3D Models and Terragen - Basics

Started by j meyer, July 03, 2015, 03:17:21 PM

Previous topic - Next topic

j meyer

These approaches are not really suited for close ups,but to see the differences
more clearly here are some close ups.

Hard edges
[attachimg=1]

Bevel and hard edges
[attachimg=2]

Soft/uncreased edges
[attachimg=3]

And with 'real' displacement in TG (Ray trace objects unchecked)
[attachimg=4]

Hard edges and 'real' displacement don't play together well.

If I'd want to apply 'real' displacement I'd use high poly meshes without any
creasing for best results.But since TG's ray tracer doesn't render the shadows
of the displaced object,only the shadows of the undisplaced object the
displacement should be kept rather moderate.

In case you want to know how to model the soft/uncreased edges cube I can
illustrate that,too.

Dune

Jochen: the last edge sample; are the extra 'edges' besides the angled corner polys needed for calculation of the angle softness, the corner normal? I mean, in the second sample, you have one bevel poly, in the third there are two, but also the adjacent polys, which are not angled but in line with the sides. I sometimes use the one bevel for a low poly distant 'soft edge', but the double edge for more close-up work, but not the adjacent poly's. Simply said, is that good or bad?

TheBadger

So what was that all about normals then?
It has been eaten.

j meyer

Ulco - the extra edge loops in the 3rd example are needed indeed as the faces
around the big center polygon all have a slight angle relative to that center poly.
Of course you could get away with one or two loops less,but the transition would
be harsher,more noticeable.
I'll prepare some more illustrations to show.


Michael - could you please be a bit more specific,what exactly you are refering to?

TheBadger

Quote from: TheBadger on July 12, 2015, 03:00:39 AM
So what was that all about normals then?

Quotecould you please be a bit more specific,what exactly you are refering to?

Well, if you take the model I sent you and you open it any modeler than it looks fine. If you send it from any other modeler to TG, it works fine. But from Maya it breaks. What was partly determined is that the normals had to be smoothed, not the geo. Some programs like Modo for one do this auto, but not maya.

This is in reference to the Architecture thread in image sharing. I think the next part of your thread here should cover normals, because that has turned out to be the confusing part for me.
It has been eaten.

paq

Hi Badger,

I did some really quick test using maya 2015 sp2, and I don't see really problem to export normals.
Have you tried to apply an 'average normal' command before exporting to avoid hard edges and thus displacement gap ?

Gameloft

TheBadger

Hi paq,

Can you give the steps to "apply normals", please. This is the part I can't find...

I found how to smooth normals, but not how to apply it, only several references that I should apply it but never any steps how to do it. Normals are for some reason, not much discussed when I search for some reason.
It has been eaten.

j meyer

#37
Michael,I'll better wait to see if you two can figure it out to your liking.
Maybe my view won't be needed anymore then.



Noticed a little mistake in my above posts on the 3 approaches.
In the third example (soft rounded edges) I didn't remove all edge loops
that are possible to delete without affecting the rounding.Forgot one pair
per axis.
Sorry,what can I say,I was young and needed the money....errh,
oops,wrong phrase,well then,it was late and I was tired. ;)
Or I'm just the usual scatterbrain.

Here is the correct wireframe and polycount
[attachimg=1]

The result when rendered in TG is visually the same as before.
Just the polycount is not 486 ,but now 294.

The announced illustrations will follow in the next posts.

bobbystahr

Quote from: j meyer on July 13, 2015, 03:09:09 PM

Sorry,what can I say,I was young and needed the money....errh,
oops,wrong phrase,well then,it was late and I was tired. ;)
Or I'm just the usual scatterbrain.

hee hee hee
something borrowed,
something Blue.
Ring out the Old.
Bring in the New
Bobby Stahr, Paracosmologist

paq

#39
[attachimg=1]
Some random geometry, with a mix of hard/soft edge.
If you apply a displacement on this kind of mesh, every hard edges will create a gap.

[attachimg=2]
You can average the normals in the 'normal' menu.

[attachimg=3]
Now the mesh should be gap proof ... the global shading might look a little bit weird too.

[attachimg=4]
These are the options for the .obj exporter ... seems to work fine ... but I never used maya myself.
Looks like a common song on every forum is 'first, delete you construction history' ... might help too :)

[attachimg=5]
This is mostly the kind of topology you need to control the overall shading without using edge crease (break normals) in your model.
It's in fact very close to subdivision modeling topology, where you try to keep tension by duplicating edges .

It's very time consuming, but very useful if you want to prevent displacement artifact ...

hope it helps.
Gameloft

TheBadger

It has been eaten.

j meyer

Cool! Thanks paq!

Now for some more pics and stuff.
You will find me using the term smooth subdivision here.The terminology can
vari from one modeling app to the other.That's why I want to show you the
difference between subdivision and smooth subdivision.

[attachimg=1]

Next are some steps to show how you can get the kind of topology on your
cube like shown earlier in this thread.Like on the corrected version of the soft
rounded cube to be precise.

[attachimg=2]

An edge loop is a closed ring of edges btw.
As you can see the added edge loops (2. + 3.in the image) prevent the cube
from that shrinking effect when smooth subdividing,shown in the first image
of this post.
Since I don't know the 'official' name for these additional loops I called them
stopping loops and will refer to them as such in this thread.
On complex shapes installing these loops can get tricky,but they are a good
way of keeping the edges of a model from being smoothed too much.

To be continued in the next post




bobbystahr

something borrowed,
something Blue.
Ring out the Old.
Bring in the New
Bobby Stahr, Paracosmologist

bobbystahr

keep em coming Jochen...and thanks
something borrowed,
something Blue.
Ring out the Old.
Bring in the New
Bobby Stahr, Paracosmologist

j meyer

Now that you have your 2x smooth subdivided cube you can start to reduce the
polycount,if needed/desired.
There is an area where the polys are perfectly even (or in other words where
adjacent faces have an angle of 0° or !80° from one to the other) on each
of the cube's sides.
They are all in the same plane,so you can delete all edge loops within these
areas without altering the result of the smoothing.

The areas
[attachimg=1]

and the reduced result
[attachimg=2]

You have seen the reduced result rendered in TG some posts above already.
But you'll see it again in my next post where I'll show some different versions
of the cube for you to compare.