The experiments to get the best images out of the x-ray machine have produced some useful results.
Producing a good x-ray relies on getting the x-rays to just pass through the material. You want them to pass through the slightly less dense areas and not to pass through the denser areas. This is how you pick up the internal structure. You have to judge the power so that the x-rays just pass through the material so that these differences in density are significant enough to affect the image on the x-ray plate. 
The x-rays generated by the machine can produce lower energy secondary x-rays when they hit other things (like the bottom of the x-ray machine). These low energy x-rays can also affect the x-ray plate and fog the image slightly. You can put different filters in, in different places to remove these secondary x-rays but you have to judge it so that you get the best results. 
I have been doing a lot of human bone, at the lower end of the exposure range of the machine. Bone is not very dense and lower energy x-rays are needed (between 65kv and 80kv). Having tried exposures with and without a filter underneath the plate, the general consensus is that the image is slightly sharper with the filter in place. However, putting a filter on top at these low kv values, produces a very ‘thin’ result (see figure 1) and detail is lost. 

At the other end of the scale, you definitely need a top filter at high kv values. The image is very much fuzzier if you do not use a top filter when x-raying thicker metal (see figure 2). The trick has been to work out at what point you need to add the top filter. The cut-off point appears to be 100kv.