Select Faces Inside An Object

[pete1973]

hy,

got a problem which i don't know how to tackle exactly.
i got geometry which got surfaces inside the visible geometry, which i need to separate.
e.g. two spheres a bigger one with a small one inside, which are combined.
is there a mel command to query if a face is visible to the camera or not?
i thought about measuring the distances of the faces to the camera and do it that way and
how can i query a faces position

urgent ! help please

[joojaa]

QUOTE(pete1973 @ 02/11/08, 12:48 PM) [snapback]280903[/snapback]
is there a mel command to query if a face is visible to the camera or not?

not in default maya. In fact the computer doesn't know this.

[pete1973]

hy,

i thought it wouldn't be easy. what would you suggest to tackle that problem?
i know scanline does that in fact. is it possible to use a d-map for it?

[joojaa]

QUOTE(pete1973 @ 02/11/08, 06:14 PM) [snapback]280917[/snapback]
i thought it wouldn't be easy. what would you suggest to tackle that problem?
i know scanline does that in fact. is it possible to use a d-map for it?

Well in actuality the scanline doesnt know this (its a bit surprising but true), it knows that any given pixel isnt visible after its darwn ALL traingles, but it cannot know if its just the inaccuracy of the system. What it does it just paints each and every face and looks if each raster is behind or not. Truly when it ends up it doesnt know one way or other.

But yes you can do it with scanline, what you do is tag a color for each face and sum up wich tags yu saw the ones you didnt must be hidden. Sample sufficently much and thats it. However i wouldn do this with mel and render way too slow.

What you CAN do however is bake occlusion into per vertex colors. The ones that are black are on the inside rest are on the out. Then expand the shell and delete ones that are predominately black if your interested in whole shells.

(the occlusion works its a bit fo a sledgehammer in speed but does the job)

PS: still neither method knows for sure, what they know is that the samples they've taken are in or out. (analyzing this is a bit of a pain to do and much slower than raytracing)

[pete1973]

hello jooja,

thanks for your suggestions.
i think i try the occlusion-solution, which sounds most practical for me.
one more question though. the paint select-tool seems to be able to distinguish between
seen and unseen faces in shaded mode. do you think it would be possible to utilize that
aproach?

[joojaa]

Yes, what it does is it works on the end result of rasterisation like the tag thing (only it renders pointers to faces) but the user would need to paint around the object. Problem is again that yu have no mel access to this. Thats teh way i would do it if youd be asking how t do it in api. But i dont want to make api codes for others, too much mental pain about thatkind of things.

[pete1973]

right,

not to worry anyway, did it the way you suggested and it works like a charm.
you are THE man!

cheers!

[pete1973]

back again;)
got one little problem. i need to change to vertex-selection-mode in order to convert the vertex selection into
faceselection, otherwise it will select all faces when converting.
i could not find a mel command for that so i got this one from the history:
doMenuComponentSelection($sel[0], "vertex");
works ok when i changed the mode manually once otherwise maya will say it cannot find the procedure.

[joojaa]

weell you can look into the procedure definition (use whatIs whateveryou dont know what is) but the command is actually not there its in PolySelectConvert, anyway its :

select polyListComponentConversion -tf;

Theres also a second way that uses a bitwise operation mask and works directly on seletion but list is even better

[pete1973]

hy,

sorry i can't get it to work. if i execute it, it selects all objects faces, but does not actually change into
the component selection mode.
maybe i do it the wrong way.

string $Vertexname = $sel[0] + ".vtx[" + $i + "]";
select -add $Vertexname;

doMenuComponentSelection($sel[0], "vertex");
PolySelectConvert 1; // Face

that will do the trick when i change the mode once manually.

another problem is that i can't get the convertLightmap command to work properly.
i can't tell it to bake occlusion, wtf!

[joojaa]

umm commands work on selection lists

so doing:

CODE
select polyListComponentConversion -tf $yourVertList;

woudl operate on just that data, likewise if you just wanted the oobject list you'd do

CODE
$facelist=polyListComponentConversion -tf $yourVertList;

But thats basic mel syntax logic which i assumed you knew. Works on pretty much any mel command.

So you could do something like aswell as select:

sets -n inner polyListComponentConversion -tf $yourVertList;

And if you dont specify a data it works on main selection