I noticed that by accesing directly the mesh value of geometry I can achieve things faster and easier. For example:
m=myMesh.mesh
for v in m.verts do
(
p=v.pos
...
)
In the for loop I have an easier acces to the verts as using getVert, and I don´t even have to care about node transformation. If tallking about transormation there some other very useful posibilities like this:
m=myMesh.mesh
move m [x, y, z]
rotate m rot
myMesh.mesh
This alows me to manipulate only mesh transformation without the one from the node.
Wondering if there is some other such really usefull things I can perfor on a trimesh? Unfortunatelly nothing from this stuff is mentioned in the documentation.
wondering if it´s posible to iterate the faces same way as verts?
Of course, you can do it like this:
m = $.mesh
for f in 1 to m.numFaces do
(
faceVerts = getFace m f
print faceVerts
)see this:
Sorry, what I was meaning is something like this
m = $.mesh
for f in m.faces do
(
...
)
so without using geter or seter fns. I know how to get faces with the getFace fn.
Okay, but why do you need that? As far as I know there’s no performance differences between for v in m.verts do and for vi in 1 to m.numVerts do. Both functions work with triMesh, so you don’t have to bother the node transformation.
Posibly I understood something wrong, but have heared somewhere that for v in m.verts do should be faster then for vi in 1 to m.numVerts do. Also a geter fn, as far I know, is a bit slower then direkt access, but my knowledge about that thing is very low, so can be complete wrong.
Whenever you hear such statements it is a good practice to try it by yourself.
Sometimes, performance varies depending on the situation, including OS, Max version, the whole code and the scene in question.
For instance, I could say that BitArrays are much faster than Arrays, and while that may be true in most situations, it can’t be taken as a general statement as in many cases Bitarrays will be much slower than Arrays.
In your case, if you take the time to write a little test script, you’ll find that what you heard don’t hold true in these examples.
(
delete objects
gc()
obj = converttomesh (geosphere radius:50 segs:256)
st=timestamp(); sh=heapfree
for j = 1 to obj.numverts collect getvert obj j
format "obj.numverts\n"
format "\ttime:% heap:%\n" (timestamp()-st) (sh-heapfree)
st=timestamp(); sh=heapfree
for j in obj.verts collect j.pos
format "obj.verts\n"
format "\ttime:% heap:%\n" (timestamp()-st) (sh-heapfree)
tmesh = snapshotasmesh obj
st=timestamp(); sh=heapfree
for j = 1 to tmesh.numverts collect getvert tmesh j
format "trimesh.numverts\n"
format "\ttime:% heap:%\n" (timestamp()-st) (sh-heapfree)
st=timestamp(); sh=heapfree
for j in tmesh.verts collect j.pos
format "trimesh.verts\n"
format "\ttime:% heap:%\n" (timestamp()-st) (sh-heapfree)
)obj.numverts
time:215 heap:25640504L
obj.verts
time:252 heap:64000680L
trimesh.numverts
time:129 heap:25600384L
trimesh.verts
time:148 heap:64009704L
So I was wrong, thanks for explanation.