Hello Marc, The sensor data goes through some processing steps and then recorded to the memory in line scan order (padded to multiple of 512 bytes). It can be recorded in either 8 bits per pixel (after scaling, optional FPN subtraction and gamma correction) or in 2 bytes per pixels. It is 16 bits signed, with no subtraction MSB is always 0 and 12 bit sensor data is left shifted by 3 bits (other types of sensors have different shift - older 10 bit - by 5 bits, CCD output in 323/363 cameras - by 1 bit).
You can copy the whole video memory to the computer by ftp (downloading through the web server of this file does not work) from the link /usr/html/fsdram (same as /dev/fsdram). When copying in konqueror it will report an error at the end and leave fsdram.part , but it will be a correct file. Both drivers /dev/ccam_img and /dev/fsdram are slower than reading the compressed image that goes through DMA - these drivers are made for development/testing. The location of the frame in the memory is defined in x353.h file (very end), currently it is in the middle of the memory (leaving first half for FPN data). You can read the first line in telnet session as fpcf -sr 1000000 600 # there will be garbage after 1000512 - line is padded to 0x600, there will be 0x512*2=0xa24 pixels read from the sensor line (default settings) fpcf -sr8 2000000 c00 # same data in byte format. The pixel data goes through the sensorpix353.v and them to mcontr353.v (it implements a 4-port SDRAM controller), every frame. port 0 - sensor data (memory write) port 1 - fpn data (memory read) - not used by current software port 2 - data to compressor (20x20 tiles, memory read) port 3 - data to/from the CPU in PIO mode (memory read/write), used in /dev/ccam_img and /dev/fsdram Andrey
_______________________________________________ Support-list mailing list [email protected] http://support.elphel.com/mailman/listinfo/support-list_support.elphel.com
