After we got our camera system working on the Orin thanks to this forum post, we are seeing very different image captures between it and our Xavier using the same hardware setup.
We adjusted our PWL curve to fit within the specified limits and applied it to both our Xavier and Orin. The Xavier image looks reasonable:
But the Orin image is very dark:
We are using the same camera board and the same image sensor settings for both captures.
We are using a simple logarithmic PWL curve:
mode_type = "bayer_wdr_pwl";
dynamic_pixel_bit_depth = "16";
csi_pixel_bit_depth = "12";
num_control_point = "9";
control_point_x_0 = "0";
control_point_x_1 = "446";
control_point_x_2 = "1617";
control_point_x_3 = "3469";
control_point_x_4 = "6597";
control_point_x_5 = "13365";
control_point_x_6 = "21217";
control_point_x_7 = "37300";
control_point_x_8 = "65535";
control_point_y_0 = "0";
control_point_y_1 = "97";
control_point_y_2 = "255";
control_point_y_3 = "452";
control_point_y_4 = "732";
control_point_y_5 = "1243";
control_point_y_6 = "1758";
control_point_y_7 = "2684";
control_point_y_8 = "4095";
Should we expect captures from an Orin to be the same to captures from a Xavier with the same general settings? We have not had ISP tuning done by our imaging partner yet. Will tuning remedy this?
We would like to use the Xavier to perform data capture for models that will eventually be used on the Orin, so we need to make sure the captures will be identical between the two.