Hi hexbabe
The OAK-D camera supports various frame formats. For the AI model input frame, the SDK uses BGR color order and Planar / CHW (Channel, Height, Width) layout conventions. If the model accepts color images, it should accept 3 channels (B, G, R), and if it accepts grayscale images, it should accept 1 channel.
The VideoEncoder node is used to encode ImgFrame into either H264, H265, or MJPEG streams. Only NV12 or GRAY8 (which gets converted to NV12) format is supported as an input. All codecs are lossy (except lossless MJPEG).
The ColorCamera and Camera nodes output ImgFrame messages that carry various formats including BGR/RGB planar/interleaved, NV12 encoded (YUV420, UV plane interleaved), and RAW10-packed (MIPI CSI-2 format) frame data.
The ImageManip node supports input formats including RGB/BGR, planar/interleaved, YUV/NV12, RAW8, and GRAY8. It can convert format to any other format mentioned above.
Thanks,
Jaka