I implemented the simple generic decode using the dcraw model. I see that the resulting image quality is sometimes fine, sometimes problematic (brightness and color conversion incorrect). What am I missing? Am I expected to understand each camera's characteristics and add post-processing to correct the image?

Outline of implementation:

librawdata->params.use_camera_wb = 1;
librawdata->params.output_color = 1;
err = libraw_dcraw_process(librawdata);
RAWimage = libraw_dcraw_make_mem_image(librawdata,&err);