You can shape ice with streams of hot air for rapid, inexpensive fabrication.
Water is one of the cheapest materials available, and it freezes and melts at a temperature that’s extremely convenient for shaping, although it has its drawbacks for structural use. But it should be possible to use a form made of water ice to give shape to a mold made of some other material that can harden below 0°, or to make an easily-removable mold for such a material.
Ice has the advantage that, because it’s used over such a narrow temperature range, it has excellent dimensional stability.
Ice is fairly fragile (about 1 MPa tensile strength, which depends greatly on strain rate and very little on temperature), so precisely stamping things with it is pretty much out of the question; to transfer its shape to something else, that other thing pretty much needs to be liquid. By the same token, though, it’s very easy to cut, leading to the ice sculpture competitions in various places every winter, as well as table centerpieces at many corporate events. Other possible ways to shape ice include additive manufacturing (by adding water just above freezing to a below-freezing workpiece) and carving it with high-speed streams of hot air.
For additive manufacturing, it probably would help a lot to include some impurities in the water that increase viscosity and perhaps make the water thixotropic, so that it stays put on the workpiece from the time it comes out of the nozzle until the time it freezes, but without lowering its freezing point too much or promoting large ice crystals. Possibilities that occur to me include tiny air bubbles, gelatin, alginate, agar-agar, carrageenan, konnyaku, gum arabic, xanthan gum, and grease (oil with surfactant), possibly with acids or bases to promote gel formation.
There are a few different hardening processes that could harden the secondary material.
First, obviously, there’s freezing, where the liquid cools down below its freezing point and becomes a solid. This has the disadvantage that it releases a large amount of heat, which can melt the surface of the ice, although this is a less serious problem with low-crystallinity materials such as Dairy Queen dip-cone chocolate-flavored paraffin wax. And maybe, since it will happen first at the surface of contact, it won’t cause any loss of detail. Hot-beeswax casting has long been used to “record the shapes of delicate ice accretions on aircraft components”, according to Reehorst and Richter in 1987.
Second, there’s the hydration-driven recrystallization of plaster of Paris. This happens more slowly at 0°, but I’m pretty sure it still happens. It also has the disadvantage that it releases a lot of heat.
Third, there’s polymerization, which is what Reehorst and Richter reported good results with, starting with thinned silicone rubber resin at -5° and recooled after degassing. Many silicones will not cure at these low temperatures; Reehorst and Richter found that Dow Corning 3110 or 3112 with “catalyst 4” worked well, and Dow Corning HS RTV worked best (10 parts base to 1 catalyst with 1 wt% 20-cSt-viscosity DC 200 silicone thinner). Smooth-On Corporation sells a silicone molding compound recommended by Freeze Cast Engineering for this purpose.