An SLVS-EC sensor sends one packet per line. Each packet carries a header (sent three times), the packed pixel payload padded to a multiple of 16 bytes, and, depending on the sensor mode, a CRC32 footer or Reed-Solomon parity. The receiver uses these header fields:
| Field | Use in the core |
|---|---|
| Frame Start / Frame End | Open and close a frame. Every frame on the output begins with a start-of-frame mark. |
| Line Valid | Marks packets that carry image lines. Blanking packets are not sent as video. |
| Embedded data line | Lines of sensor metadata. Never part of the video output; delivered on the embedded data output when it is built. |
| Line length (sensor-specific header field) | When the sensor reports the line length in the header, the core uses it instead of the LINE_LENGTH register. |
One or two bad header copies are recovered from a good one and counted in
HDR_CRC_ERR_CNT[15:0], an early sign of a marginal link. A packet whose three header copies all
fail CRC16 is lost: it is not used as video and is counted in HDR_CRC_ERR_CNT[31:16]. If that packet was a Frame End the frame joins the next one; if it was a Frame
Start that frame is dropped. The next good Frame Start resynchronises the output, so a single bad header
never shifts the image.