Hello. I have a question about the common topic of what a good polycount is. I am currently working on a model that I have bumped up from around 7,100 to 31,000 in ZBrush to use as my low poly instead. I have extruded a portion of the model, so now they don't show up so well compared to the higher versions. I have tried editing the topology in ZBrush, but I am still new to it, and am approaching a deadline for my portfolio.

So my question is to save time, if I decide to use the next lowest subdivision as my low poly instead, would 125k polys be waaaay overboard? (The model is a large monster) Normally, I usually stick with doing the best I can with low polys. But with hearing ridiculously high polys for next gen stuff, would it really be that bad?

My understanding is that 20-30k is alright for your main characters (kratos is 20k) Not sure what the "next-gen" will be pushing though. And I've been working on tablet stuff for a while, so my counts are a bit low these days :P But i think the next push for assets seems to be displacement maps. so you would still have the lower geo (~20k) and a new texture map would do that work.

Next gen characters might actually be lower poly with a bigger focus on topology for efficient, clean tessellation.

Originally Posted by JamesWild

Next gen characters might actually be lower poly with a bigger focus on topology for efficient, clean tessellation.

I would believe that, with some new tricks or more maps!

A good idea would be to post wireframes of the model and a nice render of it in-engine or at least a scree. People can figure out a lot about the topology and context of the character (and whether its features merit that 125K of polies).

I'm not a character artist, but 125K seems high (even if it were raw triangles count, which I don't know if you mean triangles or polygons).

I was wondering if we should have a wiki entry with that kind of information, because it keeps popping up. Have a list of games with polycount average for characters, props, weapons, vehicles, etc.

I don't mean to high jack the thread but I agree with what McGreed says.

As for DropOfTime, if you go looking in the Polycount Showcase threads you'll see they've put up their final tri counts so you can use those as benchmarks.

Originally Posted by DropsOfTime

would 125k polys be waaaay overboard? (The model is a large monster) Normally, I usually stick with doing the best I can with low polys. But with hearing ridiculously high polys for next gen stuff, would it really be that bad?

Yes, it would be terribly bad. It doesn't matter how powerful the hardware is. Your models should always we well optimized no matter what.

You're definitely way off the mark if you think next-gen consoles will let you use 10 times as many tris.

You essentially just want to be lazy and avoid having to make a dedicated lowpoly mesh, and you want us to approve of that?

No and no, and to echo what Perna said, 125K is no way feasible, even UDK has a hard limit of 60K vertices for a mesh, and that is still no excuse to even reach that number, unless you're talking about a fully kitted out character, which requires hyper detail for a cinematic in a game engine.

Simply put, you can import all the way upto a half a millions vertices in a game engine, but to what end, when that amount of extra 'data' can be used elsewhere?

Unless you want to work for cinematic's, in which case, you can run away with it, but even that is debatable since more times then not you have a LP model which has turbo smooth run on it and Disp. mapped.

Thank you very much guys. I am going to stick with 31k polys and use a displacement map (and even 31k seems high!) I truly appreciate everyone taking the time to help me out

Originally Posted by McGreed

I was wondering if we should have a wiki entry with that kind of information, because it keeps popping up. Have a list of games with polycount average for characters, props, weapons, vehicles, etc.

http://wiki.polycount.com/PolygonCou...riangle_Counts

Could also edit in some more info if needed.