Hello.
I was wondering which piece has the correct topology flow. When building environments i usually run into this issue
A: Has little bit more tris that are wasted from a silouette standpoint. Has better envenly space vertex which will give it better vertex lighting then B. However I don't think it would matter if it was normal mapped since normal map lights everything on a per pixel level. Am I correct about this?
B: Has less tris more efficient use of polys. However more points are jointed together which might lead to lighting issues?? Or is this not a big deal for nex generation stuff since normal mapping and light mapping pretty much makes vertex lighting obsolete?
Which topology flow is better to use for next gen stuff?
Replies
However, you see much worse topology of non-organic models in shipped games quite often. The main thing is that it doesn't cause any messy shading.
If it was all one smothing group B would ok, but as I am finding out with vertex light baking, A would be the way to go.
Not to mention you'd probably break up that mesh differently based on how you would have to texture it anyway so it would look more like A.
mmm?
http://boards.polycount.net/showthread.php?t=50588
This is more in line with what I had in mind.
Saving texture space is key, then comes saving polys but whats more important is building modules that work together that can be rearranged to create new and interesting things. I would actually increase the poly count and give more detail to this, but the textures and the over all module shape wouldn't change.
But the way he's broken it down seems ok.
I might have even just made one generic arch model, and used it as a facade over the other pieces.
Check out the current Environment Modular challenge over in the Contests forum here, there are several threads of tips and tricks for this kind of work.
Another super modular approach could be:
note how some of the segments are rotated to prevent repeating. You can also just work with more different segments of course if you can afford the texture space
Each side would have its own unique UV space so the flipping wouldn't cause a seam. It would solve the need for a second UV channel in which all pieces have their own space, (already taken care of) for light and normal maps.
At that point you might as well create the whole arch, but I would break it in half because you can do more with half arches then whole. Depending on how you laid out the UV's it would be pretty easy to paint across the seams and deal with any tiling issues.
The trim would also have 4 unique sides so it can be rotated. I would also make sure that the pieces where square, so I could rotate the trim and use it as vertical trim as well. Possibly using 2 opposite sides for vert trim and the other 2 opposite sides for horiz trim.
I might even create a few boxes of different lengths to go in the key stone spot between the arches, so I could change up the width of the arches.
I'm not saying MoP is wrong because there are half a dozen ways to tackle this, and I can already see 3 other ways but they depend on the engine and the purpose of the pieces. MoP would make the right call based on what is needed for the given project just like any good artist would.
Thanks for the help and taking the time to draw out examples for me. This has definetly put things in better perspective for me. Ya there is a dozen of ways to tackle this so i'll try my best to make good judgement. I'll keep in mind texture space is most important. Modulate it as best as I can without sacraficing wierd shading or seams. I can probably even use decals to hide seams