Hey Guys I am modelling the Arecibo, Puerto Rico Observatory. It is the area where I was born and has peaked my interest for a game. Now I have a few questions, I was wondering how to properly approach this and ask before I finish so I can make necessary corrections as I go. I am going to model everything. It is all quaded and have detached some parts so that I may keep low and clean topology. I was wondering if this would be game ready for the estimated polys or is it too much. Tower one will prob. be about 800, then duplicated twice. so were looking at maybe 2400 and thats excluding the dish and and the scope receiver and walkways. This will be a major role in game so Im estimating a total of 5000 polys for everything, is that too much? I was thinking about baking some wire into a plane to minimize but im trading quality for quantity, What would you guys suggest technical wise and some other approaches be better for final product? and is it looking okay so far?
Another is I want to minimize the use of textures in UDK. So I can UVW Unwrap some pieces but it would be very time consuming and may cause lots of visible seams. I can also UVW map some stuff but it would be added texture space in the content browser but it would look cleaner and easier to do. Please add suggestions and approaches. I appreciate you guys. Thanks!!
Replies
-target platform (Mobile, desktop)
-what is the genre? (A first person shooter will require a lot more detail compared to a RTS)
-what else will be visible on screen (some weapons, characters etc...)
Unless you're targeting mobile then you need to concern yourself less of tri count and more so of drawcalls. Hope that helps a bit!
Every little thing helps.
Well I have a 3rdPersonShooter Demo in Mind I want to do in UDK. (Deff Desktop) I want it to look really clean and up to current generation standards. So I wasn't to sure if I should go crazy with detail or simplify. On the level there will be alot of foliage plus character and weapon. Will have a few buildings but very minimal and will be targeting a bit of sci fi. So electrical high tech fence and props of that nature. Its secluded so mainly wilderness.
Tri counts are no longer an issue. Textures/lighting/materials will bottleneck far quicker than to many triangles on scene.
Well to be honest I've been focusing lately on quads and never really explored the world of tri's too much. I hear it looks cleaner and more efficient on quads especially if you are using smoothing groups but do you know if the engine when it automatically breaks will keep the estimated tri count the same as said in max or will I have to do it manually to get better results? Then again, a square is two tri's so that kind of answers my own questions. lol sorry.
Modeling in quads is really only relevelant if you are doing sub-div modeling or preparing meshes for Zbrush with no intention of using Dynamesh.
Another thing to bear in mind, that usually the tri count is not relevelant at all. Vertices count is what really matters.
Might be of some interest :
http://www.moddb.com/mods/goldeneye-source/images/cradle3#imagebox
Now you poly count is fine, level of detail seems decent for a fps view.
You might want to make a quick mockup of everything tho (blocking), before going into details, to check proportions, be sure everything work together..
You'll need tiling textures to keep a fairly high texture resolution. Everything that doesn't tile (details, whatever) could stand on the same map. For instance, that lower block could be on its own map, and the tower used a tiled map, or 2 blended together with vertex alpha...Lots of possibilities. You might want to take a look at udk tutorials all over the web, cause really it's a vast subject.
Yeah I saw a model ripped from halo 4 of master chief today pushing exactly 20k polys. I mean a scene in general including characters, assets, foliage, etc. what would the average poly count be in total be? if known..
Polycount isn't your biggest concern, draw calls are much bigger of a performance concern.