normal map baking: reason for exploding elements / matching material IDs?

pablohotsauce

Noob questions.

I watched that old Racer445 video on baking normal maps in Max. In it, he explodes (separates) various elements of his model before baking a normal map. I learned soon after that there's a much easier way to accomplish the same thing w/out exploding, by assigning matching material IDs to elements in your low-poly and hi-poly, and then checking on "Hit Only Matching Material ID" in Projection Options when baking.

Questions:

1. What's the point of baking these elements separately in the first place? What problem(s) does it solve? Is it something to do w/ getting better normals for nooks and crannies or something? Or is it sometimes hard to adjust the verts on a projection cage to fit around certain areas, or ... ?

2. How do you decide what parts of a given model to bake separately? I.e., how do you decide where to create material ID separations? (Maybe this is answered by the answer to #1?)

Thanks. Maybe I should experiment more w/ baking normals and figure it out for myself. :poly142:

SuperFranky

1. Elements baked separately to avoid artefacts on the bake. Try to bake two intersecting pieces and you'll see what I mean.

2. You bake them like that so pieces don't intersect or touch each other.

I'm simplifying here. Surely, somebody can explain better than I can.

D4V1DC

If you keep things together they will project onto each other which is not what you want.

In example, my most complicated model even though some might not think so anyway, was a "rock" based character.
I had to bake everything separately and so for example his fingers are separate objects and i had to bake the palms separate from the fingers, plus he has an inside and not just a front so it gets more complicated to explain, it was just a lot of parts.

2.
I'd say everything should be baked separately.

3.
Yea that is usually the way to learn once you run into some problems you'll get the idea better than we can explain it.

pablohotsauce

Thanks for the answers. OK, after playing around a little, I kinda get it. You get little artifacts when you don't explode / match matIDs. They're not terrible, but still.

After thinking about it a little more, I guess overall this is really more of a modeling question. It's a given you're going to bake out normals for an object's elements separately (to avoid artifacts), but when you're first making your models, how do you decide whether any given parts of an object are modeled as separate elements or as one big mesh? That's what I was really asking w/ question #2 in the first post. Is game modeling basically one big cheat-fest, where you just model tons of stuff as separate elements and thus don't have to worry about topology on a large scale? That's probably not true for organic stuff, but maybe it's easier to get away w/ on inorganic: props, enviros, hard surface, etc.

Kimon

There's actually lots of reasons why it's good to keep things as a continuous mesh!

- Easier to bake, and you get better result.
- Save texture space (those intersecting pieces adds up!)
- Possible higher vert count if you keep stuff seperated.

And probably a bunch more

ZippZopp

i tend to bake things as separate pieces personally. it does use up some extra UV space, but it is easier to control each individual bake.

also, when working in production, many assets need to be mobile and re-usable. working with characters specfically, it is great to have a library of various game ready assets to dress new character with.

sargentcrunch

n/a