I find the time to create the 3D drawing is not much different to that required to scratchbuild a similar quality model. More than half the time is working out the measurements, etc. from photos and drawings which would have to be done anyway, rather than the actual drawing time. And having taken the time to create a 3D drawing, I can produce multiple copies of a model without minimal further effort. Some of the wagons I am doing for my own layout were very common in real life, so I need 10 or more of each type.
I don't use the grainy WSF/Polyamide material for rollingstock models. The only thing I use it for is 3D print sheep loads for sheep wagons, because the grainy texture is an advantage. I also use the Alumide material for sheep, which has a slightly rougher texture still, and it also somewhat heavier which helps add weight to a sheep wagon. (Alumide is a mix of powdered Nylon and aluminium, WSF/Polyamide is just powdered Nylon.)
Some of the 3D printed models on offer have "facetted" surfaces on parts which should be round, e.g. a loco boiler. This is not a limitation of the 3D printing process but a limitation of some entry-level software packages which can't actually draw a proper circle or cylindrical object. There may well be some stepping on curved surfaces due to the 3D printing process, depending on the print orientation. I use i.Materialise in Belgium and they will at least allow the customer to request a particular print orientation, although they don't guarantee to use that orientation. I find that the print orientation is as requested at least 90% of the time.
My favourite material is "Prime Gray" from i.Materialise (not available from Shapeways) which is fairly similar to conventional modelling plastics in many respects. It is quite stable (My oldest test prints are now 2 years old and in perfect condition.) and can be cut, drilled and tapped much the same as styrene. It is flexible to some extent, but small parts can break suddenly if over-stressed. Prime Gray is done using a Stereolithography process (liquid resin cured by a laser, layer-by-layer).
Most of my models have 1 or 2-piece bodies. I generally use 2-piece bodies for models with curved roofs to avoid stepping on the roof. Due to the pricing structure for Prime Gray, it is not practical to create a "kit" with many parts as there is a starting price of 2.5 - 5 Euros per part (plus a charge based on the volume of material in the model) and i.Materialise don't allow "sprued" parts. I try to include as much detail as is practical on the model, and add handrails, etc. later from wire. Prime Gray reliably reproduces details down to 0.5 mm. The layer thickness is 0.1 - 0.125 mm.
Some of the models I have done using 3D printing are pretty complex and would be difficult to produce by scratchbuilding or conventional low-volume casting techniques. I have generally avoiding doing models which are already available from local "cottage industry" modellers which are typically resin castings.
Due to the relatively small number of people actively modelling WAGR narrow gauge in S scale (about 50) there is virtually no prospect of anyone putting up the cash for injection moulded models or kits, so 3D printing offers a way forward for us modellers of the more obscure railway systems. One of the good things about 3D printing is that there are minimal start-up costs and that models can be economically produced in small-medium quantities "on demand" so there is no major financial outlay for moulds, etc. or having to do a "run" of a certain number of models which may or may not sell. Also, minor variations found on the prototype can be easily included, without having to make new moulds, etc.
Also, if there are significant advances in quality vs. price in the future, the models I am designing now can always be reprinted to a higher quality level. I don't see any harm in building up a "library" of 3D designs.