Hello Gergely, Are you using monochrome sensors? The monochrome sensors are just no-mosaic version of the color ones, so there are still separate analog paths and so color gains for the 4 channels (corresponding to Bayer pattern) and I suspect they are not balanced - that may cause 2x2 pixel pattern.
One of the reasons the color channels are not balanced that I suspect is that the clock phase (pixel data to pixel clock) is not set correctly - in stereo camera the cables are long and the signal "eye" is not that wide, so the moment when the incoming pixel data is latched in the FPGA should be set correctly. It is probably our fault - ourselves we only use color sensors where such problems immediately reveal themselves by wrong colors (when the pixels are shifted by 1, green becomes purple. And when pixels go to the wrong color channels, the automatic white balancing (it should work with monochrome sensors too) gets positive (instead of negative) feedback, so instead of balancing colors they go to maximal dis-balance You should be able to see that in the color gains (they are also encoded in the MakerNote field of the Exif header). We'll think of the best procedure to set the clock phase correctly for the monochrome sensor and send you the recommendations later today, Andrey PS. For monochrome sensors I would recommend you to use green band-pass filters - one of the reasons we use color sensors is that we can process each color channel separately and compensate chromatic aberration of the lens. _______________________________________________ Support-list mailing list Support-list@support.elphel.com http://support.elphel.com/mailman/listinfo/support-list_support.elphel.com
