Female (stylized) basemesh + joints
Hi Polycount!
I've been working on a female (stylized-realism) basemesh. Joints will be added for body and facial animation.
I hope I did well on edge loops etc. for animation - I'd like to ask you guys for honest critique. Please don't hold back
I'll be adding progress images along the way.
Thanks in advance!
I've been working on a female (stylized-realism) basemesh. Joints will be added for body and facial animation.
I hope I did well on edge loops etc. for animation - I'd like to ask you guys for honest critique. Please don't hold back
I'll be adding progress images along the way.
Thanks in advance!
Replies
topologywise looks good and clean. I don't think there would be any nasty deformation problem.
The one area that doesn't really convince me is the top of the hand: you have four unnecessary poles near the middle and ring fingers. There are a couple of ways to clean them up.
One way is as follow: when you have a rectangular patch and one side has more subdivisions than the opposite side, you simply make the side with more subdivisions converge towards the one with less subdivisions. This will always give you two poles as illustrated in the image below.
Anothe way is to obliterate all the topology surrounding the area that you're trying to clean up and try to move the problem closer to an already existing pole nearby that area. There's usually one way of merging the whole problem into that single pole. You then just need to relax the topology a bit.
If you plan on using this base mesh for sculpting, the least poles you have around, the better.
It would be also much better to avoid any pole with valence greater than 5. You have plenty of those between all the fingers. Dealing with those inside of ZBrush is usually quite painful!
Also I was wondering, is there a specific reason why you're using the T-pose instead of A-pose?
I updated the top hand topology to your 'one pole left' example.
About the poles/5-6 legged stars: I was aware and unhappy with those, but there are a couple of reasons.
- The basemesh is not meant for base for sculpting, but for Skinning and that I can copy the skin weights to all game characters I'll make.
- Since I'm developing for mobile applications, I have to keep the polycount as low as possible, I don't see a good way to fix these poles between the fingers. Ideas are very welcome.
To answer you last question:
Yes, I am starting with a base from Mixamo to have fast designs and clothing. Mixamo characters are in T-pose. Then I can use my basemesh Skin to copy the weights and skin to my custom skeleton. Plus, I am using Mocap data from the Rokoko Suit, which is also in T-pose.
Thanks again!
No problem. Topology is right in my confort zone.
If I got this right, this mesh itself is not ment to go in the game engine. If so, I wouldn't worry too much about it's polycount. You only have to worry about the polycount of whatever ends up inside the game engine.
Keep also in mind that "full quad topology" rules are only needed for SubDiv models. I don't think those would be of any help if you only plan on transferring skin weights from your base mesh.
Although full quad topology is always a good exercise!
1. You have a good point there, the mesh itself won't be used in game. I was planning to use it as a base mesh for the in game characters as well (like the hands are reusable), but I guess I should make two separate meshes from now on: one for skinning and one for in game with a lower polycount.
While typing this, a question arises: it does not add anything to the cheaper mesh to copy skin weights of the expensive mesh (right?), so why should I make an expensive mesh in the first place anyway?
The cheaper in-game mesh would have to deform nicely too, and there the polycount does matter..
2. Another thing I'm struggling with:
In my Facebook post there is a discussion going on about using tris or not for deformation. I'm really confused because some people swear by A and some by B (see image). Could anyone shed some light on this?
What I've tried now is this geometry, it deforms this way:
3. Yep for SubDiv I'm definitely practicing full quad - that makes me confused about what to do between the fingers, I feel like poles are unavoidable...
Thanks for your insights!
If those character are very different from one another, maybe having a slightly higher res mesh could help to get more accurate results while copying the skin weights. I'm not so sure about that though. Some experimentation will probably be needed.
2. In a realtime environment there simply no reason not to use triangles. If you're on a tight budget and you're only relying on linear skinning, which is very common in games, 'A' is a better choice since it gives better deformations by default while also being slightly cheaper. Your solution should work just fine.
Triangles are used all the time in order to lighten the geometry. As long as the topology doesn't get too messy, you can use as many triangles as you want.
I have never developed for mobile applications. Do you have a polycount budget for those characters?
3. Yes, poles are unavoidable in such area. Still, you can minimize the number and valence of them. Something like this should do it.
(about the poly budget: I'd say around 10k, but it depends on how many and how complex the characters are at the same time in the scene)
You guys are awesome, I went with all your feedback and the deformations work well in Unity
If your goal is around 10k triangles, this boundle of tutorials could be just for you. The 3rd part is all about optimization and is pretty cheap by itself. But I think the whole thing is a very good deal too. I learned a lot from it back in the days!
@jacopoS Woa I've had that tut saved for a long time, always wanted to follow it. Now I have even more reason (: