Using Bayer CFA Decoder
Prerequisites
| 1. | Frame grabber
Coaxlink card and firmware-variant with CFA decoder. Open Firmware Variants per Product and search for CFA in the Advanced Processing column. |
| 2. | Camera
|
| a. | Bayer CFA area-scan |
| b. | Up to 16384 pixels per line |
| c. | Having the color registration (GR, RG, GB, or BG) of the first two first transmitted pixels of the first transmitted line and the pixel bit depth ( 8-bit, 10-bit, 12-bit , 14-bit or 16-bit) correctly specified in the PixelF field of the CoaXPress image header. |
When the fields Xoffs and/or Yoffs are greater than 0, the camera must report an adapted PixelF value corresponding to the transmitted data!
Bayer to RGB Pixel Processing Configurations
When the Bayer CFA decoder is enabled:
| ● | thePixel Component Unpacking control is inoperative, the Coaxlink card unpacks 10-bit, 12-bit or 14-bit pixels to lsb |
| ● | the Pixel Component Re-Ordering feature allows to swap the Red and Blue components and deliver either BGR or RGB pixels |
| Input Pixel Format | FFC | RedBlueSwap | Output Pixel Format |
|---|---|---|---|
| Bayer**8 | off | on | off | RGB8 |
| on | BGR8 | ||
| Bayer**10pmsb | off | on | off | RGB10 |
| on | BGR10 | ||
| Bayer**12pmsb | off | on | off | RGB12 |
| on | BGR12 | ||
| Bayer**14pmsb | off | on | off | RGB14 |
| on | BGR14 | ||
| Bayer16 | off | on | off | RGB16 |
| on | BGR16 |
Enabling the Bayer CFA Decoder
In the Data Stream module, set the BayerMethod feature value to Legacy or Advanced according to the desired interpolation method.
Disabling the Bayer CFA Decoder
In the Data Stream module, set the BayerMethod feature value to Disable .
Bayer CFA Decoder Performance
| Product | Firmware variant | Peak pixel processing rate |
|---|---|---|
|
|
1-camera | 1,000 Mpixels/s |
|
|
1-df-camera | 1,000 Mpixels/s |
| |
1-camera |
2,000 Mpixels/s |
| 2-camera | 1,000 Mpixels/s per stream |
The peak pixel processing rate the Bayer CFA decoder approximately matches the performance of the DMA transfer when delivering RGB8 or BGR8 pixels. It is significantly lower than the maximum aggregated input data rate achievable by the board.
The effective pixel processing rate may differ significantly from the peak pixel processing rate when the PCI Express available bandwidth is insufficient or when processing very short lines.
PCI Express Performance
| PCI Express Interface | Sustainable PCIe data rate | RGB8 | RGB10, RGB12, RGB14, RGB16 |
|---|---|---|---|
| 4-lane Rev 3.0 PCIe End-point | 3,350 MB/s typical | ~1,117 Mpixels/s | ~558 MPixels/s |
| 8-lane Rev 3.0 PCIe End-point | 6,700 MB/s typical | ~2,238 Mpixels/s | ~1,117 Mpixels/s |
The PCI interface is not the bottleneck for 8-bit bit depths!
For 10-, 12-, 14- and 16-bit bit depths, the sustainable data output rate is further limited by the performances of the PCI Express Interface on the Host PC.
Latency
The hardware CFA decoder performs on-the-fly conversion with a negligible latency when the data throughput is NOT limited by the available PCI Express bandwidth!
See also in the GenICam Reference (Data Stream Module)