I am thinking of something similar. I think I will indeed attach the wedge to the bottom of the glass to be in contact with the film, rather than above the glass like the original. But rather than let the glass be loose and depend on its weight for pressure, I will put thin black paper under the film to press it upwards. The glass is not heavy enough to do much, so shimming under the film is probably the best way to go with this design. I think I will have to filter the light to reduce exposure rather than increase, so I can lay the ND filters on top of the glass and keep the time to the original 1 second. Longer than that and I start getting into reciprocity failure. I could also replace the original spacer pictured above with a piece of processed film, which would make the gap smaller. The spacer serves no other purpose.