Note the 3d crystallographic model of penicillin. These were not meant to be exhibition pieces, it was what people used before interactive computer graphics for

Note the 3d crystallographic model of penicillin. These were not meant to be exhibition pieces, it was what people used before interactive computer graphics for structure refinement. You had imperfect phases from somewhere and would calculate a Fourier map (perhaps only a projection along the axes as seen here), then you would construct your model and make it conform to the calculated density and chemical knowledge as best as you could, you'd measure atomic positions with calipers and from these derive new phases that you would feed into another cycle.