Hello,
1) The range of the detection (1 and 2 meter) depends greatly at which range the object is added to find_object. Features may be different with an object looking smaller (farther) in the image. A solution can be to add the same object in find_object at different scales. You may then have detection inside the range of the XTION (<4 meters). Increasing the image resolution will not help for the distance for the 3D version of find_object (which is more dependent to depth resolution).
2) Theoretically it can work with any resolution. But in practice, it depends on the sensor. Here the depth image of the XTION cannot have a resolution higher than VGA. If you setup OpenNI2 with SXGA, it may not publish a registered depth image. In that case, the
ROS callback of find_object node won't be called. When you say the image is fix, is the GUI still responding?
cheers