Briefly about Retopology
polycounter lvl 8
Hello PC!I have a very brief question about topology and workflow.
I watched some tutorials (Gnomon and DTuts etc.) and major concept about workflow was:
Base Mesh > Mid Detailed Mesh > Re-Topology > High Detailed Mesh
So question:
Why re-topology has to be done in the middle stage of modeling and why not after finishing whole model?
Is it because that very high detailed model is harder to re-topologize then in the middle stage?
Many thanks in advance for replies.
All the best to PC and all of you!
Replies
Dynamesh alleviates most of the need to do this, but it can only retain so much detail so don't go to far into detailing before you dynamesh.
Also with Dynamesh you lose your sub division levels thus using the ability to drop down levels and make broad changes using your starting topology.
In my point view (I should admit that now it’s mostly more theoretical knowledge then practical experience), so in my point of view, I need re-topology for one main reason: a) to get a Low-Poly Mesh; b) with proper edge flow, ready for rendering/animating/texturing;
Previously I thought, that it is better, if I will have Low Poly Mesh, which is made (re-topo) from my best possible model (highest level with all final touches and details).
And if I understand m4dcow correctly about Dynamesh, in this case workflow could look like:
Base Mesh > Dynamesh > Re-Topology > High Detailed Mesh
Just to insert my two cents regarding dynamesh: it's a basemesh creation tool that retopologizes a model. If you're working with an existing basemesh before using dynamesh, it's topology is of no relevance, only it's volume.
After dynameshing, you're left with a pretty consistant distribution of faces across a surface, so you shouldn't have to retopologize after running dynamesh: it will be perfectly fit for sculpting as-is. If you like having a simpler mesh to make larger changes to the form at once, you might find it beneficial to keep the dynamesh resolution towards a lower setting, and then turn it off and subdivide like normal. The only real need to retopologize the mesh afterwards is for creating the final low poly piece that everything gets baked to, or if you were just sculpting a rough form out and then building hard-surface pieces on top.
The most basic way to view the workflow is just do whatever you need to do to get your high-poly surface looking the way you want (which may or may not involve retopology along the way), then retopologize that for a more optimized/animation friendly edge flow.
I always thought it was:
1) low poly mesh (whether this is a simple block out, or using the zsphere in zbrush to create the required mesh)
2) Sculpt, sculpt, sculpt - anatomy; muscle detail etc..
3) Create a base clothes/armour & Import into zbrush to place on the mesh
4) sculpt the new clothes/armour models until finally happy with the model
5) retop
6) texture
Am I wrong?!
Is there some retop in the middle section too!?!
Please can some just post a step by step like i just did for noobs like myself!
Cheers
The thing to keep in mind is that it's art, and there are many ways to go about creating anything. Different projects may suit some approaches better than others, and different artists might prefer different methods depending on how they work.
Some people use a workflow like you described with no problem. Other's might want to use existing basemesh's with an existing edgeflow in order to ensure that all their characters have the same topology/uv layout/general scale.
As for "Is there some retop in the middle section too!?!"
And some artists just like to hammer out the initial shape first because they find it easier to sculpt something into existence rather than use traditional methods. Instead of worrying about box modeling a face and dealing with the shape and edgeflow at the same time, they might find it easier to just push and pull on a sphere until it resembles a head. The problem there (at least a few years ago) was that if you started modeling a head from a sphere, you'll get to a point where the polygon's just aren't working in your favor. They get stretched and squished, and you'd have to subdivide to get a better density in certain regions, but by doing that you'd add a few million unneeded verts to the back of the head or some place where they just simply aren't being put to good use. So you'd be better off quickly sculpting in the rough shape, and then retopologize it so that you can further sculpt it better (this approach can be see in http://eat3d.com/zbrush35_intro).
Lately its not so much of an issue as things like Dynamesh do it instantly at the click of a button. Every time someone remeshes the tool, they're retopologizing it.
Legend.
Thank you so very much for clarifying that!
Im currently building a base mesh at the moment; In fact I did post on the technical support forum asking whether my topology was ok as I had a few "poles" (on the hand, thigh and shoulder).
I didn't gain much insight from any replies (bar 1 of the 2 replies).
Could you quickly explain;
If I'm currently building a base mesh, with some detail to keep the characters form and then I plan on sculpting its definition in the muscles before modelling the base clothes/Armour and then re-topping.. does it really matter about the topology at the base mesh stage? I mean as long as its neat obviously but do I really have to go crazy?
I have since updated my mesh but here is my thread for a better understanding:
http://www.polycount.com/forum/showthread.php?t=102991
If you're dynameshing it at some point, it doesn't matter at all IMO since it will just be thrown out the window. On the other hand if you're going to subdivide it and sculpt on it as is, then you'll want to make sure the topology is mostly consistently placed quads. Sometime, anytime, just load up your mesh into a sculpting program, subdivide it a few times, and start doing some (quick) sculpts on it, solely to get a feel for how the brushes behave. You'll be left with a general feel for how its working out for you, and can make changes accordingly. You might decide you want more geometry in an area, you might think you have more than enough. Definitely sculpt around the poles to see how they behave. I'm not sure what program you're using to sculpt with, but I know with zbrush you might find that smoothing works better around poles if you release shift once you actually start to smooth.
Otherwise, it's just one of those things you need to experiment with to see how you like it. Speaking for myself, sometimes I like when basemeshes have edgeloops that flow with the muscles, and other times they just get in the way of sculpting an area the way I want to. My Lightbox is currently filled with different basemeshes (and even different variations of those meshes) depending on what I think might be best for what I'm doing.
Personally, you could probably go for a much simpler knee and get rid of the small edge loop that makes the cap. Leaving it as more of a basic cylinder should provide more than ample geometry. Simple base meshes can be pushed quite far (see http://www.pixologic.com/zclassroom/artistinaction/scotteaton/modeling/).
You..
Are a life saver, honestly i've waited for a reply like that for 2 days! I've actually saved out your quote on a .txt just for future reference in case I forget
Thank you