I wondered when it is best to add the subsurf modifier :
If i add it at the beginning of my modeling process, i can see what my final model will look like, but as i need to add supporting edges to prevent my model to be too “limp”, it creates a lot of geometry which makes the modeling more complicated.
On the other hand, if i don’t add it before the end of the modeling, my object is hard edged, and thus do not represent what it will finally look like.
For me it doesn’t matter when it gets added because I already planned for it and modeled such that I already know what it’s going to do, which is round the curved surfaces, all preferably with similar density.
+1 to SkpFX and Macser
My workflow when modeling tends to go with a very basic shaping or outline of an object (so box modeling or with a low poly circle), then I turn on the modifier and model out the problematic spots with extra loop cuts, faces, whatever’s needed.
There’s a tutorial on youtube (https://www.youtube.com/watch?v=VnCFumSfg_s) about modeling a human skull pretty low poly. The guy’s kind of boring, and it’s not the newest version of Blender - plus I don’t care for how he set up his reference images - but the key here is, if you go through a couple of the videos (which has multiple parts), he explains very well how to follow the flow of topology.
Yes, it’s not that critical when the modifier gets added. Might feel awkward working on the model when the modifier gets added early on but there are options to help with that:
Modifier header has these visibility toggles, first one is for render visibility but when working
modifier visibility in the viewport. Edit mode has its own toggle which is next, but you could toggle between edit and object modes to see the subdivision result
Modifier result also visible in edit mode, the default, so you see the control cage and the subdivided version
Subdivision result enabled on the control cage
Could also enable matcap from the 3D view properties (hotkey N, opens on the right by default) which can help viewing the forms when the default visibility option is used on the modifier
It might be tempting to just enable modifier visibility on the control cage (3) and think that is the answer, but that might be the worst option to choose. I strongly recommend not to do that while building the model structure and leave that option for adjusting the forms, after you already built the structure for it.
When the control cage gets subdivided it no longer show you the actual vertex positions, and you’re likely to end up with a mess: geometry overlapping, vertices too close to each other so they get merged, severely distorted faces where you should have a different structure.
(Don’t mean her model is a prime example of bad modeling, not even close, just an example of what happens when people get too fond with the option)
Also worth noting, if you haven’t already, that there are three separate things you pay close attention to but work on them separately: structure, interpreted form, and the form you build. You want to describe a form in 3 dimensions by building the structure, but the form you end up with is the result of the structure you built. Would be great to only do one of those things but the structure usually has to fulfil many other requirements than just the form it describes.
When the subdivision result is enabled on the control cage you’re only concentrating on the form. Not enabling it is when you can work with both. It’s also possible to separate the two in the workflow by first concentrating on the forms and proportions (sculpting, blocking) and then build an appropriate structure on top of those existing forms. Technique is called retopology.
Edit: Got carried away a bit again but hopefully there’s something that helps.