Will this be a problem ? (LP question)

MisterSande

Hey guys,

I have ve been wondering this for some days now:

Will these tiny spaces between my LP pieces become a problem ? (I am exploding the bake so it shouldn't be a problem but I am entirely sure).



I am curious if I should just try to "close" or / straighten these gaps in my HP/LP.

Visceral

Try it with a test bake and see what happends. I have never any of theese issues since i tend to close up my LP meshes, but not sure what result it will yeild.

Make a little test bake to see what happends, thats allways a good way to learn.

joeriv

Yes that will give problems, your cages are gonna intersect, and so stuff will bake onto eachother (or you have to make your cage extremely small).

So yes, exploding will be your only option.

Or the better one, don't model things like that in, it's so tiny, and so small that it could just be normal mapped in with some black Ao at the end.

AlecMoody

that will also alias like crazy. Is ther a reason why it isn't one contiguous mesh? Do the parts move part of an animation?

Mark Dygert

If this is the low poly you're going to have trouble. You could explode-o-bake like joeriv pointed out but that is normally more trouble then its worth.

Unless it animates like Alec pointed out, you're better off going with a contiguous mesh.

If this is your high poly, do whatever works best it won't really matter, the baking process just takes a picture of the surface like an orbiting satellite. The surface of the high poly can be broken up and fractured and it won't matter.

MisterSande

Or the better one, don't model things like that in, it's so tiny, and so small that it could just be normal mapped in with some black Ao at the end.

hmm that might be the best way indeed, but how would you get full control of the desired insets ? If I unwrap the model it is not that easy to tell where to paint in these insets.

I am doing this workflow because the tutorial I am following uses it (without real explanation, at least I know now why).

that will also alias like crazy. Is ther a reason why it isn't one contiguous mesh? Do the parts move part of an animation?

no, no animation whatsoever. I think you do not mean 1 actual mesh but rather several "big chunks" right ?

ill stick to the tutorial's exploded bake method until it becomes a little clearer.

EarthQuake

Doing an explode bake doesn't mean you need to use a large number of seperate mesh chunks, and using a continunouse mesh chunk for everything that doesn't animate(or just 2-3 larger chunks) doesn't mean you shouldn't do an explode bake.

My general rules of thumb:
1. Use as few mesh chunks as possible
2. Use separate mesh chunks for anything will animate, cause overlapping errors, or has some other special case where attaching it will be a pain
3. Explode all mesh chunks that will cause overlapping/intersecting raytrace errors when you bake

The area you show there, there's really no reason that should be a separate mesh chunk. You use more triangles in places you'll never see, and more uv space is wasted because you have to texture the little bits in-between the pieces, plus you will get more aliasing as Alec says.

Heres an example asset:
https://dl.dropbox.com/u/499159/3PointStudios_Brink_Maya_ar02_04.jpg
https://dl.dropbox.com/u/499159/3PointStudios_Brink_Maya_ar02_03.jpg

You can see that the majority of the gun is one large continuous mesh chunk(ie: all the parts are merged together without intersections). Exceptions:
1. The magazine, trigger, cocking handle and bolt(+ bolt flap), because they need to animate.
2. The iron sights because you can swap them in game
3. The rails on top of the rifle, client requirement.

So all these secondary elements were exploded to bake so they wouldn't cause intersecting issues, but for the most part the static pieces can be merged into one chunk which makes uvs and baking a lot easier, and gives you a more seamless end result without aliasing.

MisterSande

thanks Earthquake It seems very time consuming to make such detailed LP meshes but I understand why it is done.

I will give it a shot and post some results tomorrow for you guys to judge!

bblotus

This threads helping me dig through my HP to LP issues. Thanks

EarthQuake

MisterSande wrote: »

thanks Earthquake It seems very time consuming to make such detailed LP meshes but I understand why it is done.

I will give it a shot and post some results tomorrow for you guys to judge!

Once you get into the habit it gets easier and starts to make more sense. Its also generally easier/faster to uv a big continuios mesh than a lot of smaller little chunks, for me at-least, though uv distortion can sometimes be harder to deal with on more complex shapes.

Mark Dygert

EarthQuake wrote: »

Once you get into the habit it gets easier and starts to make more sense. Its also generally easier/faster to uv a big continuios mesh than a lot of smaller little chunks, for me at-least, though uv distortion can sometimes be harder to deal with on more complex shapes.

Once you encounter those problems, you learn how to side step them in the future, its all a learning process. EQ rocks so fuckin hard and knows so much because he has probably made (and fixed) more mistakes than anyone around, he just doesn't tell anyone about them :poly124:

You'll hit that point to if you stick with it.

MisterSande

thanks guys, with the help of Polycount members like yourself ive been able to learn allot faster

Here is where I am now:

I know it is not 1 contiguous mesh yet but I am working on it =P For me it is actually allot easier and faster to make seperate meshes for UVing. UVLayout is ultra fast when you have seperate meshes.

Few questions:
- Should I exclude the cylinders and small detail meshes from the "big" mesh ?
I presume it is better to detach identical detailed objects from the main mesh so that I can copy the UV's easier.

- My biggest worry of using these big LP meshes was that you did not have enough control on the texturing process. But I now think this is not true because you can just seperate the different type of elements how you would like when UVing. Is this correct ?

- In the tutorial I was following the guy only UVed/Baked out half of the asset. Is this indeed the best method or would you guys prefer it otherwise ?

Thanks again ^^

*EDIT* Should I attach the handle also ? Awaiting advice.

MisterSande

bumping for reply.

Mark Dygert

I lean toward merging them because that is how I would have originally modeled it, but you do have a pretty natural break there and since it has been modeled as two separate pieces and the topology doesn't really match, you might want to cut your losses leave it as separate and explode/bake.

Then moving forward keep in mind that it will be easier to model it from one solid mass, rather than join two meshes together later. The topology will flow much better, it will unwrap easier with less wasted space bla bla bla all the stuff you just figured out.

As for the "only bake half", I think the person had a lot of overlapping UV shells? That's pretty common for bits and pieces that will probably share the same texture. Especially if they won't ever be seen side by side, like two sides of a gun.

By stacking UV shells you can make the remaining shells bigger allowing them to use more pixels and have a better resolution. If everything is unique then all of the UV shells are small and they barely have enough pixels to define anything. Most newbies will then crank up the size of the texture maps to insanely high levels, which you can't do in actual games, so eventually they learn to stack shells and offset them when they bake.