These encoders are built on our OBC-Polar computing platform series (PolarFire SoC based), with an FPGA RTL-based video encoder pipeline integrated alongside the Linux control/software stack.

VREC supports two H.264 operating modes depending on the mission trade-off between compression efficiency and robustness / per-frame access:

• H.264 I/P (Inter-frame) encoding
  Uses I + P frames to achieve higher compression efficiency (better quality per bitrate). The published capability supports up to 4K (3840×2160) @ 60 fps.
• H.264 I-Frame (All-Intra) encoding
  Encodes I-frames only, meaning each frame is independent (useful for deterministic per-frame access and more localized error impact). The published capability is up to 1080p @ 60 fps.

Both modes generate an ITU-T H.264 Annex-B compliant NAL byte stream output

Chroma subsampling / pixel format

• Encoder input (accepted): YCbCr 4:2:2, 8-bit per component (Y, Cb, Cr).
• Encoded bitstream (output): H.264/AVC, YCbCr 4:2:0 video content.
• Output stream framing: ITU-T H.264 Annex B compliant NAL byte stream.

Input support / pre-processing stage (before the encoder)

Many camera sources provide RAW Bayer, RGB, or alternative YUV layouts. Therefore, VREC includes an FPGA RTL-based input conditioning stage to convert common input formats into the encoder-acceptable format (YCbCr 4:2:2, 8-bit/component).

Supported input families (via configurable pre-processing/adapters):

• RAW Bayer: RAW8 / RAW10 / RAW12
  Typical chain: RAW Bayer → debayer/demosaic → RGB → color-space conversion → YCbCr 4:2:2 (8-bit).
• RGB: RGB888, RGB565, RGB8 (mono/greyscale)
  Typical chain: RGB → color-space conversion → YCbCr 4:2:2 (8-bit).
• YUV / YCbCr formats (from bridges and camera interfaces):
  • YCbCr 4:2:2 packed (e.g., YUYV/UYVY) → direct feed (or light adapter) to encoder input
  • YCbCr 4:2:0 semi-planar (e.g., NV12/NV21) → converted to YCbCr 4:2:2 prior to encode. NV12 is defined as a two-plane YUV 4:2:0 format (Y plane + interleaved CbCr plane).

This approach keeps the external camera input flexible while maintaining a deterministic encoder-facing format.

Encoder operating envelope

• Maximum encode capability: up to 4K (3840×2160) @ 60 fps per stream.
• Low-latency behavior: designed for minimal latency; for example, a low-latency configuration ~252 µs at 1080p.
• Bitrate range: Configurable 1 Mbps to 80 Mbps per stream.

Simultaneous streams

Simultaneous streams are achieved by deploying multiple parallel encoder pipelines in the FPGA logic. For the common operating point of 4K @ 30 fps:

• CubeSat platform (OBC-Cube-Polar / MPFS250 class): up to 2 simultaneous streams
• Larger platform (OBC-HYPR-Polar / MPFS460 class): up to 4 simultaneous streams

Video input and platform context

• Video input (selectable front-end): CSI-2 / HDMI / CameraLink (Base/Medium/Full) / SDI / CoaXPress / SD / Composite / etc.
• Networking: Dual 1G Ethernet
• Storage: dual 256 GB eMMC configured in pSLC mode (80 GB + 80 GB)
• Plus the rest of the general OBC-Cube-Polar / OBC-HYPR-Polar platform capabilities (SpaceWire, LVDS, CAN, UART variants, etc.).