Is there anything to the process of 3D printing that might make the dice unfair/land on a specific side more frequently? Or will any imperfections be negligible, thus making the dice fair? I'm more concerned about making a d20 than anything else, so focus on that shape if need be.
I am using a MakerBot Replicator+ with FDM, so take this into consideration if it affects your answer, and if you believe a different printing process would have an advantage over this don't hesitate to include it.
Incidentally, I have a decent amount of experience with 3D printing.
It depends heavily on the particular technology that you're using to do the 3D printing, but nearly all forms of printing aren't perfect--depending on the design you make (solid? honeycombed? hollow?), there will be slight, or not-so-slight variations across the faces, which will result in an unfair die. In particular, the faces that are parallel to the build surface will probably be different compared to the ones that are angled.
You could probably deal with some of these problems with some kind of post-processing, but honestly at that point buying dice would be easier. 3D printing is for making custom or prototype objects, and isn't really suited to mass production of simple shapes like dice.
Even manufactured dice have some imperfections, and a slightly unfair die probably won't make much of a difference in gameplay, but it will be hard for you to determine whether the imperfections from 3D printing are severe enough to make the die unfair without rolling it many times and looking at its distribution. Again, at that point, it's easier to just buy the dice.
That being said, custom d20s can be super cool looking, and if you're going to 3D print something, might as well make it look cool...
I've printed a few dice, and thrown them ~100 times each to check -- they were very far off from fair. One reason, I think, is that the inner fill (and thus a lot of weight) is oriented a certain way. It might help to do solid-fill or no fill, but I haven't compared yet.
Custom 3D printed dice are distinctly lower quality than the dice from a FLGS you see in 'standard' dice sets, and around the same quality of die as the cheaper plastic dice available online. It depends somewhat on your printer, but you can get very uniform results in terms of material density with the issues in die fairness being limited to problems with the surfaces (e.g. the sides aren't usually all the same texture). You can fix this with statistical modelling and sanding to get a very good die, but it takes forever and is a huge pain and is totally not worth it from a practical standpoint. A quick sanding to get all faces roughly the same texture is usually good enough, in my experience. You can also just print the die so that the build-surface face and the face parallel with it (if it's not a d4) are unimportant faces, like the '10' and '11' on a d20-- being smoother they'll be rolled a little less often but the average roll will be the same regardless of what number pair that effects if you keep the 'opposite sides sum to die size + 1' rule. A quick sanding doesn't take too long and works well, though; that's definitely what I would recommend from my experience.
As a note, the main advantage in 3d printing dice is that it's affordable, whereas otherwise acquiring truly custom dice (weighted to specification, unusual number of sides, etc) requires you to make them by hand or shell out many thousands of dollars. 3d printed dice are particularly useful when you want to concoct biased dice, as you can easily alter things like the physical die size, weight (by altering how much of the interior is hollow), weight distribution (by altering where the hollow cavities are-- known as 'floaters' when these gaps are used to bias the roll, much less good than weighting), and number placement to conceal the differences between your die and a 'fair' die you 3d printed.
In my experience, 3d printed dice work fine for RPGs, but feel a little light (at least with my feedstock). I used them for a bit with a group I GMed for while on a bit of a 3d printing kick, and they definitely have some niche uses, but beautiful high-quality dice with a good feel to them are so cheap nowadays I just don't see any reason to go to the expense and trouble when you can avoid it.
I have tried to make an unfair die by putting 100% infill in the conner opposite to 6, 5 and 4.
When you float it, it always comes with expected side up. So with floating you really can check if somebody has tried to make it unfair.
The real-life tests have unfortunately shown my complete failure :)
After 104 throws the average is 3.49. Even less than "fair" 3.5!
It looks like the weight of plastic is too low to compete with hitting and rotation. The next try could be to put something metallic into it during printing.
Pretty much all die are unfair to an extent. If you are seriously concerned about accuracy then you will need to test what you produce and, if not satisfactory, modify the design reprint and retest. A simple/coarse test is to float your die in very salty water and note which face faces upwards, repeatedly perturb the die noting the upwards face. Any egregious bias should reveal itself after a handful of perturbations. Note that this test only reveals that a die has a bias, not which face it is biased to land on. It's also probably not massively accurate but should be good enough to let you gauge whether your printed die are any worse than your shop bought ones.
Well it is a bit more complicated than that. Changes are the d20 you buy in the store is a bit more favored as well. Now we are talking statically, it will like one side over the other. As the difference in weight distribution is not drastic if you print it solid, it will not matter. In use you likely will not see it.
That said if you wanted to be extra sure.. I would use the 3d printed model, make a silicon mold with it. Then use Resin to make the dice. Now you can start making whole batches of dice as good as what you get in the store.
My final answer is, technically it will favor but you will not notice.
Not likely.
Creating true 'fair' dice is fairly difficult to do even with high tolerance processes. 3D printing (hobbyist level) is much lower tolerance and will definitely be weighted incorrectly especially if the infill is anything less than full. Professional grade 3D Printers will be better and more accurate but will still likely be weighted slightly to one side.
Regardless of process though it would be possible to create a fair dice using 3D printing as the primary means of manufacturing and then weighing, balancing and doing some post processing to even them out.
