Good, if you know how to model by hand, you’ll have more options. In the best case, I’d advice you to start over, since if you plan to use hand painted textures, they’ll look ridculous on such detailed model. Poly model a new body using old as reference. I believe in using traditional poly by poly modeling for anything that isnt hyper realistic, such as cartoon or anime. In your case, by starting in Zbrush you’ve introduced the problem (absurd polycount) which you later trying to solve. I support everything that was said about low poly previously; in our days, it is less the performance requirement and more a tool to reduce your own workload. In short, use the least polycount that looks good.
But in case that you do not want to get rid of this model, commit to what you already have. Model real mesh clothes, accessories, etc, because hand painting them wont give much performance advanatge. Automatic optimization, such as Decimate, will do more harm than good at this point.
I dedicated my life to the arts, my college / university, Piano (Music). I worked for years as a designer, 3D animator and non-linear editor on a local TV and some advertising agencies. I didn’t want to get too far from the initial idea of the topic. but I think that everything in a game is ‘interconnected’. Art and design decisions are directly related to technical and coding decisions. The same goes for scripts or ‘whatever you want to tell’. I believe that every element on screen, be it static or dynamic, must be related to everything else. I don’t think about a light, for example, if it has nothing to do with the environment and engine code (here about performance). UI decisions, another example, and so on. Everything must have reasons to exist, to exist in a certain artistic and mainly technical (engineering) context. If a static object is on a wall, for example, it is for technical and then artistic reasons. I don’t know if I can express what I think (my English doesn’t help).
When I think of a character, number of polys, I also think of animation (technique) and ‘player control’ (coding). But I need to think about aesthetic elements as well, because it is they or him that links the player to the environment. There needs to be identification and it doesn’t just tell us about ‘gameplay’.
I believe that Unity and current computers can handle or run a lot of ‘in game’ elements. We have ways to mitigate this, the most diverse, from textures / shaders, to camera ‘tricks’. another important factor or factors tell us about the dynamic elements in the game or in any scenario. AI, for example, what is or is not animated, how is it animated, what function does it perform, animation loop, something responsive ?! etc…
So my initial question: is it possible, exaggerating numbers, to have models with 50 / 60k Polys (120k) ‘tris’ in relatively simple environments (static scenario that has a certain number of polys / tris. I know that here involve several things, many , but, literally, I speak of a hypothetical static environment, not responsible, only with inanimate objects)?
If the answer is positive, I think I will be able to better visualize my idea.
In short, yes, it is possible; more important question is how well how performant it will be. There will be many factors other than polycount influencing it. Number of bones in the skeleton and complexity of your shaders, to name a few
Perfect, thanks. I was thinking about it too. I have not yet delved into the topic (specifically about ‘Rig’). I want to study or need, the impact of this on the engine and how it happens. but that will be a step forward. For now I am attending to modeling and Shaders (studying ways / ways / how to mitigate performance, etc.). The retopology of the models is an important part of this, since joints (according to the general design idea) need to be in conformity (‘what moves’, such as joints of arms, legs, eyes, mouth, etc.). After this stage is over, there will be ‘bones’, IKs, ‘controls’, etc.
And I think my next question will have to do with LOD in the player character. Is there, where the idea is a small distance from the camera (controllable, limited)? I understand LOD in static and even dynamic objects, as possible NPCs. I need to understand how they still work (I will have a lot to study, despite the fact that the study material exists in English, which makes things a bit difficult here). An example, NPC A (humanoid, character) whose movements are limited to animation loops, with respect to ‘model’, A, ‘X’ polys, B, some distance, ‘1/2’ of these ‘X’ Polys , greater distance, ‘1/4’ of these ‘X’ polys … etc. These are things that I still need to dedicate myself to studying, although I have a very general (and poor) idea about.
One possible interpretation of LOD is that triangles should be bigger than onscreen pixels when rendered. If they become smaller than pixels, the model MIGHT start looking “noisy”. So if you move the model far away, at the point where the triangles become too small, you dial detail level down.
