BILINEAR = (1 << D3D12_VIDEO_ENCODER_AV1_INTERPOLATION_FILTERS_BILINEAR),
D3D12_VIDEO_ENCODER_AV1_INTERPOLATION_FILTERS_FLAG_SWITCHABLE = (1 << D3D12_VIDEO_ENCODER_AV1_INTERPOLATION_FILTERS_SWITCHABLE),
} D3D12_VIDEO_ENCODER_AV1_INTERPOLATION_FILTERS_FLAGS;typedef enum D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_BLOCK_SIZE {
D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_BLOCK_SIZE_4x4 = 0,
D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_BLOCK_SIZE_8x8 = 1,
D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_BLOCK_SIZE_16x16 = 2,
D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_BLOCK_SIZE_32x32 = 3,
D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_BLOCK_SIZE_64x64 = 4,
} D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_BLOCK_SIZE;typedef enum D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_MODE {
D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_MODE_DISABLED = 0,
D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_MODE_ALT_Q = 1,
D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_MODE_ALT_LF_Y_V = 2,
D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_MODE_ALT_LF_Y_H = 3,
D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_MODE_ALT_LF_U = 4,
D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_MODE_ALT_LF_V = 5,
D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_MODE_ALT_REF_FRAME = 6,
D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_MODE_ALT_SKIP = 7,
D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_MODE_ALT_GLOBALMV = 8,
} D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_MODE;typedef enum D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_MODE_FLAGS {
D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_MODE_FLAG_NONE = 0,
D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_MODE_FLAG_DISABLED = (1 << D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_MODE_DISABLED),
D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_MODE_FLAG_ALT_Q = (1 << D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_MODE_ALT_Q),
D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_MODE_FLAG_ALT_LF_Y_V = (1 << D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_MODE_ALT_LF_Y_V),
D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_MODE_FLAG_ALT_LF_Y_H = (1 << D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_MODE_ALT_LF_Y_H),
D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_MODE_FLAG_ALT_LF_U = (1 << D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_MODE_ALT_LF_U),
D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_MODE_FLAG_ALT_LF_V = (1 << D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_MODE_ALT_LF_V),
D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_MODE_FLAG_REF_FRAME = (1 << D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_MODE_ALT_REF_FRAME),
D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_MODE_FLAG_ALT_SKIP = (1 << D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_MODE_ALT_SKIP),
D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_MODE_FLAG_ALT_GLOBALMV = (1 << D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_MODE_ALT_GLOBALMV),
} D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_MODE_FLAGS;typedef enum D3D12_VIDEO_ENCODER_AV1_RESTORATION_TYPE {
D3D12_VIDEO_ENCODER_AV1_RESTORATION_TYPE_DISABLED = 0,
D3D12_VIDEO_ENCODER_AV1_RESTORATION_TYPE_SWITCHABLE = 1,
D3D12_VIDEO_ENCODER_AV1_RESTORATION_TYPE_WIENER = 2,
D3D12_VIDEO_ENCODER_AV1_RESTORATION_TYPE_SGRPROJ = 3,
} D3D12_VIDEO_ENCODER_AV1_RESTORATION_TYPE;typedef enum D3D12_VIDEO_ENCODER_AV1_RESTORATION_TILESIZE {
D3D12_VIDEO_ENCODER_AV1_RESTORATION_TILESIZE_DISABLED = 0,
D3D12_VIDEO_ENCODER_AV1_RESTORATION_TILESIZE_32x32 = 1,
D3D12_VIDEO_ENCODER_AV1_RESTORATION_TILESIZE_64x64 = 2,
D3D12_VIDEO_ENCODER_AV1_RESTORATION_TILESIZE_128x128 = 3,
D3D12_VIDEO_ENCODER_AV1_RESTORATION_TILESIZE_256x256 = 4,
} D3D12_VIDEO_ENCODER_AV1_RESTORATION_TILESIZE;Corresponds to the the size of loop restoration units in units of samples in the current plane. The enum values are based on lr_unit_shift and lr_uv_shift in lr_params() AV1 syntax and the RESTORATION_TILESIZE_MAX(256) AV1 spec constant.
typedef enum D3D12_VIDEO_ENCODER_AV1_RESTORATION_SUPPORT_FLAGS {
D3D12_VIDEO_ENCODER_AV1_RESTORATION_SUPPORT_FLAG_NOT_SUPPORTED = 0,
D3D12_VIDEO_ENCODER_AV1_RESTORATION_SUPPORT_FLAG_32x32 = 0x1,
D3D12_VIDEO_ENCODER_AV1_RESTORATION_SUPPORT_FLAG_64x64 = 0x2,
D3D12_VIDEO_ENCODER_AV1_RESTORATION_SUPPORT_FLAG_128x128 = 0x4,
D3D12_VIDEO_ENCODER_AV1_RESTORATION_SUPPORT_FLAG_256x256 = 0x8,
} D3D12_VIDEO_ENCODER_AV1_RESTORATION_SUPPORT_FLAGS;typedef enum D3D12_VIDEO_ENCODER_AV1_REFERENCE_WARPED_MOTION_TRANSFORMATION
{
D3D12_VIDEO_ENCODER_AV1_REFERENCE_WARPED_MOTION_TRANSFORMATION_IDENTITY = 0,
D3D12_VIDEO_ENCODER_AV1_REFERENCE_WARPED_MOTION_TRANSFORMATION_TRANSLATION = 1,
D3D12_VIDEO_ENCODER_AV1_REFERENCE_WARPED_MOTION_TRANSFORMATION_ROTZOOM = 2,
D3D12_VIDEO_ENCODER_AV1_REFERENCE_WARPED_MOTION_TRANSFORMATION_AFFINE = 3,
} D3D12_VIDEO_ENCODER_AV1_REFERENCE_WARPED_MOTION_TRANSFORMATION;Constants
D3D12_VIDEO_ENCODER_AV1_REFERENCE_WARPED_MOTION_TRANSFORMATION_IDENTITY
Identity transformation. 0 parameters in D3D12_VIDEO_ENCODER_AV1_REFERENCE_PICTURE_WARPED_MOTION_INFO.
D3D12_VIDEO_ENCODER_AV1_REFERENCE_WARPED_MOTION_TRANSFORMATION_TRANSLATION
Translational motion. 2 parameters in D3D12_VIDEO_ENCODER_AV1_REFERENCE_PICTURE_WARPED_MOTION_INFO.
D3D12_VIDEO_ENCODER_AV1_REFERENCE_WARPED_MOTION_TRANSFORMATION_ROTZOOM
Simplified affine with rotation with zoom. 4 parameters in D3D12_VIDEO_ENCODER_AV1_REFERENCE_PICTURE_WARPED_MOTION_INFO.
D3D12_VIDEO_ENCODER_AV1_REFERENCE_WARPED_MOTION_TRANSFORMATION_AFFINE
Affine transform. 6 parameters in D3D12_VIDEO_ENCODER_AV1_REFERENCE_PICTURE_WARPED_MOTION_INFO.
typedef enum D3D12_VIDEO_ENCODER_AV1_REFERENCE_WARPED_MOTION_TRANSFORMATION_FLAGS {
D3D12_VIDEO_ENCODER_AV1_REFERENCE_WARPED_MOTION_TRANSFORMATION_FLAG_NONE = 0,
D3D12_VIDEO_ENCODER_AV1_REFERENCE_WARPED_MOTION_TRANSFORMATION_FLAG_IDENTITY = (1 << D3D12_VIDEO_ENCODER_AV1_REFERENCE_WARPED_MOTION_TRANSFORMATION_IDENTITY),
D3D12_VIDEO_ENCODER_AV1_REFERENCE_WARPED_MOTION_TRANSFORMATION_FLAG_TRANSLATION = (1 << D3D12_VIDEO_ENCODER_AV1_REFERENCE_WARPED_MOTION_TRANSFORMATION_TRANSLATION),
D3D12_VIDEO_ENCODER_AV1_REFERENCE_WARPED_MOTION_TRANSFORMATION_FLAG_ROTZOOM = (1 << D3D12_VIDEO_ENCODER_AV1_REFERENCE_WARPED_MOTION_TRANSFORMATION_ROTZOOM),
D3D12_VIDEO_ENCODER_AV1_REFERENCE_WARPED_MOTION_TRANSFORMATION_FLAG_AFFINE = (1 << D3D12_VIDEO_ENCODER_AV1_REFERENCE_WARPED_MOTION_TRANSFORMATION_AFFINE),
} D3D12_VIDEO_ENCODER_AV1_REFERENCE_WARPED_MOTION_TRANSFORMATION_FLAGS;Remarks
If only D3D12_VIDEO_ENCODER_AV1_REFERENCE_WARPED_MOTION_TRANSFORMATION_FLAG_NONE is supported, reference warp motion arguments are ignored and AV1 is_global syntax is false for all references.
typedef enum D3D12_VIDEO_ENCODER_AV1_POST_ENCODE_VALUES_FLAGS {
D3D12_VIDEO_ENCODER_AV1_POST_ENCODE_VALUES_FLAG_NONE = 0,
D3D12_VIDEO_ENCODER_AV1_POST_ENCODE_VALUES_FLAG_QUANTIZATION = 0x1,
D3D12_VIDEO_ENCODER_AV1_POST_ENCODE_VALUES_FLAG_QUANTIZATION_DELTA = 0x2,
D3D12_VIDEO_ENCODER_AV1_POST_ENCODE_VALUES_FLAG_LOOP_FILTER = 0x4,
D3D12_VIDEO_ENCODER_AV1_POST_ENCODE_VALUES_FLAG_LOOP_FILTER_DELTA = 0x8,
D3D12_VIDEO_ENCODER_AV1_POST_ENCODE_VALUES_FLAG_CDEF_DATA = 0x10,
D3D12_VIDEO_ENCODER_AV1_POST_ENCODE_VALUES_FLAG_CONTEXT_UPDATE_TILE_ID = 0x20,
D3D12_VIDEO_ENCODER_AV1_POST_ENCODE_VALUES_FLAG_COMPOUND_PREDICTION_MODE = 0x40,
D3D12_VIDEO_ENCODER_AV1_POST_ENCODE_VALUES_FLAG_PRIMARY_REF_FRAME = 0x80,
D3D12_VIDEO_ENCODER_AV1_POST_ENCODE_VALUES_FLAG_REFERENCE_INDICES = 0x100,
} D3D12_VIDEO_ENCODER_AV1_POST_ENCODE_VALUES_FLAGS;Specifies for which AV1 encoding features, the underlying encoder is able to override (partially or totally) the associated AV1 syntax values or honor API Client exact configuration input otherwise.
When the bitflag is SET for a given feature, the driver receives the related API Client input and is able to override all or certain parameters of the associated structure with the given reported flag, which will then write back in D3D12_VIDEO_ENCODER_AV1_POST_ENCODE_VALUES with the final values for the API Client to re-pack the AV1 headers accordingly. API Client can compare this to the associated input structure to determine the driver changes (if any).
When the bitflag is NOT SET for a given feature, the driver honors the related API Client input exactly and copies the input values in D3D12_VIDEO_ENCODER_AV1_POST_ENCODE_VALUES. This way the client can always copy the post encode values to pack the headers directly.
Constants
D3D12_VIDEO_ENCODER_AV1_POST_ENCODE_VALUES_FLAG_QUANTIZATION
Related to D3D12_VIDEO_ENCODER_CODEC_AV1_QUANTIZATION_CONFIG values. Used to code quantization_params().
D3D12_VIDEO_ENCODER_AV1_POST_ENCODE_VALUES_FLAG_QUANTIZATION_DELTA
Related to D3D12_VIDEO_ENCODER_CODEC_AV1_QUANTIZATION_DELTA_CONFIG values. Used to code delta_q_params().
D3D12_VIDEO_ENCODER_AV1_POST_ENCODE_VALUES_FLAG_LOOP_FILTER
Related to D3D12_VIDEO_ENCODER_CODEC_AV1_LOOP_FILTER_CONFIG values. Used to code AV1 syntax loop_filter_params().
D3D12_VIDEO_ENCODER_AV1_POST_ENCODE_VALUES_FLAG_LOOP_FILTER_DELTA
Related to D3D12_VIDEO_ENCODER_CODEC_AV1_LOOP_FILTER_DELTA_CONFIG values. Used to code AV1 syntax delta_lf_params().
D3D12_VIDEO_ENCODER_AV1_POST_ENCODE_VALUES_FLAG_CDEF_DATA
Related to D3D12_VIDEO_ENCODER_AV1_CDEF_CONFIG values. Used to code AV1 syntax cdef_params().
D3D12_VIDEO_ENCODER_AV1_POST_ENCODE_VALUES_FLAG_CONTEXT_UPDATE_TILE_ID
Related to ContextUpdateTileId element in D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_SUBREGIONS_LAYOUT_DATA_TILES. Used to code AV1 element syntax context_update_tile_id in tile_info().
D3D12_VIDEO_ENCODER_AV1_POST_ENCODE_VALUES_FLAG_COMPOUND_PREDICTION_MODE
Related to D3D12_VIDEO_ENCODER_AV1_COMP_PREDICTION_TYPE values.
Remarks
- When API Client selects D3D12_VIDEO_ENCODER_AV1_COMP_PREDICTION_TYPE_COMPOUND_REFERENCE and this flag is set, the driver will return D3D12_VIDEO_ENCODER_AV1_COMP_PREDICTION_TYPE in post encode values. The returned value must be used to code reference_select = 0 (SINGLE) or reference_select = 1 (COMPOUND) syntax accordingly.
- When API Client selects D3D12_VIDEO_ENCODER_AV1_COMP_PREDICTION_TYPE_COMPOUND_SINGLE and this flag is set, the driver will return D3D12_VIDEO_ENCODER_AV1_COMP_PREDICTION_TYPE_COMPOUND_SINGLE and reference_select must be coded as 0 (SINGLE).
D3D12_VIDEO_ENCODER_AV1_POST_ENCODE_VALUES_FLAG_PRIMARY_REF_FRAME
Related to PrimaryRefFrame element in D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_CODEC_DATA. Used to code AV1 element syntax primary_ref_frame in uncompressed_header().
D3D12_VIDEO_ENCODER_AV1_POST_ENCODE_VALUES_FLAG_REFERENCE_INDICES
When the flag is reported by the driver, the driver may reorder/remap (but not change the number of references) of the D3D12_VIDEO_ENCODER_AV1_POST_ENCODE_VALUES.ReferenceIndices array output, based on the user input D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_CODEC_DATA.ReferenceIndices. Otherwise, driver must copy each array entry of this post encode output parameter as-is from D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_CODEC_DATA.ReferenceIndices.
API Client will write the picture header ref_frame_idx AV1 syntax from this output parameter.
typedef struct D3D12_VIDEO_ENCODER_AV1_CODEC_CONFIGURATION_SUPPORT {
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAGS SupportedFeatureFlags;
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAGS RequiredFeatureFlags;
D3D12_VIDEO_ENCODER_AV1_INTERPOLATION_FILTERS_FLAGS SupportedInterpolationFilters;
D3D12_VIDEO_ENCODER_AV1_RESTORATION_SUPPORT_FLAGS SupportedRestorationParams[3][3];
D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_MODE_FLAGS SupportedSegmentationModes;
D3D12_VIDEO_ENCODER_AV1_TX_MODE_FLAGS SupportedTxModes[4];
D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_BLOCK_SIZE SegmentationBlockSize;
D3D12_VIDEO_ENCODER_AV1_POST_ENCODE_VALUES_FLAGS PostEncodeValuesFlags;
UINT MaxTemporalLayers;
UINT MaxSpatialLayers;
} D3D12_VIDEO_ENCODER_AV1_CODEC_CONFIGURATION_SUPPORT;Members
SupportedFeatureFlags
Output param. Indicates which features are supported for the codec. Supported features can be set or not by the API Client.
RequiredFeatureFlags
Output param. Indicates which features the driver requires to be set.
SupportedInterpolationFilters
Output parameter. Indicates which values can be selected on input parameters of type D3D12_VIDEO_ENCODER_AV1_INTERPOLATION_FILTERS.
SupportedRestorationParams
Output parameter. Indicates which values can be selected as input parameters for FrameRestorationType and LoopRestorationPixelSize in D3D12_VIDEO_ENCODER_AV1_RESTORATION_CONFIG.
The first array indexing corresponds to the restoration filter type:
| Index i in SupportedRestorationParams[i][j] | Filter type |
|---|---|
| 0 | SWITCHABLE |
| 1 | WIENER |
| 2 | SGRPROJ |
Note the indexing of the filter types corresponds to D3D12_VIDEO_ENCODER_AV1_RESTORATION_TYPE minus 1 (skipping D3D12_VIDEO_ENCODER_AV1_RESTORATION_TYPE_DISABLED).
The second array indexing corresponds to the planes:
| Index j in SupportedRestorationParams[i][j] | Plane |
|---|---|
| 0 | Y plane |
| 1 | U plane |
| 2 | V plane |
The value returned in SupportedRestorationParams[i][j] is a bitflag mask indicating whether the i-th filter in the j-th plane is either:
- Not supported indicated by SupportedRestorationParams[i][j] = D3D12_VIDEO_ENCODER_AV1_RESTORATION_SUPPORT_FLAG_NOT_SUPPORTED.
- Supported with any of the D3D12_VIDEO_ENCODER_AV1_RESTORATION_TILESIZE as indicated by the combinable bit flags in SupportedRestorationParams[i][j].
SupportedSegmentationModes
Output parameter. Indicates which segmentation modes can be selected in D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_CONFIG.
SupportedTxModes
Output parameter. Indicates which values can be selected on input parameters of type D3D12_VIDEO_ENCODER_AV1_TX_MODE for different D3D12_VIDEO_ENCODER_AV1_FRAME_TYPE.
Note: Driver must support at least 1 mode for each frame type (ie. mask value cannot be 0).
SegmentationBlockSize
Output parameter. Indicates the block size for the segment map. This is both the for input blocks in D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_MAP for custom segmentation or the block size of the segment map written in the compressed bitstream by the driver in auto segmentation.
PostEncodeValuesFlags
Specifies for which AV1 encoding features, the underlying encoder is able to override the associated AV1 syntax values or accept API Client configurable input exactly.
MaxTemporalLayers
Specifies the maximum number of temporal layers that can be supported. The reported values must be in the range [1..MaxTemporalIdSupported + 1]. A reported value 1, there is no temporal scalability support.
MaxSpatialLayers
Specifies the maximum number of spatial layers that can be supported. The reported values must be in the range [1..MaxSpatialIdSupported + 1]. A reported value 1, there is no spatial scalability support.
typedef struct D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_SUPPORT {
UINT DataSize;
union
{
...
D3D12_VIDEO_ENCODER_AV1_CODEC_CONFIGURATION_SUPPORT* pAV1Support;
};
} D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_SUPPORT;typedef enum D3D12_VIDEO_ENCODER_AV1_FRAME_TYPE
{
D3D12_VIDEO_ENCODER_AV1_FRAME_TYPE_KEY_FRAME = 0,
D3D12_VIDEO_ENCODER_AV1_FRAME_TYPE_INTER_FRAME = 1,
D3D12_VIDEO_ENCODER_AV1_FRAME_TYPE_INTRA_ONLY_FRAME = 2,
D3D12_VIDEO_ENCODER_AV1_FRAME_TYPE_SWITCH_FRAME = 3,
} D3D12_VIDEO_ENCODER_AV1_FRAME_TYPE;typedef enum D3D12_VIDEO_ENCODER_AV1_FRAME_TYPE_FLAGS {
D3D12_VIDEO_ENCODER_AV1_FRAME_TYPE_FLAG_NONE = 0x0,
D3D12_VIDEO_ENCODER_AV1_FRAME_TYPE_FLAG_KEY_FRAME = (1 << D3D12_VIDEO_ENCODER_AV1_FRAME_TYPE_KEY_FRAME),
D3D12_VIDEO_ENCODER_AV1_FRAME_TYPE_FLAG_INTER_FRAME = (1 << D3D12_VIDEO_ENCODER_AV1_FRAME_TYPE_INTER_FRAME),
D3D12_VIDEO_ENCODER_AV1_FRAME_TYPE_FLAG_INTRA_ONLY_FRAME = (1 << D3D12_VIDEO_ENCODER_AV1_FRAME_TYPE_INTRA_ONLY_FRAME),
D3D12_VIDEO_ENCODER_AV1_FRAME_TYPE_FLAG_SWITCH_FRAME = (1 << D3D12_VIDEO_ENCODER_AV1_FRAME_TYPE_SWITCH_FRAME),
} D3D12_VIDEO_ENCODER_AV1_FRAME_TYPE_FLAGS;typedef enum D3D12_VIDEO_ENCODER_AV1_COMP_PREDICTION_TYPE
{
D3D12_VIDEO_ENCODER_AV1_COMP_PREDICTION_TYPE_SINGLE_REFERENCE = 0,
D3D12_VIDEO_ENCODER_AV1_COMP_PREDICTION_TYPE_COMPOUND_REFERENCE = 1,
} D3D12_VIDEO_ENCODER_AV1_COMP_PREDICTION_TYPE;Constants
D3D12_VIDEO_ENCODER_AV1_COMP_PREDICTION_TYPE_SINGLE_REFERENCE
Indicates that all inter blocks will use single prediction. Equivalent to AV1 syntax reference_select equal to 0.
D3D12_VIDEO_ENCODER_AV1_COMP_PREDICTION_TYPE_COMPOUND_REFERENCE
Indicates that the mode info for inter blocks contains the syntax element comp_mode that indicates whether to use single or compound reference prediction. Equivalent to AV1 syntax reference_select equal to 1.
typedef struct D3D12_VIDEO_ENCODER_CODEC_AV1_PICTURE_CONTROL_SUPPORT {
D3D12_VIDEO_ENCODER_AV1_COMP_PREDICTION_TYPE PredictionMode;
UINT MaxUniqueReferencesPerFrame;
D3D12_VIDEO_ENCODER_AV1_FRAME_TYPE_FLAGS SupportedFrameTypes;
D3D12_VIDEO_ENCODER_AV1_REFERENCE_WARPED_MOTION_TRANSFORMATION_FLAGS SupportedReferenceWarpedMotionFlags;
} D3D12_VIDEO_ENCODER_CODEC_AV1_PICTURE_CONTROL_SUPPORT;Members
PredictionMode
Input param. The requested prediction mode to be used. The driver must return the output parameters below assuming a frame will be encoded using this prediction mode in the picture params structure.
MaxUniqueReferencesPerFrame
Output param. Indicates how many unique reference frames in the DPB can be selected at the same time for a given frame from any of the reference types (LAST, ..., ALTREF, etc) in the picture control parameters from the DPB that the API Client manages. In other words, the maximum number distinct (and with ReconstructedPictureResourceIndex != 0xFF) entries in D3D12_VIDEO_ENCODE_REFERENCE_FRAMES.ppTexture2Ds[ReferenceFramesReconPictureDescriptors[ReferenceIndices[i]].ReconstructedPictureResourceIndex] for i in [0..7].
SupportedFrameTypes
Output param. Indicates the supported frame types to be used in D3D12_VIDEO_ENCODER_AV1_FRAME_TYPE.
SupportedReferenceWarpedMotionFlags
Output param. Indicates the supported types to be used in D3D12_VIDEO_ENCODER_AV1_REFERENCE_PICTURE_WARPED_MOTION_INFO.TransformationType.
typedef struct D3D12_VIDEO_ENCODER_CODEC_PICTURE_CONTROL_SUPPORT {
UINT DataSize;
union
{
...
D3D12_VIDEO_ENCODER_CODEC_AV1_PICTURE_CONTROL_SUPPORT* pAV1Support;
};
} D3D12_VIDEO_ENCODER_CODEC_PICTURE_CONTROL_SUPPORT;typedef struct D3D12_VIDEO_ENCODER_AV1_CODEC_CONFIGURATION {
D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAGS FeatureFlags;
UINT OrderHintBitsMinus1;
} D3D12_VIDEO_ENCODER_AV1_CODEC_CONFIGURATION;Members
FeatureFlags
Defines the set of enabled features. Flags can be combined based on the reported capabilities/requirements by the driver.
Remarks
typedef enum D3D12_VIDEO_ENCODER_AV1_FRAME_SUBREGION_LAYOUT_CONFIG_VALIDATION_FLAGS
{
D3D12_VIDEO_ENCODER_AV1_FRAME_SUBREGION_LAYOUT_CONFIG_VALIDATION_FLAG_NONE = 0x0,
D3D12_VIDEO_ENCODER_AV1_FRAME_SUBREGION_LAYOUT_CONFIG_VALIDATION_FLAG_NOT_SPECIFIED = 0x1,
D3D12_VIDEO_ENCODER_AV1_FRAME_SUBREGION_LAYOUT_CONFIG_VALIDATION_FLAG_CODEC_CONSTRAINT = 0x2,
D3D12_VIDEO_ENCODER_AV1_FRAME_SUBREGION_LAYOUT_CONFIG_VALIDATION_FLAG_HARDWARE_CONSTRAINT = 0x4,
D3D12_VIDEO_ENCODER_AV1_FRAME_SUBREGION_LAYOUT_CONFIG_VALIDATION_FLAG_ROWS_COUNT = 0x8,
D3D12_VIDEO_ENCODER_AV1_FRAME_SUBREGION_LAYOUT_CONFIG_VALIDATION_FLAG_COLS_COUNT = 0x10,
D3D12_VIDEO_ENCODER_AV1_FRAME_SUBREGION_LAYOUT_CONFIG_VALIDATION_FLAG_WIDTH = 0x20,
D3D12_VIDEO_ENCODER_AV1_FRAME_SUBREGION_LAYOUT_CONFIG_VALIDATION_FLAG_AREA = 0x40,
D3D12_VIDEO_ENCODER_AV1_FRAME_SUBREGION_LAYOUT_CONFIG_VALIDATION_FLAG_TOTAL_TILES = 0x80,
} D3D12_VIDEO_ENCODER_AV1_FRAME_SUBREGION_LAYOUT_CONFIG_VALIDATION_FLAGS;Constants
D3D12_VIDEO_ENCODER_AV1_FRAME_SUBREGION_LAYOUT_CONFIG_VALIDATION_FLAG_NONE
No flags.
D3D12_VIDEO_ENCODER_AV1_FRAME_SUBREGION_LAYOUT_CONFIG_VALIDATION_FLAG_NOT_SPECIFIED
When this flag is set, indicates that the requested tiles configuration is not supported due to a reason not specified by any of the other flag categories.
D3D12_VIDEO_ENCODER_AV1_FRAME_SUBREGION_LAYOUT_CONFIG_VALIDATION_FLAG_CODEC_CONSTRAINT
When this flag is set, indicates that the requested tiles configuration is not supported due to codec constraints. An example on this for AV1 would be D3D12_FEATURE_DATA_VIDEO_ENCODER_FRAME_SUBREGION_LAYOUT_CONFIG.Level.
D3D12_VIDEO_ENCODER_AV1_FRAME_SUBREGION_LAYOUT_CONFIG_VALIDATION_FLAG_HARDWARE_CONSTRAINT
When this flag is set, indicates that the requested tiles configuration is not supported due to hardware constraints.
D3D12_VIDEO_ENCODER_AV1_FRAME_SUBREGION_LAYOUT_CONFIG_VALIDATION_FLAG_ROWS_COUNT
When this flag is set, indicates that the number of tile rows requested is not supported.
D3D12_VIDEO_ENCODER_AV1_FRAME_SUBREGION_LAYOUT_CONFIG_VALIDATION_FLAG_COLS_COUNT
When this flag is set, indicates that the number of tile columns requested is not supported.
D3D12_VIDEO_ENCODER_AV1_FRAME_SUBREGION_LAYOUT_CONFIG_VALIDATION_FLAG_WIDTH
When this flag is set, indicates that one or more tiles widths in the requested configuration is not supported.
D3D12_VIDEO_ENCODER_AV1_FRAME_SUBREGION_LAYOUT_CONFIG_VALIDATION_FLAG_AREA
When this flag is set, indicates that one or more tiles areas in the requested configuration is not supported.
D3D12_VIDEO_ENCODER_AV1_FRAME_SUBREGION_LAYOUT_CONFIG_VALIDATION_FLAG_TOTAL_TILES
When this flag is set, indicates that the total number of tiles in the requested partition exceeds the total supported tiles count. Please see D3D12_FEATURE_DATA_VIDEO_ENCODER_RESOLUTION_SUPPORT_LIMITS.MaxSubregionsNumber.
typedef struct D3D12_VIDEO_ENCODER_AV1_FRAME_SUBREGION_LAYOUT_CONFIG_SUPPORT
{
BOOL Use128SuperBlocks;
D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_SUBREGIONS_LAYOUT_DATA_TILES TilesConfiguration;
D3D12_VIDEO_ENCODER_AV1_FRAME_SUBREGION_LAYOUT_CONFIG_VALIDATION_FLAGS ValidationFlags;
UINT MinTileRows;
UINT MaxTileRows;
UINT MinTileCols;
UINT MaxTileCols;
UINT MinTileWidth;
UINT MaxTileWidth;
UINT MinTileArea;
UINT MaxTileArea;
UINT TileSizeBytesMinus1;
} D3D12_VIDEO_ENCODER_AV1_FRAME_SUBREGION_LAYOUT_CONFIG_SUPPORT;Members
Use128SuperBlocks
Input parameter. Indicates if the returned values by the driver in superblock units need to be expressed as 128x128 superblocks. Otherwise the superblock default size 64x64 must be used.
TilesConfiguration
Input parameter. Desired tile configuration to check support for.
ValidationFlags
Output parameter. Indicates more details when D3D12_FEATURE_DATA_VIDEO_ENCODER_FRAME_SUBREGION_LAYOUT_CONFIG.IsSupported is false.
MinTileRows
Output parameter. Minimum number of horizontal partitions.
MaxTileRows
Output parameter. Maximum number of horizontal partitions.
MinTileCols
Output parameter. Minimum number of vertical partitions.
MaxTileCols
Output parameter. Maximum number of vertical partitions.
MinTileWidth
Output parameter. Minimum width of any tile, in superblock units.
MaxTileWidth
Output parameter. Maximum width of any tile, in superblock units.
MinTileArea
Output parameter. Minimum dimension of any tile, in superblock units.
MaxTileArea
Output parameter. Maximum dimension of any tile, in superblock units.
TileSizeBytesMinus1
Output parameter. Specifies the number of bytes needed to code each tile size. Related to the driver writing the D3D12_VIDEO_ENCODER_FRAME_SUBREGION_METADATA.bSize elements in the resolved metadata. The API Client will write tile_size_bytes_minus_1 = (TileSizeBytesMinus1) in frame_header_obu/uncompressed_header/tile_info when writing the frame header OBU, and when writing le(TileSizeBytes) tile_size_minus_1 in tile_group_obu().
typedef struct D3D12_VIDEO_ENCODER_FRAME_SUBREGION_LAYOUT_CONFIG_SUPPORT {
UINT DataSize;
union
{
D3D12_VIDEO_ENCODER_AV1_FRAME_SUBREGION_LAYOUT_CONFIG_SUPPORT* pAV1Support;
};
} D3D12_VIDEO_ENCODER_FRAME_SUBREGION_LAYOUT_CONFIG_SUPPORT;typedef struct D3D12_FEATURE_DATA_VIDEO_ENCODER_FRAME_SUBREGION_LAYOUT_CONFIG
{
UINT NodeIndex; // input
D3D12_VIDEO_ENCODER_CODEC Codec; // input
D3D12_VIDEO_ENCODER_PROFILE_DESC Profile; // input
D3D12_VIDEO_ENCODER_LEVEL_SETTING Level; // input
D3D12_VIDEO_ENCODER_FRAME_SUBREGION_LAYOUT_MODE SubregionMode; // input
D3D12_VIDEO_ENCODER_PICTURE_RESOLUTION_DESC FrameResolution; // input
D3D12_VIDEO_ENCODER_FRAME_SUBREGION_LAYOUT_CONFIG_SUPPORT CodecSupport; // input/output
BOOL IsSupported; // output
} D3D12_FEATURE_DATA_VIDEO_ENCODER_FRAME_SUBREGION_LAYOUT_CONFIG;Remarks
At the moment, this support query is only supported for input Codec == D3D12_VIDEO_ENCODER_CODEC_AV1, as it can only be compiled with CodecSupport.pAV1Support pointer of type D3D12_VIDEO_ENCODER_AV1_FRAME_SUBREGION_LAYOUT_CONFIG_SUPPORT.
Members
NodeIndex
Input parameter. in multi-adapter operation, this indicates which physical adapter of the device this operation applies to.
Codec
Input parameter. Desired codec to check support for.
Profile
Input parameter, desired profile to check support for.
Level
Input parameter, desired level to check support for.
SubregionMode
Input parameter. Desired subregion mode to check support for.
FrameResolution
Input parameter. Indicates the resolution to check support for.
CodecSupport
Input/Output parameter. Indicates the codec specific input/output arguments for this cap.
IsSupported Output parameter. Indicates if the driver supports this subregion layout configuration based on the input values, including the codec, profile, level, resolution and any other codec specific parameters.
typedef struct D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION
{
UINT DataSize;
union
{
...
D3D12_VIDEO_ENCODER_AV1_CODEC_CONFIGURATION* pAV1Config;
};
} D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION;Adds a new flag to the existing D3D12_VIDEO_ENCODER_VALIDATION_FLAGS, for usage with the new extended query cap D3D12_FEATURE_DATA_VIDEO_ENCODER_SUPPORT1.
typedef enum D3D12_VIDEO_ENCODER_VALIDATION_FLAGS
{
D3D12_VIDEO_ENCODER_VALIDATION_FLAG_NONE = 0x0,
...
D3D12_VIDEO_ENCODER_VALIDATION_FLAG_SUBREGION_LAYOUT_DATA_NOT_SUPPORTED = 0x1000,
} D3D12_VIDEO_ENCODER_VALIDATION_FLAGS;New Constants
D3D12_VIDEO_ENCODER_VALIDATION_FLAG_SUBREGION_LAYOUT_DATA_NOT_SUPPORTED
When using D3D12_FEATURE_DATA_VIDEO_ENCODER_SUPPORT1, indicates the subregions configuration passed was not valid. Please use D3D12_FEATURE_DATA_VIDEO_ENCODER_FRAME_SUBREGION_LAYOUT_CONFIG for more information. D3D12_FEATURE_DATA_VIDEO_ENCODER_RESOLUTION_SUPPORT_LIMITS.MaxSubregionsNumber and D3D12_FEATURE_DATA_VIDEO_ENCODER_RESOLUTION_SUPPORT_LIMITS.SubregionBlockPixelsSize still must be reported when driver issues no support result in D3D12_FEATURE_DATA_VIDEO_ENCODER_SUPPORT1.
Extends on the previous D3D12_FEATURE_VIDEO_ENCODER_SUPPORT query, adding new parameters at the bottom of the associated data structure. This new query can be used with all H264, HEVC and AV1 codecs and must behave exactly as D3D12_FEATURE_VIDEO_ENCODER_SUPPORT semantics.
Please note the previous D3D12_FEATURE_VIDEO_ENCODER_SUPPORT will not work for AV1 codec as Codec input, and attempting to call it for AV1 will still return S_OK with ValidationFlags=D3D12_VIDEO_ENCODER_VALIDATION_FLAG_CODEC_NOT_SUPPORTED and print a debug layer message indicating to instead use the new D3D12_FEATURE_DATA_VIDEO_ENCODER_SUPPORT1 instead.
typedef struct D3D12_FEATURE_DATA_VIDEO_ENCODER_SUPPORT1
{
/* Below match existing D3D12_FEATURE_DATA_VIDEO_ENCODER_SUPPORT */
UINT NodeIndex;
D3D12_VIDEO_ENCODER_CODEC Codec;
DXGI_FORMAT InputFormat;
D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION CodecConfiguration;
D3D12_VIDEO_ENCODER_SEQUENCE_GOP_STRUCTURE CodecGopSequence;
D3D12_VIDEO_ENCODER_RATE_CONTROL RateControl;
D3D12_VIDEO_ENCODER_INTRA_REFRESH_MODE IntraRefresh;
D3D12_VIDEO_ENCODER_FRAME_SUBREGION_LAYOUT_MODE SubregionFrameEncoding;
UINT ResolutionsListCount;
const D3D12_VIDEO_ENCODER_PICTURE_RESOLUTION_DESC* pResolutionList;
UINT MaxReferenceFramesInDPB;
D3D12_VIDEO_ENCODER_VALIDATION_FLAGS ValidationFlags;
D3D12_VIDEO_ENCODER_SUPPORT_FLAGS SupportFlags;
D3D12_VIDEO_ENCODER_PROFILE_DESC SuggestedProfile;
D3D12_VIDEO_ENCODER_LEVEL_SETTING SuggestedLevel;
[annotation("_Field_size_full_(ResolutionsListCount)")] D3D12_FEATURE_DATA_VIDEO_ENCODER_RESOLUTION_SUPPORT_LIMITS* pResolutionDependentSupport;
/* Below are new arguments for D3D12_FEATURE_DATA_VIDEO_ENCODER_SUPPORT1 */
D3D12_VIDEO_ENCODER_PICTURE_CONTROL_SUBREGIONS_LAYOUT_DATA SubregionFrameEncodingData; // input
UINT MaxQualityVsSpeed;// output
} D3D12_FEATURE_DATA_VIDEO_ENCODER_SUPPORT1;New members
SubregionFrameEncodingData
Used to calculate D3D12_FEATURE_DATA_VIDEO_ENCODER_SUPPORT1.SuggestedLevel in codecs that have subregions constraints per level and used for related validation with the new flag D3D12_VIDEO_ENCODER_VALIDATION_FLAG_SUBREGION_LAYOUT_DATA_NOT_SUPPORTED.
MaxQualityVsSpeed
Reported by driver. Used as the maximum value allowed for the QualityVsSpeed parameter in the rate control structures. When D3D12_VIDEO_ENCODER_SUPPORT_FLAG_RATE_CONTROL_QUALITY_VS_SPEED_AVAILABLE is not reported, this value must be reported as zero.
typedef struct D3D12_FEATURE_DATA_VIDEO_ENCODER_RESOLUTION_SUPPORT_LIMITS
{
UINT MaxSubregionsNumber;
UINT MaxIntraRefreshFrameDuration;
UINT SubregionBlockPixelsSize;
UINT QPMapRegionPixelsSize;
} D3D12_FEATURE_DATA_VIDEO_ENCODER_RESOLUTION_SUPPORT_LIMITS;Member already present in existing API, see semantics for AV1 below.
Members
MaxSubregionsNumber
Indicates the maximum number of tiles supported by the hardware for the associated resolution.
SubregionBlockPixelsSize
Indicates the tiles block sizes in pixels for the associated resolution. This value must be equal or a multiple of the superblock size, which is passed in the input D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION parameter.
QPMapRegionPixelsSize
Indicates for the associated resolution, the size in pixels of the squared regions that will be affected by each of the values in the QP map buffer in absolute or delta QP modes. The resolution of the frame will be rounded up to be aligned to this value when it's partitioned in blocks for QP maps and the number of QP values in those maps will be the number of blocks of these indicated pixel size that comprise a full frame. This value must be equal or a multiple of the superblock size, which is passed in the input D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION parameter.
typedef struct D3D12_VIDEO_ENCODER_AV1_SEQUENCE_STRUCTURE
{
UINT IntraDistance;
UINT InterFramePeriod;
} D3D12_VIDEO_ENCODER_AV1_SEQUENCE_STRUCTURE;This is a hint to the driver of the GOP being used for rate control purposes only.
Members
IntraDistance
Indicates the distance between intra-only frames (or key frames) in the video sequence, or the number of pictures on a sequence of inter-frame pictures. If set to 0, only the first frame will be an key-frame.
InterPeriod
Indicates the period for inter-frames to be inserted within the inter frame structure. Note that if IntraDistance is set to 0 for infinite inter frame structure, this value must be greater than zero.
typedef struct D3D12_VIDEO_ENCODER_SEQUENCE_GOP_STRUCTURE
{
UINT DataSize;
union
{
...
D3D12_VIDEO_ENCODER_AV1_SEQUENCE_STRUCTURE* pAV1SequenceStructure;
};
} D3D12_VIDEO_ENCODER_SEQUENCE_GOP_STRUCTURE;Given an encoded frame with K tiles, the driver will write the K decode_tile() AV1 syntax elements in the compressed bitstream, corresponding to the requested tiles in EncodeFrame arguments.
The API Client then builds the tile_group_obu() AV1 syntax elements with tile_start_and_end_present_flag/tg_start/tg_end elements to arrange the tiles into tile groups as desired with the condition that the tiles are placed sequentially. The tile_size_minus_1 element is coded from the related tile D3D12_VIDEO_ENCODER_FRAME_SUBREGION_METADATA information and decode_tile() elements are copied from the compressed bitstream buffer. Finally, each tile_group_obu() is wrapped around open_bitstream_unit() elements of type OBU_TILE_GROUP and prepended with an OBU_FRAME_HEADER, or in case of a single tile group an OBU_FRAME type can be used.
The API Client is responsible for inferring obu_extension_flag as !(TemporalLayerIndexPlus1 || SpatialLayerIndexPlus1) for the current frame and also code if necessary temporal_id and spatial_id in the open_bitstream_unit().
The EncodeFrame submissions are in encode order, like the other codecs implemented in D3D12 Encode API.
If D3D12_VIDEO_ENCODER_SUPPORT_FLAG_RESOLUTION_RECONFIGURATION_AVAILABLE is reported by the driver, this still only applies to resolution changes on a key frame.
The active sequence header must have the max_frame_*_minus_1 syntax set to the max resolution present in the associated ID3D12VideoEncoderHeap being used, and different frames using resolutions also present in the associated ID3D12VideoEncoderHeap can use th AV1 syntax frame_size_override_flag in frame_size() to convey change of resolution.
If D3D12_VIDEO_ENCODER_AV1_FRAME_TYPE_FLAG_SWITCH_FRAME is supported, the reference frames must point to higher or equal resolution than the current switch frame being encoded and the different resolutions must be all present in the associated ID3D12VideoEncoderHeap being used.
Similarly, if spatial scalability is supported, the different resolutions of the reference frames must be all present in the associated ID3D12VideoEncoderHeap being used.
The accepted range for D3D12_VIDEO_ENCODER_RATE_CONTROL_QVBR.ConstantQualityTarget is [0..63]. The lowest the value, the highest the quality.
In general, D3D12_VIDEO_ENCODER_SUPPORT_FLAG_RATE_CONTROL_RECONFIGURATION_AVAILABLE applies to the quality vs speed tweaking and the following RC parameters of the different RC modes: QP in constant QP, bitrates and quality levels in CBR, VBR and QVBR. Driver can return D3D12_VIDEO_ENCODER_ENCODE_ERROR_FLAG_RECONFIGURATION_REQUEST_NOT_SUPPORTED in D3D12_VIDEO_ENCODER_OUTPUT_METADATA.EncodeErrorFlags for other unsupported RC params reconfiguration.
typedef struct D3D12_VIDEO_ENCODER_AV1_REFERENCE_PICTURE_WARPED_MOTION_INFO
{
D3D12_VIDEO_ENCODER_AV1_REFERENCE_WARPED_MOTION_TRANSFORMATION TransformationType;
INT TransformationMatrix[8];
bool InvalidAffineSet;
} D3D12_VIDEO_ENCODER_AV1_REFERENCE_PICTURE_WARPED_MOTION_INFO;Remarks
Related to warped motion transformation/global motion type. Transform to be applied to motion vectors.
#define D3D12_VIDEO_ENCODER_AV1_INVALID_DPB_RESOURCE_INDEX 0xFF
typedef struct D3D12_VIDEO_ENCODER_AV1_REFERENCE_PICTURE_DESCRIPTOR
{
UINT ReconstructedPictureResourceIndex;
UINT TemporalLayerIndexPlus1;
UINT SpatialLayerIndexPlus1;
D3D12_VIDEO_ENCODER_AV1_FRAME_TYPE FrameType;
D3D12_VIDEO_ENCODER_AV1_REFERENCE_PICTURE_WARPED_MOTION_INFO WarpedMotionInfo;
UINT OrderHint;
UINT PictureIndex;
} D3D12_VIDEO_ENCODER_AV1_REFERENCE_PICTURE_DESCRIPTOR;Members
ReconstructedPictureResourceIndex
Maps the current reference picture described by this structure to a resource in the D3D12_VIDEO_ENCODER_PICTURE_CONTROL_DESC.ReferenceFrames array.
If the associated slot in ReferenceFramesReconPictureDescriptors containing this structure has a valid ReconstructedPictureResourceIndex reference to the D3D12_VIDEO_ENCODER_PICTURE_CONTROL_DESC.ReferenceFrames array then the allowed range of values is [0..254]. Otherwise, if it corresponds to an empty/unused slot in the DPB, then the value must be set to D3D12_VIDEO_ENCODER_AV1_INVALID_DPB_RESOURCE_INDEX.
TemporalLayerIndexPlus1
Picture temporal layer index plus one of the previously encoded frame now used as reference.
SpatialLayerIndexPlus1
Picture spatial layer index plus one of the previously encoded frame now used as reference.
FrameType
The type of frame used to encode the described reference frame associated to this entry.
WarpedMotionInfo
Global motion params. Only used if supported in D3D12_VIDEO_ENCODER_AV1_REFERENCE_WARPED_MOTION_TRANSFORMATION_FLAGS.
OrderHint
The reference ref_order_hint AV1 syntax. Even when reference order hints are not coded in the AV1 bitstream, OrderHint here must be set to the value used in D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_CODEC_DATA.OrderHint for the previously encoded frame that is now being held as reference by this descriptor. This information will hint the driver, in a frame type that allows frame references, which are from past frames and which are from future frames (in display order) when comparing against the current frame value of D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_CODEC_DATA.OrderHint.
PictureIndex
The unique picture index of the previously encoded frame that is stored in this entry as a reference. This parameter is not related in any way to the AV1 standard syntax, but merely used for API client implementation tracking instead.
typedef enum D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_FLAGS
{
D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_FLAG_NONE = 0x0,
D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_FLAG_ENABLE_ERROR_RESILIENT_MODE = 0x1,
D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_FLAG_DISABLE_CDF_UPDATE = 0x2,
D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_FLAG_ENABLE_PALETTE_ENCODING = 0x4,
D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_FLAG_ENABLE_SKIP_MODE = 0x8,
D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_FLAG_FRAME_REFERENCE_MOTION_VECTORS = 0x10,
D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_FLAG_FORCE_INTEGER_MOTION_VECTORS = 0x20,
D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_FLAG_ALLOW_INTRA_BLOCK_COPY = 0x40,
D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_FLAG_USE_SUPER_RESOLUTION = 0x80,
D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_FLAG_DISABLE_FRAME_END_UPDATE_CDF = 0x100,
D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_FLAG_ENABLE_FRAME_SEGMENTATION_AUTO = 0x200,
D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_FLAG_ENABLE_FRAME_SEGMENTATION_CUSTOM = 0x400,
D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_FLAG_ENABLE_WARPED_MOTION = 0x800,
D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_FLAG_REDUCED_TX_SET = 0x1000,
D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_FLAG_MOTION_MODE_SWITCHABLE = 0x2000,
D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_FLAG_ALLOW_HIGH_PRECISION_MV = 0x4000,
} D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_FLAGS;Defines the set of flags for the codec-specific picture control properties.
Constants
D3D12_VIDEO_ENCODER_PICTURE_CONTROL_FLAG_NONE
No flags
D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_FLAG_ENABLE_ERROR_RESILIENT_MODE
Related to error_resilient_mode AV1 syntax in frame header.
D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_FLAG_DISABLE_CDF_UPDATE
Related to AV1 syntax for disable_cdf_update.
D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_FLAG_ENABLE_PALETTE_ENCODING
Enables the usage of palette encoding for this frame.
D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_FLAG_ENABLE_SKIP_MODE
Related to AV1 syntax skip_mode_present. skip_mode element will be present for this frame if this flag is set. Please check support in AV1 query caps before enabling this feature.
D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_FLAG_FRAME_REFERENCE_MOTION_VECTORS
Related to AV1 syntax use_ref_frame_mvs. Equal to 1 specifies that motion vector information from a previous frame can be used when encoding the current frame. use_ref_frame_mvs equal to 0 specifies that this information will not be used.
D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_FLAG_FORCE_INTEGER_MOTION_VECTORS
Equal to 1 specifies that force_integer_mv may be enabled on a per frame basis. Equal to 0 specifies that force_integer_mv syntax element will not be used. Please check support in AV1 query caps before enabling this feature.
D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_FLAG_ALLOW_INTRA_BLOCK_COPY
Indicates if intra block copy is supported or not at per frame basis. Related to allow_intrabc syntax in AV1 spec.
D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_FLAG_USE_SUPER_RESOLUTION
Related to AV1 syntax use_superres.
D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_FLAG_DISABLE_FRAME_END_UPDATE_CDF
Related to AV1 syntax disable_frame_end_update_cdf.
D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_FLAG_ENABLE_FRAME_SEGMENTATION_AUTO
Enables automatic (performed by driver without API Client input) segmentation for the current frame. Requires D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_AUTO_SEGMENTATION. This flag must not be combined with D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_FLAG_ENABLE_FRAME_SEGMENTATION_CUSTOM.
D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_FLAG_ENABLE_FRAME_SEGMENTATION_CUSTOM
Enables customized segmentation with the API Client sending the driver segmentation config and segment map. Requires D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_CUSTOM_SEGMENTATION. This flag must not be combined with D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_FLAG_ENABLE_FRAME_SEGMENTATION_AUTO.
D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_FLAG_ENABLE_WARPED_MOTION
Related to AV1 syntax allow_warped_motion to be coded in the frame header. Requires D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_WARPED_MOTION.
D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_FLAG_REDUCED_TX_SET
Related to AV1 syntax reduced_tx_set. Requires D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_REDUCED_TX_SET.
D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_FLAG_MOTION_MODE_SWITCHABLE
Related to AV1 syntax is_motion_mode_switchable. Requires D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_MOTION_MODE_SWITCHABLE.
D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_FLAG_ALLOW_HIGH_PRECISION_MV
Related to AV1 syntax allow_high_precision_mv. Requires D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_ALLOW_HIGH_PRECISION_MV.
typedef struct D3D12_VIDEO_ENCODER_AV1_RESTORATION_CONFIG {
D3D12_VIDEO_ENCODER_AV1_RESTORATION_TYPE FrameRestorationType[3];
D3D12_VIDEO_ENCODER_AV1_RESTORATION_TILESIZE LoopRestorationPixelSize[3];
} D3D12_VIDEO_ENCODER_AV1_RESTORATION_CONFIG;Remarks
Related to AV1 syntax lr_params(). The array entries correspond to Y, U, V planes.
// 64bit aligned access for SetPredication as part of resolved metadata memory layout
typedef struct D3D12_VIDEO_ENCODER_AV1_SEGMENT_DATA {
UINT64 EnabledFeatures;
INT64 FeatureValue[8];
} D3D12_VIDEO_ENCODER_AV1_SEGMENT_DATA;Members
EnabledFeatures
Accepts a bit mask combination of values from D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_MODE_FLAGS.
FeatureValue
For the enabled bit flags features in EnabledFeatures, the array FeatureValue is indexed by D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_MODE - 1 for it's associated feature value.
// 64bit aligned access for SetPredication as part of resolved metadata memory layout
typedef struct D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_CONFIG {
UINT64 UpdateMap;
UINT64 TemporalUpdate;
UINT64 UpdateData;
UINT64 NumSegments;
D3D12_VIDEO_ENCODER_AV1_SEGMENT_DATA SegmentsData[8];
} D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_CONFIG;Related to AV1 syntax segmentation_params()
NumSegments
When using D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_FLAG_ENABLE_FRAME_SEGMENTATION_AUTO and the driver writes it back on post encode values, NumSegments = 0 indicated that segmentation_enabled must be 0 in the frame header. Otherwise, the API client codes segmentation_params() in the frame header accordingly with the other parameters in this structure.
When using D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_FLAG_ENABLE_FRAME_SEGMENTATION_CUSTOM, indicates the input number of segments.
typedef struct D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_MAP {
UINT SegmentsMapByteSize;
UINT8* pSegmentsMap;
} D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_MAP;Members
SegmentsMapByteSize
Byte size of pSegmentsMap buffer.
pSegmentsMap
In raster order, contains the AV1 syntax segment_id between [0..7] for each block in the frame. The block size is SegmentationBlockSize as reported by driver in D3D12_VIDEO_ENCODER_AV1_CODEC_CONFIGURATION_SUPPORT.
// 64bit aligned access for SetPredication as part of resolved metadata memory layout
typedef struct D3D12_VIDEO_ENCODER_CODEC_AV1_LOOP_FILTER_CONFIG {
UINT64 LoopFilterLevel[2];
UINT64 LoopFilterLevelU;
UINT64 LoopFilterLevelV;
UINT64 LoopFilterSharpnessLevel;
UINT64 LoopFilterDeltaEnabled;
UINT64 UpdateRefDelta;
INT64 RefDeltas[8];
UINT64 UpdateModeDelta;
INT64 ModeDeltas[2];
} D3D12_VIDEO_ENCODER_CODEC_AV1_LOOP_FILTER_CONFIG;Related to AV1 syntax loop_filter_params().
Remarks
AV1 syntax loop_filter_delta_update is derived from members as (UpdateRefDelta || UpdateModeDelta).
Members
LoopFilterLevel
Related to AV1 syntax loop_filter_level[0], loop_filter_level[1]
LoopFilterLevelU
Related to AV1 syntax loop_filter_level[2]
LoopFilterLevelV
Related to AV1 syntax loop_filter_level[3]
LoopFilterSharpnessLevel
Related to AV1 syntax loop_filter_sharpness
LoopFilterDeltaEnabled
Related to AV1 syntax loop_filter_delta_enabled. Requires D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_LOOP_FILTER_DELTAS supported/enabled.
UpdateRefDelta
Related to AV1 syntax update_ref_delta
RefDeltas
Related to AV1 syntax loop_filter_ref_deltas
UpdateModeDelta
Related to AV1 syntax update_mode_delta
ModeDeltas
Related to AV1 syntax loop_filter_mode_deltas
// 64bit aligned access for SetPredication as part of resolved metadata memory layout
typedef struct D3D12_VIDEO_ENCODER_CODEC_AV1_LOOP_FILTER_DELTA_CONFIG {
UINT64 DeltaLFPresent;
UINT64 DeltaLFMulti;
UINT64 DeltaLFRes;
} D3D12_VIDEO_ENCODER_CODEC_AV1_LOOP_FILTER_DELTA_CONFIG;Related to AV1 syntax delta_lf_params().
Members
DeltaLFPresent
Related to AV1 syntax delta_lf_params(). Requires D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_DELTA_LF_PARAMS supported/enabled.
DeltaLFMulti
Related to AV1 syntax delta_lf_params()
DeltaLFRes
Related to AV1 syntax delta_lf_params()
// 64bit aligned access for SetPredication as part of resolved metadata memory layout
typedef struct D3D12_VIDEO_ENCODER_CODEC_AV1_QUANTIZATION_CONFIG {
UINT64 BaseQIndex;
INT64 YDCDeltaQ;
INT64 UDCDeltaQ;
INT64 UACDeltaQ;
INT64 VDCDeltaQ;
INT64 VACDeltaQ;
UINT64 UsingQMatrix;
UINT64 QMY;
UINT64 QMU;
UINT64 QMV;
} D3D12_VIDEO_ENCODER_CODEC_AV1_QUANTIZATION_CONFIG;Remarks
- AV1 syntax separate_uv_delta_q will be coded as 1 always
- AV1 syntax diff_uv_delta can be inferred from below if U and V AC/DC components are the same.
Members
BaseQIndex
Related to AV1 syntax in quantization_params().
YDCDeltaQ
Related to AV1 syntax in quantization_params().
UDCDeltaQ
Related to AV1 syntax in quantization_params().
UACDeltaQ
Related to AV1 syntax in quantization_params().
VDCDeltaQ
Related to AV1 syntax in quantization_params().
VACDeltaQ
Related to AV1 syntax in quantization_params().
UsingQMatrix
Related to AV1 syntax in quantization_params().
QMY
Related to AV1 syntax in quantization_params().
QMU
Related to AV1 syntax in quantization_params().
QMV
Related to AV1 syntax in quantization_params().
// 64bit aligned access for SetPredication as part of resolved metadata memory layout
typedef struct D3D12_VIDEO_ENCODER_CODEC_AV1_QUANTIZATION_DELTA_CONFIG {
UINT64 DeltaQPresent;
UINT64 DeltaQRes;
} D3D12_VIDEO_ENCODER_CODEC_AV1_QUANTIZATION_DELTA_CONFIG;Members
DeltaQPresent
Related to AV1 syntax in delta_q_params().
DeltaQRes
Related to AV1 syntax in delta_q_params().
// 64bit aligned access for SetPredication as part of resolved metadata memory layout
typedef struct D3D12_VIDEO_ENCODER_AV1_CDEF_CONFIG {
UINT64 CdefBits;
UINT64 CdefDampingMinus3;
UINT64 CdefYPriStrength[8];
UINT64 CdefUVPriStrength[8];
UINT64 CdefYSecStrength[8];
UINT64 CdefUVSecStrength[8];
} D3D12_VIDEO_ENCODER_AV1_CDEF_CONFIG;Members
CdefBits
Related to AV1 syntax in cdef_params().
CdefDampingMinus3
Related to AV1 syntax in cdef_params().
CdefYPriStrength
Related to AV1 syntax in cdef_params().
CdefUVPriStrength
Related to AV1 syntax in cdef_params().
CdefYSecStrength
Related to AV1 syntax in cdef_params().
CdefUVSecStrength
Related to AV1 syntax in cdef_params().
#define D3D12_VIDEO_ENCODER_AV1_SUPERRES_DENOM_MIN 9
typedef struct D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_CODEC_DATA
{
D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_FLAGS Flags;
D3D12_VIDEO_ENCODER_AV1_FRAME_TYPE FrameType;
D3D12_VIDEO_ENCODER_AV1_COMP_PREDICTION_TYPE CompoundPredictionType;
D3D12_VIDEO_ENCODER_AV1_INTERPOLATION_FILTERS InterpolationFilter;
D3D12_VIDEO_ENCODER_AV1_RESTORATION_CONFIG FrameRestorationConfig;
D3D12_VIDEO_ENCODER_AV1_TX_MODE TxMode;
UINT SuperResDenominator;
UINT OrderHint;
UINT PictureIndex;
UINT TemporalLayerIndexPlus1;
UINT SpatialLayerIndexPlus1;
D3D12_VIDEO_ENCODER_AV1_REFERENCE_PICTURE_DESCRIPTOR ReferenceFramesReconPictureDescriptors[8];
UINT ReferenceIndices[7];
UINT PrimaryRefFrame;
UINT RefreshFrameFlags;
D3D12_VIDEO_ENCODER_CODEC_AV1_LOOP_FILTER_CONFIG LoopFilter;
D3D12_VIDEO_ENCODER_CODEC_AV1_LOOP_FILTER_DELTA_CONFIG LoopFilterDelta;
D3D12_VIDEO_ENCODER_CODEC_AV1_QUANTIZATION_CONFIG Quantization;
D3D12_VIDEO_ENCODER_CODEC_AV1_QUANTIZATION_DELTA_CONFIG QuantizationDelta;
D3D12_VIDEO_ENCODER_AV1_CDEF_CONFIG CDEF;
UINT QPMapValuesCount;
[annotation("_Field_size_full_(QPMapValuesCount)")] INT16 *pRateControlQPMap;
D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_CONFIG CustomSegmentation;
D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_MAP SegmentsMap;
} D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_CODEC_DATA;Constants
D3D12_VIDEO_ENCODER_AV1_SUPERRES_DENOM_MIN
As defined by the AV1 codec standard. Associated to SuperResDenominator param below.
Members
Flags
Configuration flags for this frame to be encoded.
FrameType
Sets the picture type. See D3D12_VIDEO_ENCODER_AV1_FRAME_TYPE values.
CompoundPredictionType
Specifies whether single or compound prediction is used for the given frame. Related to AV1 syntax reference_select.
InterpolationFilter
InterpolationFilter to be used for inter prediction on the current frame, related to syntax interpolation_filter.
FrameRestorationConfig
Indicates the restoration config to be used.
InterIntraConfig
Indicates the configuration for inter intra.
FilterIntraMode
Indicates the configuration for filter intra.
MotionMode
Indicates the configuration for motion mode.
CompoundTypeConfig
Indicates the configuration for compound config.
SuperResDenominator
Indicates the configuration for super resolution. Has to be greater or equal than D3D12_VIDEO_ENCODER_AV1_SUPERRES_DENOM_MIN when super resolution is enabled.
OrderHint
Current frame order_hint AV1 syntax. For this API purposes, OrderHint must be always passed even when not coding the order hint in the AV1 bitstream, and it must reflect the display order of the frame.
PictureIndex
The unique picture index for this frame that will be used to uniquely identify it as a reference for future frames. This parameter is not related in any way to the AV1 standard syntax, but merely used for API client implementation tracking instead.
API Client should initialize this value at 0 for the first D3D12_VIDEO_ENCODER_AV1_FRAME_TYPE_KEY_FRAME and increment it by one on each subsequent frame until the next D3D12_VIDEO_ENCODER_AV1_FRAME_TYPE_KEY_FRAME, when it should be reset to zero and follow the same process.
Note: OrderHint cannot be used for this purpose as it has a max range of [0..2^(OrderHintBitsMinus1+1)], which can wrap around and not work as unique identifier of the frames and their references.
TemporalLayerIndexPlus1
Picture temporal layer index plus one. A value of zero indicates temporal scalability not used. This value must be within the range [0..D3D12_VIDEO_ENCODER_AV1_CODEC_CONFIGURATION_SUPPORT.MaxTemporalLayers].
SpatialLayerIndexPlus1
Picture spatial layer index plus one. A value of zero indicates spatial scalability not used. This value must be within the range [0..D3D12_VIDEO_ENCODER_AV1_CODEC_CONFIGURATION_SUPPORT.MaxSpatialLayers].
ReferenceFramesReconPictureDescriptors
Describes the current state snapshot of the complete (ie. including frames that are not used by current frame but used by future frames, etc) DPB buffer kept in D3D12_VIDEO_ENCODER_PICTURE_CONTROL_DESC.ReferenceFrames. The reference indices (ie. last, altref, etc) map from past/future references into this descriptors array. The AV1 codec allows up to 8 references in the DPB.
This array of descriptors, in turn, maps a reference picture for this frame into a resource index in the reconstructed pictures array D3D12_VIDEO_ENCODER_PICTURE_CONTROL_DESC.ReferenceFrames.
The size of this array always matches D3D12_VIDEO_ENCODER_PICTURE_CONTROL_DESC.ReferenceFrames.NumTextures for the associated EncodeFrame command.
ReferenceIndices
Corresponds to the ref_frame_idx[i] AV1 syntax. For a reference type i, ReferenceIndices[i] indicates an index between [0..7] into ReferenceFramesReconPictureDescriptors where the current frame i-th reference type is stored in the DPB. In other words ReferenceFramesReconPictureDescriptors[ReferenceIndices[i]] contains the DPB descriptor for the i-th reference type.
The i-th entry of ReferenceIndices[] corresponds to each reference type as follows:
| Index i | Reference type | ReferenceFramesReconPictureDescriptors[ReferenceIndices[i]] |
|---|---|---|
| 0 | Last | DPB Descriptor for Last |
| 1 | Last2 | DPB Descriptor for Last2 |
| 2 | Last3 | DPB Descriptor for Last3 |
| 3 | Golden | DPB Descriptor for Golden |
| 4 | Bwdref | DPB Descriptor for Bwdref |
| 5 | Altref | DPB Descriptor for Altref |
| 6 | Altref2 | DPB Descriptor for Altref2 |
PrimaryRefFrame
Corresponds to the AV1 element syntax primary_ref_frame in uncompressed_header(). Specifies which reference frame contains the CDF values and other state that must be loaded at the start of the frame. The allowed range is [0..7] and the values corresponds as follows:
| PrimaryRefFrame value | AV1 syntax value (primary_ref_frame) | Reference frame selected |
|---|---|---|
| 0 | 0 | Last |
| 1 | 1 | Last2 |
| 2 | 2 | Last3 |
| 3 | 3 | Golden |
| 4 | 4 | Bwdref |
| 5 | 5 | Altref |
| 6 | 6 | Altref2 |
| 7 | 7 (PRIMARY_REF_NONE) | None |
RefreshFrameFlags
Corresponds to the refresh_frame_flags AV1 syntax element.
QPMapValuesCount
Contains the number of elements present in QPMapValuesCount. This must match the number of coding blocks in the frame, rounding up the frame resolution to the closest aligned values.
LoopFilter
Specifies the loop filter parameters.
LoopFilterDelta
Specifies the loop filter delta parameters. Related to delta_lf_params AV1 syntax.
Quantization
Specifies the quantization parameters.
QuantizationDelta
Specifies the quantization delta parameters.
CDEF
Specifies the constrained directional enhancement filtering parameters.
pRateControlQPMap
Array containing in row/column scan order, the QP map values to use on each squared region for this frame. The QP map dimensions can be calculated using the current resolution and D3D12_FEATURE_DATA_VIDEO_ENCODER_RESOLUTION_SUPPORT_LIMITS.QPMapRegionPixelsSize conveying the squared region sizes. The range for Delta QP values is [-255;255].
CustomSegmentation
Only used when D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_FLAG_ENABLE_FRAME_SEGMENTATION_CUSTOM is set for current frame.
SegmentsMap
Only used when D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_FLAG_ENABLE_FRAME_SEGMENTATION_CUSTOM is set for current frame. Segment map to be used if CustomSegmentation.UpdateMap is set. Otherwise, the segment map is inherited from ref frame.
Remarks
DPB Management expectations
The following is the contract the API client and the driver must conform to when using this API:
-
When encoding D3D12_VIDEO_ENCODER_AV1_FRAME_TYPE_KEY_FRAME:
-
PrimaryRefFrame must be 7 (PRIMARY_REF_NONE)
-
RefreshFrameFlags must be 0xFF (conforming to the AV1 codec standard bitstream syntax)
-
For all array entries in ReferenceFramesReconPictureDescriptors, ReconstructedPictureResourceIndex must be set to D3D12_VIDEO_ENCODER_AV1_INVALID_DPB_RESOURCE_INDEX
-
The values in ReferenceIndices[7] are indistinct as any value [0..7] will point to an unused DPB slot in ReferenceFramesReconPictureDescriptors as per the point above.
-
D3D12_VIDEO_ENCODE_REFERENCE_FRAMES will be filled as:
-
NumTexture2Ds = 0
-
ppTexture2Ds = NULL
-
pSubresources = NULL
-
-
-
When encoding a frame type with references (ie. SWITCH/INTER):
-
PrimaryRefFrame specifies which reference frame contains the CDF values and other state that must be loaded at the start of the frame.
-
RefreshFrameFlags must be set accordingly to indicate on which ReferenceFramesReconPictureDescriptors DPB slots the current encoded frame reconstructed picture will be placed after encoding execution. This value must match exactly what the API client will code for refresh_frame_flags AV1 syntax element in the associated picture header for the current frame.
-
The array entries in ReferenceFramesReconPictureDescriptors for the current frame will follow exactly what RefreshFrameFlags indicated for the previous frame.
- For example: if RefreshFrameFlags indicates the current frame N will be placed in slots 2, 3 and 6, then in the next EncodeFrame call for frame N+1, the ReconstructedPictureResourceIndex parameter in the entries ReferenceFramesReconPictureDescriptors[1], ReferenceFramesReconPictureDescriptors[2], ReferenceFramesReconPictureDescriptors[5] param must point to frame N reconstructed picture in D3D12_VIDEO_ENCODE_REFERENCE_FRAMES.
- Note this includes considering RefreshFrameFlags=0xFF for D3D12_VIDEO_ENCODER_AV1_FRAME_TYPE_KEY_FRAME as well, by marking all ReferenceFramesReconPictureDescriptors entries pointing to the KEY frame reconstructed picture.
- If a (non key frame) picture won't be marked with D3D12_VIDEO_ENCODER_PICTURE_CONTROL_FLAG_USED_AS_REFERENCE_PICTURE, their RefreshFrameFlags must be zero to indicate this.
-
The values in ReferenceIndices[7] are within [0..7] and point to DPB slots in ReferenceFramesReconPictureDescriptors.
- The values of ReferenceIndices[7] must match exactly what the API client codes in the picture header for ref_frame_idx.
- As per AV1 syntax definition of ref_frame_idx there is no requirement that the array entry values must be unique.
- If ReferenceFramesReconPictureDescriptors[ReferenceIndices[i]].
ReconstructedPictureResourceIndex == D3D12_VIDEO_ENCODER_AV1_INVALID_DPB_RESOURCE_INDEX
- This indicates the reference picture i-th won't be used for the current frame and must be ignored.
- Otherwise (Valid ReferenceFramesReconPictureDescriptors[ReferenceIndices[i]].
ReconstructedPictureResourceIndex)
- This indicates the reference i-th will point to the DPB slot ReferenceIndices[i] and the reconstructed picture to be used is D3D12_VIDEO_ENCODE_REFERENCE_FRAMES.ppTexture2Ds[ReferenceFramesReconPictureDescriptors[ReferenceIndices[i]].ReconstructedPictureResourceIndex]
-
D3D12_VIDEO_ENCODE_REFERENCE_FRAMES will be filled as:
-
NumTexture2Ds = {number of unique values of ReferenceFramesReconPictureDescriptors.ReconstructedPictureResourceIndex[j] != D3D12_VIDEO_ENCODER_AV1_INVALID_DPB_RESOURCE_INDEX for j in {0..7} }
-
ppTexture2Ds = { compact array (no null entries) containing the reconstructed picture from the previously encoded frames that will be used as references }
-
[Texture Array mode only] pSubresources = { compact array (no null entries) containing the subresource index of the texture array from the previously encoded frames that will be used as references }
-
-
-
When encoding a frame type without references but without clearing the DPB (ie. INTRA_ONLY)
-
PrimaryRefFrame must be 7 (PRIMARY_REF_NONE)
-
RefreshFrameFlags must be set accordingly to indicate on which ReferenceFramesReconPictureDescriptors DPB slots the current encoded frame reconstructed picture will be placed after encoding execution. This value must match exactly what the API client will code for refresh_frame_flags AV1 syntax element in the associated picture header for the current frame.
-
The array entries in ReferenceFramesReconPictureDescriptors for the current frame will follow exactly what RefreshFrameFlags indicated for the previous frame.
-
The values in ReferenceIndices[7] are ignored as the intra-only frame doesn't use any references.
-
D3D12_VIDEO_ENCODE_REFERENCE_FRAMES contains the DPB snapshot and will be filled as:
-
NumTexture2Ds = {number of unique values of ReferenceFramesReconPictureDescriptors.ReconstructedPictureResourceIndex[j] != D3D12_VIDEO_ENCODER_AV1_INVALID_DPB_RESOURCE_INDEX for j in {0..7} }
-
ppTexture2Ds = { compact array (no null entries) containing the reconstructed picture from the previously encoded frames referred to by ReferenceFramesReconPictureDescriptors.ReconstructedPictureResourceIndex[j] for j in {0..7} }
-
[Texture Array mode only] pSubresources = { compact array (no null entries) containing the subresource index of the texture array from the previously encoded frames referred to by ReferenceFramesReconPictureDescriptors.ReconstructedPictureResourceIndex[j] for j in {0..7} }
-
-
typedef struct D3D12_VIDEO_ENCODER_PICTURE_CONTROL_CODEC_DATA
{
UINT DataSize;
union
{
...
D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_CODEC_DATA* pAV1PicData;
};
} D3D12_VIDEO_ENCODER_PICTURE_CONTROL_CODEC_DATA;constexpr UINT MAX_TILE_ROWS = 64u;
constexpr UINT MAX_TILE_COLS = 64u;
// 64bit aligned access for SetPredication as part of resolved metadata memory layout
typedef struct D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_SUBREGIONS_LAYOUT_DATA_TILES
{
UINT64 RowCount;
UINT64 ColCount;
UINT64 RowHeights[MAX_TILE_ROWS];
UINT64 ColWidths[MAX_TILE_COLS];
UINT64 ContextUpdateTileId;
} D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_SUBREGIONS_LAYOUT_DATA_TILES;Remarks
This operates in different ways using different D3D12_VIDEO_ENCODER_FRAME_SUBREGION_LAYOUT_MODE configurations.
For D3D12_VIDEO_ENCODER_FRAME_SUBREGION_LAYOUT_MODE_CONFIGURABLE_GRID_PARTITION,
- Input parameters: RowCount, ColCount, RowHeights, ColWidths within the reported tile caps. The integer values must match the AV1 codec standard expectations (ie. power of two, etc).
- Driver honors exactly and copies the exact same structure after EncodeFrame execution.
For D3D12_VIDEO_ENCODER_FRAME_SUBREGION_LAYOUT_MODE_UNIFORM_GRID_PARTITION
- Input parameters: RowCount, ColCount. The integer values must match the AV1 codec standard expectations (ie. power of two, etc).
- Driver copies RowCount/ColCount as passed by API Client, and returns also RowHeights, ColWidths after EncodeFrame execution.
For ContextUpdateTileId, is an input parameter from the API Client corresponding to the frame header context_update_tile_id AV1 syntax and if D3D12_VIDEO_ENCODER_AV1_POST_ENCODE_VALUES_FLAG_CONTEXT_UPDATE_TILE_ID was reported then the driver is able to overwrite the API client input after EncodeFrame execution, otherwise must be copied by the driver from the input to the post encode values.
Constants
MAX_TILE_ROWS
As defined in the AV1 codec standard.
MAX_TILE_COLS
As defined in the AV1 codec standard.
Members
RowCount
Number of tile rows.
ColCount
Number of tile cols.
RowHeights
Heights of tile rows.
ColWidths
Widths of tile cols.
ContextUpdateTileId
Related to AV1 syntax context_update_tile_id.
Remarks
Please check the limitations for tile configuration returned in D3D12_FEATURE_DATA_VIDEO_ENCODER_FRAME_SUBREGION_LAYOUT_CONFIG for AV1 codec.
For the width and height parameter lists, please make sure they fall under the codec standard defined dimension limitations specified above. The units of these arrays are measured in block of size D3D12_FEATURE_DATA_VIDEO_ENCODER_RESOLUTION_SUPPORT_LIMITS.SubregionBlockPixelsSize. For example if ColHeights[0] = 2 and SubregionBlockPixelsSize = 64,
then the ColHeightInPixels[0] = ColHeights[0] * SubregionBlockPixelsSize = 2 * 64 = 128.
typedef struct D3D12_VIDEO_ENCODER_PICTURE_CONTROL_SUBREGIONS_LAYOUT_DATA
{
UINT DataSize;
union
{
...
const D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_SUBREGIONS_LAYOUT_DATA_TILES* pTilesPartition_AV1;
};
} D3D12_VIDEO_ENCODER_PICTURE_CONTROL_SUBREGIONS_LAYOUT_DATA;typedef struct D3D12_VIDEO_ENCODER_AV1_POST_ENCODE_VALUES
{
D3D12_VIDEO_ENCODER_AV1_COMP_PREDICTION_TYPE CompoundPredictionType;
D3D12_VIDEO_ENCODER_CODEC_AV1_LOOP_FILTER_CONFIG LoopFilter;
D3D12_VIDEO_ENCODER_CODEC_AV1_LOOP_FILTER_DELTA_CONFIG LoopFilterDelta;
D3D12_VIDEO_ENCODER_CODEC_AV1_QUANTIZATION_CONFIG Quantization;
D3D12_VIDEO_ENCODER_CODEC_AV1_QUANTIZATION_DELTA_CONFIG QuantizationDelta;
D3D12_VIDEO_ENCODER_AV1_CDEF_CONFIG CDEF;
D3D12_VIDEO_ENCODER_AV1_SEGMENTATION_CONFIG SegmentationConfig;
UINT64 PrimaryRefFrame; // Aligned to 64 for use of post encode metadata with predication
UINT64 ReferenceIndices[7]; // Aligned to 64 for use of post encode metadata with predication
} D3D12_VIDEO_ENCODER_AV1_POST_ENCODE_VALUES;Remarks
If D3D12_VIDEO_ENCODER_AV1_POST_ENCODE_VALUES_FLAGS was reported for the respective structures, driver must write the values after EncodeFrame execution with or without modifications (copy API client input) done by the driver. Otherwise, driver must copy the values from the associated API Client input when calling EncodeFrame.
Members
CompoundPredictionType
Associated flag is D3D12_VIDEO_ENCODER_AV1_POST_ENCODE_VALUES_FLAG_COMPOUND_PREDICTION_MODE.
LoopFilter
Associated flag is D3D12_VIDEO_ENCODER_AV1_POST_ENCODE_VALUES_FLAG_LOOP_FILTER.
LoopFilterDelta
Associated flag is D3D12_VIDEO_ENCODER_AV1_POST_ENCODE_VALUES_FLAG_LOOP_FILTER_DELTA.
Quantization
Associated flag is D3D12_VIDEO_ENCODER_AV1_POST_ENCODE_VALUES_FLAG_QUANTIZATION.
QuantizationDelta
Associated flag is D3D12_VIDEO_ENCODER_AV1_POST_ENCODE_VALUES_FLAG_QUANTIZATION_DELTA.
CDEF
Associated flag is D3D12_VIDEO_ENCODER_AV1_POST_ENCODE_VALUES_FLAG_CDEF_DATA.
SegmentationConfig
This operates in different modes depending the segmentation mode selected.
- D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_FLAG_ENABLE_FRAME_SEGMENTATION_AUTO is on for current frame
- Driver will calculate and write SegmentationConfig which will be used by the API Client to code the segmentation_params() syntax in the frame header.
- Driver will calculate and write the read_segment_id() map information directly in the compressed bitstream
- D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_FLAG_ENABLE_FRAME_SEGMENTATION_CUSTOM is on for current frame
- Driver will copy here the SegmentationConfig sent by the API Client in the encode frame parameters for API Client to code the segmentation_params() syntax in the frame header.
- Driver will copy the read_segment_id() map sent by the API Client in the encode frame parameters directly in the compressed bitstream
- Otherwise (both segmentation modes are off)
- Driver should write all zeroes
- API Client will write segmentation_enabled = 0 accordingly in the segmentation_params() section of the frame header
PrimaryRefFrame
Associated flag is D3D12_VIDEO_ENCODER_AV1_POST_ENCODE_VALUES_FLAG_PRIMARY_REF_FRAME. If the flag was set, the driver controls the selection of the primary reference frame from which the segment id map, CDF, etc are inherited from.
When used together with D3D12_VIDEO_ENCODER_AV1_FEATURE_FLAG_AUTO_SEGMENTATION, allows the driver to force a value other than PRIMARY_REF_NONE when applicable, giving the driver full control of the AV1 syntax: segmentation_update_map, segmentation_temporal_update and segmentation_update_data in segmentation_params().
ReferenceIndices
When the flag is reported, the driver may reorder/remap (but not change the number of references) the ReferenceIndices array, based on the user input D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_CODEC_DATA.ReferenceIndices. Otherwise, driver must copy each array entry of this parameter as-is from D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_CODEC_DATA.ReferenceIndices.
API Client will write the picture header ref_frame_idx AV1 syntax from this output parameter.
void ResolveEncoderOutputMetadata(
[annotation("_In_")] const D3D12_VIDEO_ENCODER_RESOLVE_METADATA_INPUT_ARGUMENTS* pInputArguments,
[annotation("_In_")] const D3D12_VIDEO_ENCODER_RESOLVE_METADATA_OUTPUT_ARGUMENTS* pOutputArguments);Resolves the output metadata to a readable format. Already present in existing Video Encode API.
Remarks
- The API Client must interpret the contents of pOutputArguments as a memory blob that contains the data as specified in the buffer layout defined for ResolveEncoderOutputMetadata below. All the information in the layout is positioned in memory contiguously. API Client will parse accordingly depending on the value of D3D12_VIDEO_ENCODER_RESOLVE_METADATA_INPUT_ARGUMENTS.EncodeCodec.
- For D3D12_VIDEO_ENCODER_RESOLVE_METADATA_INPUT_ARGUMENTS.EncodeCodec equal to values supported before this extension to the API (ie. H264/HEVC) the resolved buffer memory layout will **not** be changed.
- For D3D12_VIDEO_ENCODER_RESOLVE_METADATA_INPUT_ARGUMENTS.EncodeCodec == D3D12_VIDEO_ENCODER_CODEC_AV1, a new resolved buffer layout is defined below.
- Same as existing ResolveEncoderOutputMetadata layout:
- D3D12_VIDEO_ENCODER_OUTPUT_METADATA structure
- WrittenSubregionsCount elements of type D3D12_VIDEO_ENCODER_FRAME_SUBREGION_METADATA indicating each tile in the same order they're written in the compressed output bitstream.
- The maximum size of the resolved metadata buffer can be inferred by the H264/HEVC layout for ResolveEncoderOutputMetadata.
maxSliceNumber = D3D12_FEATURE_DATA_VIDEO_ENCODER_RESOLUTION_SUPPORT_LIMITS.MaxSubregionsNumber (driver reported output);
MaxEncoderOutputMetadataResolvedBufferSize = sizeof(D3D12_VIDEO_ENCODER_OUTPUT_METADATA) +
(maxSliceNumber * sizeof(D3D12_VIDEO_ENCODER_FRAME_SUBREGION_METADATA));
- New memory layout defined as:
- D3D12_VIDEO_ENCODER_OUTPUT_METADATA structure
- WrittenSubregionsCount elements of type D3D12_VIDEO_ENCODER_FRAME_SUBREGION_METADATA indicating each tile in the same order they're written in the compressed output bitstream.
- bSize = tile_size_minus_1 + 1 + bStartOffset
- bStartOffset = Bytes to skip relative to this tile, the actual bitstream coded tile size is tile_size_minus_1 = (bSize - bStartOffset - 1).
- bHeaderSize = 0
- The i-th tile is read from the compressed_bitstream[offset] with offset = [sum j = (0, (i-1)){ tile[j].bSize }] + tile[i].bStartOffset
- D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_SUBREGIONS_LAYOUT_DATA_TILES structure indicating the encoded frame tile grid structure
- D3D12_VIDEO_ENCODER_AV1_POST_ENCODE_VALUES indicating encoding metadata values that are only obtained post-execution of EncodeFrame on the GPU.
- The maximum size of the resolved metadata buffer can be inferred by the layout for AV1 in ResolveEncoderOutputMetadata.
MaxTiles = D3D12_FEATURE_DATA_VIDEO_ENCODER_RESOLUTION_SUPPORT_LIMITS.MaxSubregionsNumber (driver reported output);
MaxEncoderOutputMetadataResolvedBufferSize =
sizeof(D3D12_VIDEO_ENCODER_OUTPUT_METADATA) +
(MaxTiles * sizeof(D3D12_VIDEO_ENCODER_FRAME_SUBREGION_METADATA));
+ sizeof(D3D12_VIDEO_ENCODER_AV1_PICTURE_CONTROL_SUBREGIONS_LAYOUT_DATA_TILES)
+ sizeof(D3D12_VIDEO_ENCODER_AV1_POST_ENCODE_VALUES)