Hi, this is my first post on the forums...

As we all know the in the game industry time is money...so where do we stop optimizing??? take the image below



...we can either leave the model with a T-Junction or be neat and clean like the one to the right...
Now do we need to do that? or just leave the T-Junction as it as, as its a minor thing(assuming no one will see it in-game). Also should the models(inorganic, like buildings) be snapped to the grids...
i prefer to make clean model which are snapped to the grid...but there are few people at my work place who argue that we should not be bothered about shuch trivial issues...

so what is your opinion?

I always bother. To me, making clean models is not a trivial issue.

I don't know exactly what you mean by snapping to a grid. Being a slave to a grid isn't necessarily important, but making sure that buildings and props are touching the ground are.

Many times, you're not the only one who's ever going to touch a model. And if what you create isn't clean then it's going to cause problems for others down the road that have to work on what you created. Either that, or some QA tester to going to report a bug that you or someone else is going to have to fix. And it always takes more time to fix problems than to avoid them in the first place.

You should definitely avoid t-junctions like in the pic you posted: it will not shade as nicely when it's rendered on-screen in a lot of engines. Plus, I'd say it makes for an unwieldlier uv-map.

[ QUOTE ]
I always bother. To me, making clean models is not a trivial issue.

I don't know exactly what you mean by snapping to a grid. Being a slave to a grid isn't necessarily important, but making sure that buildings and props are touching the ground are.

Many times, you're not the only one who's ever going to touch a model. And if what you create isn't clean then it's going to cause problems for others down the road that have to work on what you created. Either that, or some QA tester to going to report a bug that you or someone else is going to have to fix. And it always takes more time to fix problems than to avoid them in the first place.

[/ QUOTE ]

What i mean by snap to grid is to have the model aligned in the orthographic viewports. The model on the left is properly aligned and all the vertices are snapped to the grid...while to model on the right is not



I insist on this as it is very frustrating for someone else to work on such models(like urs truely)...i understand that while doing organic modelling one need not and cannot use snapping, but in inorganic modelling why not use snapping as it makes the model nice and clean...

I've never used snapping when making inorganic objects... I just make sure all the points line up (not hard if you start right) ... failing that, Absolute transform type-ins get the verts in the right place. Then again, I don't like anything but a 0.5 degree angle snap [img]/images/graemlins/smile.gif[/img]

In the last example, it depends on the thing you are building. If your building was a 21st century concrete and steel building, then hell yeah, snap away. If it was a 3rd centrury cottage, then make it uneven.

_shank: I never use the grid, but if I wanted to something like in your example, I'd just planarize the selected elements.

Okay, I see what you mean. I do think it's important to make sure all the edges are straight and the verts are aligned, I just don't use snap to grid to do this.

But it also depends on what kind of game you're making. I would most likely keep everything lined up to a certain point, and then move verts around to achieve a more natural look (not having everything perfectly lined up and squared out).

But, to get back to your original point, people that don't take the time to make sure models are clean and neat are just being lazy and inconsiderate, IMO. It really doesn't take that long to implement good modeling practices.

For the allignment thing use the flatten x/y/z tools which Meshtools provides (hmmm....somehow I think that comment is redundant on this site [img]/images/graemlins/smile.gif[/img]).

[ QUOTE ]
For the allignment thing use the flatten x/y/z tools which Meshtools provides (hmmm....somehow I think that comment is redundant on this site [img]/images/graemlins/smile.gif[/img]).

[/ QUOTE ]

This is also built into Max 7 now.

If something needs to be straight, make sure that it is. I don't generally use snapping to grid, but rather do a Non-Uniform Scale type-in to scale them to 0 on the axis in question. It really is nicer to work with models that are properly constructed.

[ QUOTE ]
I always bother. To me, making clean models is not a trivial issue.

I don't know exactly what you mean by snapping to a grid. Being a slave to a grid isn't necessarily important, but making sure that buildings and props are touching the ground are.

Many times, you're not the only one who's ever going to touch a model. And if what you create isn't clean then it's going to cause problems for others down the road that have to work on what you created. Either that, or some QA tester to going to report a bug that you or someone else is going to have to fix. And it always takes more time to fix problems than to avoid them in the first place.

[/ QUOTE ]

I agree 100%. Being a slave to a grid, often gives you a "gravity doesn't matter" look to models. Rooves don't sag alittle on old buildings. Car tires don't squat out a bit where they touch the ground (like real life radials do).

Clean models are good, but you don't need a grid to do so.

I'm no 3D expert but its my impression that t-juntions such as the one in the image at the top is an industry no-no. It'll cause lighting errors and looks sloppy overall.