Heya. Was hoping someone could tell me if I'm approaching low poly models in a suitable way.
I'm working on a gun at the moment. I'm doing the low poly for the grip and am wondering about how to correctly optimize. Take a look at the pic and tell me what you think.
The one on the left is 50 or so tris less than the right one, but it clearly doesn't look as clean. Does that matter? Its all tris and quads. the silhouette is the same for both.
Thanks.
Replies
Have a read of this: http://tech-artists.org/wiki/Beveling
So you need to minimise the number of verts, and carefully choose where your UV breaks will be, and what faces will be apart of what smoothing groups.
You could also collapse a few more edges.
Eh? Maybe I'm misunderstanding, but make an edge selection > hold shift down + right click > Soften/Harden Edge.
This way you can have a certain selection being soft edged and rest hard edged. Or the other way around, whatever you're looking for.
perna - Not sure what I meant by it not looking clean. I guess I'm just used to high poly modeling, where I stick to quads and try to keep everything nice and even spaced. I'm aiming for somewhere between 3.5-4k tris. You can see the actual gun here: http://www.polycount.com/forum/showthread.php?t=105856
At the moment I'm aiming for it to be a first person gun.
I didn't realize console specs were that high.
I'm glad I asked this question. Seems I knew nowhere near as much about optimization as I thought I did. I'll go through and try and do it better.
This is the whole low poly at the moment. it's only 2600 tris, so as you suggested I'll add much more.
I didn't think you had to have seamless models. I had planned to just bake down each part separately. Is having a seamless mesh just for convenience purposes? Intersecting geo always seemed easier to model and to keep the tris down.
I'll give it a try though.
I'm going to try to start merging parts where I can. I doubt I could get the whole thing merged, but reading through a few other threads on the issue, I'm starting to see the benefits of at least trying to get as much merged as I can, even if it does cost a few more polies.
Hell, even iPhone games try and put a majority of the costs on polies as opposed to textures.
So yep, listen to what Perna's says!
I'm glad this has all came up. Shows me how much I have to learn about actual modeling.
I'm going to go over the entire high poly and see what I can do about making as much of it seamless as possible. I have a feeling there's just some parts that will be beyond my skill level, but I'll post back when I've done what I can.
You don't need to make your highpoly watertight. You can do it with your low though, how you construct your high shouldn't have much to do with how you construct your low.
Generally I merge all of the lowpoly bits together as much as I can, except for pieces that would animate or make bakes more complex(complex overlaps etc).
But the high, model it in as many chunks as you need to model efficiently.
Try the same with intersecting highpoly and merged lowpoly.
We can get into the aesthetics of merging highpoly meshes, or moreso limiting nasty intersections, but there isn't any technical reason why your high would need to be merged.
I tried what you said, and did a quick test. The result isn't as nice without the merged high poly, but its better than what I thought it would be.
Yeah it confuses me sometimes too, its all part of Per's diabolical plan to confuse as many people as possible.
Pretty much.
In the future you can put more though into how parts are intersected. When you start to think about how objects are created in real life you realize its always one of a few things:
1. Two complex shapes will form a solid object. Here it is straight forward, you just want a soft intersection.
2. Two complex shapes, one is cut into the other. Here you can simply "imply" seam/shape lines on one of the objects(you don't have to really cut a big hole into it, a simple beveled inset is usually enough).
3. Two complex shapes, one is bolted to the other. Here the smaller shape will lay flush with the larger shape, and you'll have a bevel on that smaller shape where they butt together.
For original concepts you can vary this up quite a bit for effect.
If any area of a mesh is going to be seen up close, its worth it to think about how you handle the intersections. If its not going to really be seen in detail, you can skip and just do a simple intersection.
image to show what I mean:
And the third example, that's just having the two meshes just touching?
I think I'm starting to get all this. I really do just need to be a bit more observant when I'm modeling.
Yep, and the first is just a basic intersection, 2nd is a soft intersection.