I have this model https://prnt.sc/ghu721 where it has 3.5m polygons, should i load the 3.5m model to a renderer or make a less heavy model and then apply displacement maps of the details?
I see many people spend a lot of time with creating good topology to make static poses without animation. I don’t see why do that.
Some people make a good topology because it can be usefull to make a pose without having much distortion in texture maps or the mesh itself.
It also depends what the purpose is of the model, if it is supposed to be for games or not.
A good topology is also usefull if you plan to rig and animate it after all when you change your mind.
“Good” topology is instrumental in clean deformation, but it does go beyond that.
Well planned topology, which is really what we’re talking about, can ultimately yield cleaner UV islands as well as an overall lower polygon count. It would seem that the latter is only important for games, but high poly work can benefit greatly too. Why waste system on a 60k poly model when its slimmer 6k equivalent will do just fine and absolutely nobody will notice?
More over, a well planned poly flow can also help you avoid basic waste. For example, on the human head, a lot of physical detail goes into the front facing elements. That maybe only accounts for 40% of the entire head. Why waste unnecessary poly detail on the back then? Only create just enough to preserve smooth and even curvature.
As for your second question, WorldDevour, I will refer you back to the beginning of my answer.
Take a 60k lamp model vs a 6k version. Both models may look 100% identical, especially with the camera at a certain distance. The lamp doesn’t deform. However, by planning your topology and cleaning waste, you just saved 54k polys. It might seem like a small savings, but it adds up. When you’ve got a (hypothetical) room filled with 100 such lamps, you just saved 5.4M polys. If you can go from a 6M poly filled room to a 600k poly filled one, why wouldn’t you? Think of the savings to RAM and processing, not to mention storage.
I’ll reframe the issue another way. Take a simple cube, for example. You can have a cube that’s subdivided 100x along every axis. Put it next to a cube at its absolute lowest detail level, one quad per side. Neither will deform. In the former case, you’ve got a 1M quad (2M tri) cube. In the latter, your cube is only made of 6 measly quads, or 12 triangles. Which would you want to have in your scene? Would you rather a 2M tri cube or a 12 tri one? 10 such cubes in your scene… 20M triangles versus 120.
Higher subdivision can also aid in issues of rendering or map building, but those are very specific issues.
For the purpose of this thread, never argue against clean topology. Just because you have a mega system with the latest GPU or access to cloud rendering, don’t think that topology doesn’t matter. It always does. It “reads” cleaner. It leads to swifter rendering of scenes. It saves system resources, which will ALWAYS be at a premium. It aids in easier unwrapping. ETC and so on.
Just be aware of one crucial thing. When retpologizing or optimizing your mesh, don’t sacrifice visual quality if you don’t have to. If extra poly detail aids in curvature/silhouette or deformation, then go with it - within reason. If you’ve got a flat or “flat enough” surface that will never deform then consider lowering the detail level - also within reason. Ultimately, it’s a balance between function and form.
Thanks for the detailed answear, although after these weeks I understand the importance of it!
Great responses all around here. Do remember though that in the case of cubes, and any object with hard edges, you’ll want to bevel the edges a bit. There may be cases where you don’t want to, or where it won’t matter (camera far away might make bevels less useful?), but even a very small scale bevel to round edges makes things more realistic as almost nothing has a non-rounded edge.