Skip to content

Applications

4K60 encoding with on-board AI.

YUAN's 4K60 smart encoder combines broadcast-grade capture, AV1 compression, and on-board AI that can write analysis results into stream metadata during encoding.

Video production control room with multiview displays
15 µs
YUAN-reported capture latency in the tested GPUDirect RDMA configuration
4K60P
broadcast-grade capture: 4:4:4, 10-bit, VANC
70/100
TOPS options on Jetson Orin NX for in-encoder AI

The challenge

Reduce manual tagging in 4K workflows.

Broadcast workflows are moving from HD to 4K while contribution networks move from 4G to 5G. The resulting H.264 and H.265 streams create large storage and conversion workloads, and footage often still requires manual tagging before editing or replay.

Edge-AI systems can now extract text and metadata from video during processing. Placing that analysis in the encoder reduces the need for a separate post-processing step.

Broadcast control room with equipment racks and multiview screens

The capture

YUAN reports 15-microsecond capture in a tested configuration.

The encoder pairs NVIDIA Jetson Orin NX with YUAN SC710 capture cards for 4:4:4 chroma subsampling, 10-bit depth, and VANC capture. YUAN reports capture latency as low as 15 microseconds with NVIDIA GPUDirect RDMA in its tested configuration.

Jetson Orin NX adds AV1 compression. Bitrate and quality relative to H.265 depend on the encoder settings, source content, decoder, and measurement method.

Four-input video capture card on a control-room desk

The transport

One SDK for broadcast transport protocols.

The NexVDO SDK supports RTSP, RTMP, TS, NDI, SRT, Dante AV, and YUAN's SkyLink X ultra-low-latency protocol. This keeps capture, encoding, and transmission in one software workflow.

CBR streaming keeps network transmission rates under stable control for production broadcast use.

Video equipment racks inside a production truck

The analysis

Write AI analysis into stream metadata.

With 70 or 100 TOPS on board, the encoder runs AI and generative-AI algorithms during encoding. Stream Information Extraction synchronizes encoded audio with AI-generated information and writes the analysis into stream metadata.

Post-production teams can receive searchable footage with important timestamps already marked.

Video capture card displayed on an electronics workbench

Pipeline

The signal chain

  • 12G-SDI + HDMI in
  • SC710 4:4:4 10-bit capture
  • GPUDirect RDMA 15 µs
  • AV1 encode on Orin NX
  • In-stream AI metadata
  • NDI + SRT + SkyLink X out

Hardware

Related products

SC750N1-L HDMI2.1

Capture I/O

SC750N1-L HDMI2.1

1 CH 8K60/4K120 HDMI2.1

View product
SC750N1O1 12G-SDI Genlock

Capture I/O

SC750N1O1 12G-SDI Genlock

1 CH 4K60 12G-SDI Genlock

View product
SC6T0N4 12G-SDI AI+

AV over IP

SC6T0N4 12G-SDI AI+

4 CH 12G-SDI 4KP60 Encoder + Decoder + Switch + AI

View product
SC6F0N1 HDMI2.0 PRO

AV over IP

SC6F0N1 HDMI2.0 PRO

1 CH HDMI2.0 4KP60 Encoder + Decoder + Loop

View product
HDMI2.0 to Dante AV Ultra

AV over IP

HDMI2.0 to Dante AV Ultra

Transfer 4K60 Image with Low Latency

View product

Review a broadcasting system requirement

Share the signal sources, required output, host constraints, and project stage with the YUAN USA product team.

Send system requirements