The background:
While helping PoisonCourtesan with a model problem of his I was asked:
And responded:
Spoiler Alert, click show to read:
Then Gigantus posted:
In response to that I am writing this mini tutorial.
Later on I was providing modelling trouble shooting for TheQ here.
In post#15 of that thread I posted something new that I learned since I started this tutorial.
So this is an update of this tutorial.
You are going to need msAlignNormals.rar.
Here are the steps:
A) Unhide all (under the edit tab),
B) Select all (under the edit tab),
C) Weld together (under the vertex tab),
D) Clean (under the tools tab),
E) Smooth all (under the face tab) and finally,
F) Align normals (under the vertex tab).
G) Save and exit.
A.)
The "Unhide all function" ensures that there are not going to be any hiden parts of the model that will not be affected by the following processes.
B.)
"Select all" is self explanatory.
C.)
The "Weld together" function is the reason for this latest update, you can thank Araval for his inquiry for it.
Here is the explanation for it's usefulness.
A cube (in it's simplest MS3D form) whose surfaces are welded together is made up of 12 triangles (2 per face) and eight vertices (one per corner).
If none of the triangles are welded together but still similarly positioned, you still get to see a cube of identical dimensions that is made up of the same triangles as before but it now has 36 vertices (three for every triangle).
If the GPU has to rotate that cube it will have to execute more than quadruple the calculations and that's just to reposition the vertices.
Sometimes, for the sake of easier UV mapping you make an object the "hard" way, that is you do not use primitives.
This increases the number of overlapping vertices.
The weld together function eliminates unneeded (for 3D representation) vertices but should always be followed by the "clean" function which adds the vertices that are necessary for proper display of the UV mapping in those cases where adjucent surfaces in a model are UV mapped on non adjucent areas of the texture map.
All in all, the weld together merges overlapping vertices into one, thus reducing their number but the clean adds only the vertices that are absolutely needed.
You understand that those two functions maximize their utility when used together, first the weld then the clean.
D.)
The "clean" function mostly eliminates faces and vertices that are redundant.
However, where it is solid gold, is in the cases that it adds vertices in order to "protect" the integrity of the UV mapping of 3D objects that are skinned on non-adjacent areas of the texture map.
For example, a shield is usually a solid object but it is very difficult to UV map it as such (nigh impossible I would say). Instead, we first select only the front triangles and UV map them as one object, then we select the back triangles and UV map them as another object and finally do the same with the side faces.
To make things more complicated for those (like me) who like composite UV mapping, say you want to make shields whose left half will depict heraldic beasts (say one with a lion rampant, one with a dragon and one with a griffin) and combine them with a variety of right halves that are painted in chequers, chevrons, barrys, you name it, so that with a limited area of an attachments texture map you can equip a model with a very large variety of shield (re)combinations. In such a case (but even in a much more moderate one) you would absolutely have to use the "clean" function in order not to end up with a totally messed up UV map.
PoisonCourtesan only had one object mapped in the attachments texture (the cleaver) but as you can see in the linked screenshot the UV map was messed up because the cleaver was one object (blade and handle) but in the attachments texture they were not one but two, the blade at the left of the handle. And so he got the problem he got.
The "clean" function would have "cleaned" that problem.
Just make sure you "Unhide All", then "select all" before you "Clean" so as not to leave parts of the model unaffected, you never know which parts of the model need that the most.
E.)
About the "Smooth All" function, I don't want to come out as a "know-it-all" dork, so I must say that I am not absolutely certain about this but I believe that it enables a built-in effect that allows for surfaces that are flat in MS3D (and mark this, all the triangles of a model are dead-flat in MS3D) to look like convex in-game. In any case my models look better with it than without.
F.)
The "Align Normals" function provides a partial remedy for the split mesh effect that is usually somewhat aggravated after the "Smooth All" function has been applied.
Model, right after conversion to MS3D from CA .mesh:
Same model, right after the "Smooth All" function has been applied, the split mesh line is clearly visible (and that's just one of them).
Same model, right after the "Align Normals" function has been applied, the split mesh line has clearly been smoothened:
Now, I suppose the more experienced modellers did not need this tutorial that much, but given the number of newcomers who after all these years keep coming to make models for M2TW - may the Good Lord keep them well for it! - and so many of them are posting questions about the same problems, I sure hope that this one is a useful addition to the noob modeller's arsenal of tutorials.






Reply With Quote











































