Yo everyone!
This is my first post so I want to say you hallo
I have some questions about next gen cars - like in GT5, NFS, DIRT etc. Do I need to make a Hi Poly model and after that model a low poly with normalmap from it? Yeah it's cool - new cool renders but... do I really have to? Yesterday I downloaded model of Pagani Zonda F from NFS Undercover to check how it's made and it doesn't have normal map, just diffuse textures.
http://speedy.sh/rqwDQ/Zonda-Test-2.rar
Here is "little game" made in Unity 3d and in my opinion this car looks just good.
So what I have to do? Start with low or hi poly?
Replies
Here's a recentish thread on the issue: http://www.polycount.com/forum/showthread.php?t=94200&
mLichy - maaaan thanks, now I know how to model. I must tell my friend that he was wrong
GTA IV uses same methods as Forza, but I think Saints Row does both methods. I'd say anything more realistic, just do higher poly with shaders, except interior, ect. But for something like the warthog, do HP and bake down. As always, you don't have to follow this, but it's more often how things go. Just depends on the style.
Doing a "mid-poly" model is less work (if you know how to model this sort of thing) and still looks better.
Today I have another problem.
The problem are these ugly black places. I know that I have to make more edges around the extruded region but is there something more to do?
Any idea how to solve it?
If you have a Box for example, and set it all to smoothing group 1, it has to try and normalize/wrap all the Normals around to an average, so you get darkening and artifacts. But sometimes this is done in a way, to help remove/fix normal map seams/issues.
http://matthewlichy.com/Personal_Belair_Coupe_57.html
Maybe my vehicle will give you some ideas. I'm not saying what I did is right/good, but it's what I did at the time. I'd prob. do some things different now, but I added more geo in places for proper deformation for GTA IV damage.
The best example I can find right now is David Finlay's work:
http://3dfinlay.com/html/veh/mcla.html
to summarize a few key things about the method:
-Like high-poly modeling, you chamfer/buffer your sharp edges with more edges, to correct the shading
-Unlike highpoly modeling, you can use tri's and quads all the same, it doesn't matter as long as the shading is good.
-Unlike highpoly modeling, you can't rely on the subdivision algorithm to round things for you, so you need to model them as round/curved as you want them. This is what makes the modeling so difficult, as you have to keep things smooth on a very dense mesh. Proper workflow is essential!