VideoDecodeConfig

width

int

Buffer width for video decoding. The value must be an even number, which ranges from 128 to 4096 and defaults to 1920. It is automatically rounded up to a multiple of 16.

Set this parameter based on the actual video frame data. The value must be greater than or equal to the actual video frame data width. Otherwise, no decoding output is available. If the value is too large, redundant memory resources are generated.

height

int

Buffer height for video decoding. The value must be an even number, which ranges from 128 to 4096 and defaults to 1080.

Set this parameter based on the actual video frame data. The value must be greater than or equal to the actual video frame data height. Otherwise, no decoding output is available. If the value is too large, redundant memory resources are generated.

inputVideoFormat

StreamFormat enumerations

Input video format. H.264 videos support base.h264_baseline_level, base.h264_main_level (default), and base.h264_high_level. H.265 videos support only base.h265_main_level.

outputImageFormat

image_format enumerations

Output image format after decoding. The Ascend 310 AI Processor supports base.nv12 (default) and base.nv21. The Atlas inference products also support base.rgb and base.bgr besides the aforementioned formats.

skipInterval

int

Skipped frames during video decoding. The value ranges from 0 to 250 and defaults to 0.