3 ms·
To try and give a slightly better answer then others, without complicating too much. My thought it is because they use the default shaders (and/or the artists
by gens 9y ago
To try and give a slightly better answer then others, without complicating too much.
My thought it is because they use the default shaders (and/or the artists do glossy too much). That said..
Most, if not all, modern big engines use "deferred shading" and support normal/bump maps, gloss/reflection maps in addition to "normal" textures (lately every rendering engine that focuses on being realistic is switching to "Physically based rendering", that is very similar to "normal" rendering except that it obeys the "amount of (light) energy out can not be greater then the amount of energy in"(this also means linear light space, instead of logarithmic like we humans perceive)).
All the big rendering engines can, with the same equations in shaders, produce exactly the same result. The same results in regards to the direct "shading".
The other part of how a thing looks is lighting (aka "global illumination"), where the actual engine code is a lot more involved. Since you can't make, like raytracing does, light bounce off of things, you have to cheat (it uses too much processing power). Rendering engines cheat by, for example, plunking "light probes" around (unity had a good post about them, i can't find it now). The slight differences in lighting may not seem like much in a side by side comparison, but our brains use those little lighting clues meaning they(we) are sensitive to them.
There is also how many lights a rendering engine supports (and the distance they can cast light), how it does particles and can they emit light, how it does volumetric.. things (like smoke, maybe water and such), how good it is at reflections (are they precise, do they include dynamic things; note that they probably don't bounce twice), subsurface scattering (notable on skin), and probably more that i forgot.
Note that i simplified plenty. For example lighting and shading are closely related.
PS Reality is not like basic theory. Physically based rendering does give real world results but in reality everything has dust on it, or an oxide layer (aluminium especially), or a sulfide layer (copper likes sulphor). The other imperfection is the minor imperfections in smoothness, what the collective "they" do take into equation (see https://www.cs.cornell.edu/~srm/publications/EGSR07-btdf.pdf https://www.cs.cornell.edu/~srm/publications/EGSR07-btdf.pdf).
PPS "Physically based rendering".. as if Phong shading is based on something else. (PBR is still no where near reality, for reasons above)