So I’ve gotten around to playing with the tessellation settings in UE4 and I have to admit, it’s not as resource hungry as I first thought it would be.
However it made me think, is it really practical to use?
The whole purpose of tessellation is to increase polycount, but for the resources it costs, why not just model a higher poly object instead?
For example, take a standard Fire Hydrant (what I was using to test with). The low poly mesh is 1,300 polygons and the high poly is 40,000 polygons. Now, you could just apply a couple of tessellation multiplers on it. But at the cost of one hydrant, why not just model a 5,000 poly hydrant instead and spread them throughout a level?
Same goes for displacement mapping. You could tessellate the ocean, or why not just make the ocean higher poly to begin with?
If I had to imagine a practical scenario for tessellation, it would have to be in a very linear/corridor like game. Trying to model something like individual bricks on a house, or stone pavement would be much more easier using a displacement modifier and since the level is so small, you will recognize those details.
I also think another huge problem with tessellation is animation. At least static objects will always remain still and thus require less processing cycles. But actual human characters will need to move and express themselves.
