Selecting objects inside of a sphere

[shcmack]

I wasn't sure whether to post this in the general forum or here as this isn't an actual programming question. But, since this have to be done with MEL/Python I thought I'd just post it here.

I want to make a tool/script that enables me to get all the objects which are inside another (specified) object.

So lets say I have a sphere that should be the "selector"-object, the script has to be pretty accurate to distinguish between the objects inside and outside of the sphere because of the roundness of the sphere. I guess it could be easier with a pCube, I don't know.  
Anyway, I've thought about querying the wordspace position of each vtx of the "selector"-object, and then somehow select all of the objects within the worldspace range of the vertexes.

Thing is, I'm not sure how to approach this task. So I was wondering if anyone had any ideas?

bendingiscool from cgcociety mentioned a way to do this with a cube, but that way won't work with e.g. a sphere (atleast I can't wrap my mind around it). The method he explained was to query each vtx of the cube and store the values, then query the position of each object, and if they were more than the cube's vtx min values and less than the max values, then select that object.

I can't say that I'm eager to get into the API stuff, I've had a quick look in the devkit/docs, and it seems pretty darn complicated

Or maybe a tool like this already exists?

[joojaa]

I want to make a tool/script that enables me to get all the objects which are inside another (specified) object.

sphere is easy, any object not so much harder just a LOT heavier. Heavier translates to slower,

For a sphere idea is the same as for cube (assuming that's your definition of where a object IS!). instead of the formula for cube cheeking  you just use mag(p2-p1)* < radius as selector condition. For any object its a bit heavier, one approach is to shoot a ray out of the location (any direction but orthogonal directions are easier to process), and if the ray hits a odd number of time your inside else your not, you need to check for glancing hits tough. Now ray-tracing IS a bit heavy so this is not going to be the fastest solution. Theres a few other tricks too but i dont think your prepared to do what it takes.

and it seems pretty darn complicated

Its not. Not really. Its just that it assumes you understand such things as programming architecture, node networks, how maya operates and what of the 3 things api exposes you to you will want to use. *!

in this case you want to make a either node** that outputs a selection or a command***. or just stay in mel/python.

*mag is a built in mel command essentially sqrt(x^2+y^2+z^2)

** recommended as:

*** making commands usualy signals your doing something worng, tough granted there are some uses. In this case its borderline.

*! assumption is that you have a master in computer science or comparable know how. Which trabsltes to 5-10 years of serious programming experience. Which basically means 10,000 ways which work but are very stupid and bad for your helath.

[shcmack]

Thanks for your reply Joojaa

For a sphere idea is the same as for cube (assuming that's your definition of where a object IS!). instead of the formula for cube cheeking  you just use mag(p2-p1)* < radius as selector condition.

I'm not sure if I understood this one, could you elaborate a little for me? p1/p2, is that two random vertex position of the selector object? My math skills is pretty poor, so my attempt to play around with it didn't help me much as I don't understand the logic

For any object its a bit heavier, one approach is to shoot a ray out of the location (any direction but orthogonal directions are easier to process), and if the ray hits a odd number of time your inside else your not, you need to check for glancing hits tough. Now ray-tracing IS a bit heavy so this is not going to be the fastest solution. Theres a few other tricks too but i dont think your prepared to do what it takes.

Woha! I think I understood what you meant, but is this even possible with MEL? You may be right when you don't think I got what it takes, atleast if your other tricks are more advanced, heheh

Its just that it assumes you understand such things as programming architecture, node networks, how maya operates and what of the 3 things api exposes you to you will want to use. *!

I guess everything's easy when you know how to do it I think I'll stick with MEL a while before I try to grasp over C++ and the Maya API.

*! assumption is that you have a master in computer science or comparable know how. Which trabsltes to 5-10 years of serious programming experience. Which basically means 10,000 ways which work but are very stupid and bad for your helath.

Uhm yeah.. I guess my carpenter-skills aren't of much use here

[joojaa]

I'm not sure if I understood this one, could you elaborate a little for me? p1/p2

Oh ok, i tought that would have triggered something in you. I mean this is the kind fo stuff you were taught in later clases of primary school or early/junior college and again at every turn after that.

p1 is point one

p2 is point 2

p2 - p1 = difference or vector distance between 2 points (this si the convention see your old math textbooks)

mag(vector) = distance as a scalar (mag is just trigonometry)

< radius = boolean operation signals true if inside sphere false if not.

ergo

p1 is center of sphere

p2 whatever vertex you are testing for

PS. just the kind of stuff youd expect carpenters to do daily. Ever used a tape measure, thats what you did.

but is this even possible with MEL?

yes but you need to proxythe data trough a node. Slow as hell tough

[shcmack]

Hehe, ok now I feel a little stupid I get it, that was actually a really smart way to solve the problem! Thanks alot!

[joojaa]

Dont feel, at least that wasnt my intention. I just wanted to underline the fact thet you were actually taughtvery wide mathematical skills in school. Its just that they were used in reverse, you were the computer. Now you need to form the equations yourself and that's the second step. Not all will make this step,

But see now you have the weapons to start looking back at those things you were taught and understand them better. Use a slow evening flipping trough your old books if you have them. Its amazing how much you can get out of just 3 hours of retrospection.

[holmesanimation]

Hey guys, I was just curious if you were able to figure out how to do this? Did you manage to build a working script?