combinatorial blow-moulding

It is widely known that people buy bottled water for the packaging.
Clearly, making that packaging more interesting would be an advantage in the market-place. As far as I am aware, all commonly available disposable drinks bottles are highly uniform - perhaps for cost reasons, or in an attempt to imprint the brand on the consumer's subconscious.

The packaging for bottled water is ubiquitously blow-moulded. There are several interesting variations on this, but these all involve (at least) two mould parts which are transiently clamped together to create the moulding cavity.
I propose that multiple interchangable mould designs be used for each partial mould, and these be used cyclically in all combinations. The main restriction is of course that the edge profile of all moulding parts within each set must be identical. An easy way to allow all combinations to be used is to make the set size for each moulding part relatively prime - that is, have no common denominators (other than 1), and simply cyclically iterate through each set.
This has the desirable property that no two bottles in a case (of fewer than the smallest moulding set) will have a part in common.

One variation of blow-moulding which appears superficially amenable to this is "rotary wheel blow moulding". If wikipedia is to be believed, rotary wheels within these machines typically contain from six to thirty moulds. With two moulds, up to 870 combinations would be available (for wheels of 29 and 30 impressions).
It is of course the case that at least one wheel would have to vary its rate of rotation - with one speed matching the other wheel to allow casting, and a second speed to bring the next casing into position. This issue, and the need for a threaded neck, suggest that the injection stretch blow moulding process might be appropriate - which would potentially still be workable with a set of moulding wheels.