Model Topology/Method Question

cbates540

Hey all, I've been working on this mirror frame model for a project, and I wanted to hear some opinions on if my topology was really the best way to go, or if there is an easier or more efficient way to go about it.
 
This is what I've got.  I extruded from curves for the top, snapped verts to a curve guide at the bottom, and then added edge loops to the sides and snapped those verts to another curve to get the side.  It was a nightmare of quads, so I condensed most of the edges into single specific ones.  I know there had to be a better way to go about it (and if you have advice on that, I'm all ears), but is the resulting topology necessarily good for a static mesh in a game?  I'm using Blender and the final result is 1942 quads.

Obscura

It doesn't necessarily need to be one continuous mesh. It can make things easier sometimes to split it into logical pieces. You could also weld a lot of verts into corners.

cbates540

I do know that things don't always have to be one continuous piece, but where would I be better off splitting this model up? Or making it separate if I were to do it again? And I guess with welding more verts into corners, just do what I did at several points around the edge, but push it further?

musashidan

One way of splitting a model up is by logical real-world counterparts. This is pretty much a basic tenant to follow when modeling. So the frame is made from wood? Split it to elements based on that: the main block/the acorn finial/the moulding/etc.

There are still loads of tris you can optimise. The inset region can be 2 triangles when spllt to an element. Once you split those elements off there will be a lot more opportunity for optimising.

heyeye

Good rule of thumb is if an edge loop doesn't effect the silhouette and isn't being used for UV seams(mirroring,trimsheets, etc) then collapse it down. 

JayXiong

I think what you're doing is fine for the most part. What's this model for? Knowing that will help answer your question better. These are some things I would personally do (see image). You could remove a lot of edges to create quads (as seen in green).  You could also condense the parts highlighted in yellow to avoid a lot of the issues in green. The top part of the model is already low-res, so there's no need to make those lower parts (in yellow) that detailed.  One more thing:  I don't know if this bothers anyone else, but when I model, I try to avoid vertices with 6 or more edges branching from it.  Sometimes I'll allow a 6-sided vert but only as a last resort.  This could just be a personal preference, but in my experience, 6-sided+ verts can cause awful artifacts when subdividing.  I hope this helps.

[Deleted User]

The user and all related content has been deleted.

cbates540

Hey guys! Thank you all so much for your help so far! I went back and edited things to try and make it better! I split the inset part into its own piece and collapsed a lot of the edge loops down, tell me what you think! Could it still be a lot better, or would you call this a successful static mesh model for a game? Thank you all again! 

musashidan said:

One way of splitting a model up is by logical real-world counterparts. This is pretty much a basic tenant to follow when modeling. So the frame is made from wood? Split it to elements based on that: the main block/the acorn finial/the moulding/etc.

Yes, I generally try to split models into pieces as you describe.  The moulding and acorn are already separate, I think the part that got me was that I was told this was to be a mirror frame without the mirror, so I think it slipped my mind to split the center part out anyway even though its part of the main block.  It helped a lot to do that though!

JayXiong said:

I don't know if this bothers anyone else, but when I model, I try to avoid vertices with 6 or more edges branching from it.  Sometimes I'll allow a 6-sided vert but only as a last resort.  This could just be a personal preference, but in my experience, 6-sided+ verts can cause awful artifacts when subdividing.  I hope this helps.

Your advice helped a lot! Thank you for posting the image! I also try to avoid 6+ edges at one vert, and definitely when something needs to be sculpted later.  This model in particular won't be getting subdivided though, which is why I'm testing out collapsing edges together.  I actually made this post because the amount of edges at one vertex is freaking me out! I had to know if it would bite me later somehow!

musashidan

Don't worry about 6+ valence poles, that's as Teri says, related to sub-d surfaces. Unless it results in multiple very long, thin  tris which can cause baking artifacts.

This is much better and perfectly fine for a game asset.