hi_video_frame

Description

Defines a source video frame.

Prototype

typedef struct {
    hi_u32              width;
    hi_u32              height;
    hi_video_field      field;
    hi_pixel_format     pixel_format;
    hi_video_format     video_format;
    hi_compress_mode    compress_mode;
    hi_dynamic_range    dynamic_range;
    hi_color_gamut      color_gamut;

    hi_u32              header_stride[HI_MAX_COLOR_COMPONENT];
    hi_u32              width_stride[HI_MAX_COLOR_COMPONENT];
    hi_u32              height_stride[HI_MAX_COLOR_COMPONENT];

    hi_u64              header_phys_addr[HI_MAX_COLOR_COMPONENT];
    hi_u64              phys_addr[HI_MAX_COLOR_COMPONENT];
    hi_void*            header_virt_addr[HI_MAX_COLOR_COMPONENT];
    hi_void*            virt_addr[HI_MAX_COLOR_COMPONENT];

    hi_u32              time_ref;
    hi_u64              pts;

    hi_u64              user_data[HI_MAX_USER_DATA_NUM];
    hi_u32              frame_flag; 
    hi_video_supplement supplement;
} hi_video_frame;

Members

Member	Description
width	Image width.
height	Image height.
field	Field mode. Reserved.
pixel_format	Pixel format.
video_format	Video format.
compress_mode	Video compression mode.
dynamic_range	Dynamic range. Reserved.
color_gamut	Color gamut. Reserved.
header_stride	Compression header stride. Reserved.
width_stride	Width stride of an output image component. For a YUV image, this member indicates the stride of the Y, U, and V components. For an RGB image, this member indicates the stride of the R, G, and B components.
height_stride	Height stride of an output image component. For a YUV image, this member indicates the stride of the Y, U, and V components. For an RGB image, this member indicates the stride of the R, G, and B components. This parameter is reserved for VI video capture, ISP system control, VPSS video processing, VENC video encoding, and JPEGE image encoding.
header_phys_addr	Physical address of the compression header. Reserved.
phys_addr	Physical address. Reserved.
header_virt_addr	Virtual address of the compression header. Reserved.
virt_addr	Start virtual address of an image in the device buffer. For an input YUV image, virt_addr[0] is the start address of the image or Y component, virt_addr[1] is the start address of the U component, and virt_addr[2] is the start address of the V component. For an output YUV or RGB image, virt_addr[0] is the start address of the image on the device, and virt_addr[1] and virt_addr[2] are reserved.
time_ref	Sequence number of an image frame. This parameter is reserved in the decoding scenario. This parameter is reserved in the JPEGE scenario. In the VENC scenario, set this parameter to an even number in ascending order based on the frame sequence. For example, set this parameter to 2 for the first frame, 4 for the second frame, and so on.
pts	Image presentation timestamp (PTS).
user_data	Private data. Reserved.
frame_flag	In the VDEC decoding scenario, this parameter indicates whether a frame is successfully decoded. You need to free the input and output buffer in time regardless of whether the frame is successfully decoded. The system does not guarantee that the output order of frames that fail to be decoded is in the display sequence or decoding sequence. Value options of this parameter are as follows: 0: The frame is decoded successfully. 1: The frame fails to be decoded due to abnormal input stream or inconsistency between the configured stream format and the actual stream format. 2: can be ignored. In the interlaced scenario where two fields are transmitted every frame, one of every two transmitted fields has no decoding output. This is a normal phenomenon. The decoding output of an interlaced stream is stored in the output buffers of the odd fields. In this case, you need to free the buffer in time. 3: The frame fails to be decoded because the number of reference frames is incorrectly set. 4: The frame fails to be decoded due to incorrect size of VDEC frame buffer. In the JPEGD scenario, this parameter indicates whether the frame is successfully decoded. You need to free the input and output buffers in time no matter whether the frame is successfully decoded. Value options are as follows: 0: The frame is decoded successfully. 1: The frame fails to be decoded due to abnormal input stream or inconsistency between the configured stream format and the actual stream format. 4: The frame fails to be decoded because the size of the JPEGD decoding output buffer is insufficient. You can obtain the size of the output buffer by calling hi_mpi_vdec_get_jpegd_output_info in advance. 5: The JPEGD decoding configuration parameters are incorrectly set. For example, the configured width or height exceeds the limit or the configured target decoding format is not supported. This value is used only in the JPEGD decoding scenario. In the VENC scenario, different bits are used to enable different functions: Bit 0 of this parameter indicates whether to enable the slice_temporal_mvp_enabled_flag syntax (indicating whether to use the time-domain motion vector (MV) for inter-frame prediction). The value range is as follows. Only the value of bit 0 in the H.265 protocol takes effect.: 0: temporal MV prediction enabled, with slice_temporal_mvp_enabled_flag set to True. 1: temporal MV prediction disabled, with slice_temporal_mvp_enabled_flag set to False. Bit 1 of this parameter indicates whether to enable image edge expansion. The value range is as follows. The value of bit 1 takes effect in the H.264/H.265 protocol. 0: image edge expansion enabled. 1: image edge expansion disabled. Bits 2 to 7 of this parameter indicate the value of chroma_qp_index_offset (offset of the chroma component quantization parameter value) in the stream. Only the values of bits 2 to 7 under the H.264 protocol take effect. The requirements for setting bits 2 to 7 are as follows: Bit 7 is used to identify whether the value of bit 2 to bit 6 is a positive number or a negative number. If the value of bit 7 is 1, it indicates a negative number. If the value of bit 7 is 0, it indicates a positive number. The value range of bits 2 to 7 is [–6, 18]. The system subtracts 6 from the configured value to get the final value. Therefore, the value range of chroma_qp_index_offset in the encoded stream is [–12, 12]. In the JPEGE scenario, the value is 0 or 1. However, this parameter is not supported in the current version, and the parameter value is invalid.
supplement	Supplementary information about an image. On the Atlas 350 Accelerator Card, this parameter is a transparent transmission parameter. On the Atlas A3 training product/Atlas A3 inference product, this parameter is a pass-through parameter. On the Atlas A2 training product/Atlas A2 inference product, this parameter is a pass-through parameter. This parameter is used by the VI module of the Atlas 200I/500 A2 inference product and is automatically initialized by the VI module. For other functions, this parameter is a pass-through parameter. On the Atlas inference product, this parameter is a pass-through parameter.

Considerations

In general, 10-bit data is compactly stored in the memory in non-compression mode.

For 10-bit YUV data, the first 8 bits and last 2 bits are stored separately in segment-based compressed mode.

For 10-bit tile 64 x 16 data, which is video decoder hardware (VDH) output, the first 8 bits and last 2 bits are stored separately in the memory, regardless of whether the data is compressed or not.

Parent topic: Common