Base questions for all noobies

Hey all! This is my first post here. After learning Blender for about a month I thought I’d ask the answers to a few important questions.

  1. Should we always target low-poly for our meshes and add poly’s later?
    1.2 When would we add poly’s to a mesh?
  2. The documentation is nice and explains what each feature is… but where do we go to learn how and when to use these features?

That’s all for now, thanks in advance.

  1. This very much depends on what style you are creating and you personal preference. If you are targeting a high-poly and high-detail mesh, I usually start of with low poly to get the basics down and add details later. This way, if I don’t like the basic shape, I haven’t wasted hours working on minute details only to scrap the mesh all together. So this really depends on how you like to work, so the simple answer is: Do what you are most comfortable with. I simply recommend starting with low-poly when you’re just getting started with blender, but this is only a recommendation.
    1.2 Again, personal preference. There is no set way when to add details. If you are happy with your basic mesh you can add a bunch of loopcuts and move them around for more detail, or you could do it during the basic mesh modelling.
    2 What I recommend for this is to challenge yourself. What worked really well for me was starting new projects that were just a bit outside of my skill level. This way I’d often come across problems which I had to look up. Because of the huge community behind Blender I’d usually find a good tutorial for that topic. So don’t be bothered by the fact that you don’t know what every button does. Heck, I’ve been using Blender for a couple of years now and there’s still plenty I don’t know. So challenge yourself, get out of your comfort zone and you will learn these features step by step.

I would definitely suggest starting with low-poly first. This doesn’t necessarily mean blocky, just less refined than the end result. The key to efficient modeling is to only use as many polygons as you have to to get your point across. Curvier and flexible objects like expressive faces or clothes are going to need more, while flat and rigid ones like a rock or a cup will need less.

I find a way that really works out for me is to start with low-poly shapes, but keep the different sections separated in different objects, and just move them around until you get a general look that you like. The question you’re going to ask yourself is if you can tell what’s going on in a glance. If it’s a mushroom monster, you should have no hesitation in recognizing the mushroom shape from just the outline of it. This can be adjusted with your camera view to make your intention more obvious, and now is the stage you want to do it.
The idea is you’re going first for volume, sketching out what you want to detail later. Once you have something you like the look of and you’re ready to go deeper, select all your meshes and join them together (not group, make them all one mesh!).
Now we enter the retopo stage, where we’ll add more detail without sacrificing what we liked in the low poly model. Make sure you’re in object mode, and add a new mesh object (Empties and curves won’t work for this, but any mesh will!). Select this new object (make sure it’s the only object selected, since Blender has multi-object editing, and it’ll mess you up) and slip into edit mode. select everything and delete it. CTRL + click to add a vertex, and move it to above the surface somewhere. Apply a shrinkwrap modifier, with the target the low poly mesh. The vertex should snap right to the surface, but play with the setting if it’s not cooperating.
Select your vertex again, and add a new one with CTRL + click; it should add a new one connected to the first one, and continue around the feature, ultimately looping around to sit beside the first vertex. select bot the first vertex and the last vertex, and connect them with the F key. This is how you’ll be adding all your new geometry. The goal here is to section off the parts you care about keeping as they are, or at least very close to how they are, by making loops around those features. Deep crevices also count as special features! Don’t put any vertices in the bottom, just yet, just circle around those areas. Don’t share loops for each section, then should each have their own. You’ll connect them up later, and it’ll only make your work more complicated in the next step.
Once you’ve got your sections, you can tackle each feature individually, one section at a time. It helps keep you from getting overwhelmed. Study each feature and decide what it is you love about it, how you think you can make it better (as in, fit in more with your model’s theme and intention). Now you’re going to rebuild each feature, but first we need to apply the shrinkwrap modifier. Make sure you’re in object mode (don’t worry, it’ll remind you if you’re not), then apply it. You can remake each feature with more focus on the details inside its volume or silhouette. For example, if it’s a container, what kind of container is it? If it’s meant to hold some arcane treasure, think about how this feature can imply that conclusion. Bear in mind, this is with shape alone, such as a rune or fancy padlock, maybe some rivets or leather straps holding it closed. Small textures like woodgrain and rust should happen on the last or nearly last step, ideally with image maps for bumps and normals!
This brings us right to your second question: how and when to use each feature!
It boils down to two simple criteria: what does this feature do, and will it make what I’m trying to do right now easier/faster? For example, the results of the shrinkwrap modifier can absolutely be achieved manually, but it’s a lot faster to use it than to position each vertex by hand. The f button (it connect vertices and lines into lines and faces) is faster and more precise than extruding each part, and loopcuts and edge sliding make it easier to add geometry to existing geometry without have to wreck what work you’ve already done. Features like decimate or curve deform might not really have an application for what you’re doing at the moment, so you wouldn’t be using them.