I'm using a Raspberry Pi 5 with an OV9281 grayscale camera module plugged into a CSI slot. Libcamera is able to connect to the device, but whenever I attempt to select the 8-bit RAW encoding (R8), it seems to switch to 16-bit encoding instead (R16). This seems strange because the camera module advertises "fmt R8" at the resolution that I'm choosing, and matches it with a score of 0 when I enable Debug logging.
I dug around the libcamera code to try to figure it out, and noticed this suspicious line in pisp's PiSPCameraData::platformValidate: https://github.com/raspberrypi/libcamer ... 4-L1185C74
(Note for others new to this area like me: PiSP appears to be for Raspberry Pi 5+, and vc4 appears to be for Raspberry Pi 4 and earlier.)
In that line it appears to always reset streams that are not 16-bit to be 16-bit, whether they are using packing or not.
My goal with my libcamera project is basically to spit out 8-bit grayscale images that I can them manipulate in OpenCV, so having raw 8-bit grayscale is an easier place to start from as then I can just copy them directly without needing to do any conversion...
I dug around the libcamera code to try to figure it out, and noticed this suspicious line in pisp's PiSPCameraData::platformValidate: https://github.com/raspberrypi/libcamer ... 4-L1185C74
(Note for others new to this area like me: PiSP appears to be for Raspberry Pi 5+, and vc4 appears to be for Raspberry Pi 4 and earlier.)
In that line it appears to always reset streams that are not 16-bit to be 16-bit, whether they are using packing or not.
My goal with my libcamera project is basically to spit out 8-bit grayscale images that I can them manipulate in OpenCV, so having raw 8-bit grayscale is an easier place to start from as then I can just copy them directly without needing to do any conversion...
Statistics: Posted by wrall — Fri Dec 20, 2024 9:04 am — Replies 0 — Views 3