Perhaps I'm not being clear in my intentions.
Lets take 2 cylinders, an Al S80, and a Steel HP80 (these are the two I have got data for)
The dry weight of the S80 is quoted to be 31lbs or 14kg and it can be filled to 3000psi or 207 bar. It has a buoyancy swing from -2lbs (-0.9kg) full to +4lbs (+1.8kg) empty
The dry weight of the HP80 is quoted to be 28lbs or 12kg and it can be filled to 3500psi or 240 bar. It has a buoyancy swing from -9.lbs (-4kg) full to +3lbs (1.3kg) empty
So for a similar empty volume the steel weighs less, both have a similar buoyancy shift of 6lbs (2.7kg) but the steel stays negative and of course holds more gas. Both have the same internal volume , but the steel will have a smaller external diameter because the wall thickness is less but because of the extra pressure holds 10% more gas when filled
If we put a Steel HP120 into the mix which is around 15l? Then we get the following: dry weight of 39lbs or 17kg and it can be filled to 3500psi or 240 bar. It has a buoyancy swing from -10lbs (-4.5kg) full to +1.3lbs (0.5kg) empty
A 15l holds 40% more gas (if my calculations are correct HP 80 = 82cuft gas which is 2321l which if divided by 240 = 9.6l) but its dry weight is only 30% greater.
So what if we can make a cylinder out of "fantasium" it could in theory have the same dry weight as a 12l steel if it were stronger and have similar buoyancy characteristics i.e. remains negative at the end of the dive if it had the correct density.
Of course I could be talking gibberish as I am on cold and flu drugs at the moment - but in my head it makes sense