The second step in learning C++ with Maya API

[stephen_weber]

I figured out how to create a hello world which operates fine. That must be step one. But I find the next step to be incredibly difficult because I believe it can be done but can't just do this.

In Mel I can (For example) create a sphere or primitive cube by a simple command "polySphere";

In C++ I know I can just use "polySphere" (a mel command) and get one. But there must be some basic API code underneath such a command to create a primitive.

TimerX gives for a large amount of spheres
1 second for mel
1.3 for python
1.5 for c++
(using "polysphere")

so I got as close as
MFnSphereData
but maybe I am on the wrong track, or I should just go back to the basics to setting up a c++ object (MOBject).

Can anyone point me in the right direction?

[rob-bateman]

But there must be some basic API code underneath such a command to create a primitive.

Nope. No such thing exists. Use mel, and have a read of
 this

[stephen_weber]

That's not fair

I copied your work earlier...
// we need to use a light function set to do anything useful
MFnPointLight fnLight;

// create a point light that has no transform
MObject oLight = fnLight.create(false);

(with the .h)

And wrote this.....

double rad =4;
MFnSphereData nSphere;

MObject oSphere= nSphere.create(rad);

(with it's .h)

Yours created a pointlight. Mine compiled but created nothing...

Does your link work in FireFox? I am using IE and it is a broken link.

[rob-bateman]

use the search box to look for.....   "New To The Api? Then Read This"  (with the quotes), and you should find the post (It used to be a forum sticky). Links don't seem to work in this forum anymore...

Yours created a pointlight. Mine compiled but created nothing...

Mine created a single pointLight node. 

You are wanting to create a polySphere generator node, a mesh node, a transform node, set up the connections between them, insert the final mesh instance into the relevent light linking set, and assign the polys to a shading engine for rendering. 

What you've actually done is create a pure mathematical description of a sphere. No mesh. No polyGenerator. No material assignments. No light linking. Just an equation..... 

You can create nodes in the API, and you can connect them all up as well, but it is an utterly retarded way of working. If you want to issue commands to Maya (via the API) in the same way you do in mel.

To do that, there is one way, and one way only: you must use mel script!

[stephen_weber]

Excuses excuses. There is nothing wrong with being developmentally disabled (society frowns on the term "retarded"), if the goal is to understand the pieces. I do promise not to take apart the Ferrari in the future. But I would like to understand the structure of a few basic commands.

I remember in 1982 drawing a point on the screen was magic. And I find all of the computer software system magic. Some of it is powerful magic like "polyCube". But even you have spent a few moments "wasting your time" with your 350 lines of code, connecting and checking all the pieces.

You wrote, "

You are wanting to create a polySphere generator node, a mesh node, a transform node, set up the connections between them, insert the final mesh instance into the relevent light linking set, and assign the polys to a shading engine for rendering. "

Thanks for this. I just wanted to recreate a single command. I just want to waste my first few steps doing something to understand what is going on under the hood.

Perhaps I can extend this lesson to making my own customize primitives. Like hexagon geometry.. hmm..

Thanks. I appreciate you keeping my focus on doing things smart.

The mantra: Use mel in C++ first. Use mel first. Use mel first.... Got it.

[joojaa]

Well you can do this. But ist just utter waste of time. You would indeed never be actually doing it by hand. EVEN if you did a hexagon geometry primitive.

If yu want to know what happens under the hood you go and make a sphere, a cube or so and choose

Window->Hypergraph: Connections

Which would basically tell you all you need to know. If you still dont have a clue you need to abandon c++ for a while and move over to learning how maya really works because essentially you can not survive without this deduction capability. in other words back to learning mel.

You can not learn the steps unless you understand what the steps do.

The commands themselves are very straughtfoward.

they just create a node, connect it up meaningfully to ther nodes and set values to the created node. Of fcourse you need to do this 3 times since you must also do the undo of the node and redo of the undo. The reason why you never ever do this is that number one rule of any programmer is to NEVER EVER UNDER ANY CIRCUMSTANCES rewrite something you don't need to. Especially if you don't understand how it works.

Basically doing it dont increase your understanding in a meaningful way.

You can not do what mel is for in the api, thats what mel is for. theer's no reason to make them overlap and there really is no 100% overlap. So some things can only be done in mel. Theres no porting mel to api.

PS: Rob fixed your original link, hope you dont mind.

[stephen_weber]

-So I want to create a Node that can hold geometry data. Maybe MFnMesh.
- Based on your example. I would like to perhaps take my mathematical sphere "data" and use the pointarray attribute to create a point array object which I can feed into the (fingers crossed) Mesh Object above. Then
- I see a MFnTransform class where I have to parent this "shape node". And/or use the addChild attribute.
I know in the safe world of these nodes would be automaticlly given default shading . Which you warned me wasn't the case.
And I am aware there is a createNode attribute...

I am of course , having never created a stable node which shows up in the outliner through these api, taking stabs at classes that claim they create nodes. Somehow they do. That is my positive statement. Somehow they do.

I hear your criticism of my thoughts and attempts. But I DON'T have access to the source code of Maya. And what burdens my thoughts is always going to be , what is the limitations of Maya. I program best when I have learned a language , so that I can see all the steps from the trillions of 0 s and 1 s electical impulses stored in silicon to typing a word.

Cause eventually in Maya 3421 theres going to be this button called "MOVIE" with a handfull of attributes. And I'll be hearing the mantra , just press that button and move on. It's already been written . There is no point in reinventing the film industry.... Gee.

opps sorry for the bad language..
Learning through repeating what has been done is a good thing. And I don't mind walking to school a few times to see the road slowly, even though I own a car.

[joojaa]

But I DON'T have access to the source code of Maya.

Neither do we.

You dont actually know what goes behind even the polySphere node. You can assume and defeer it.

I program best when I have learned a language , so that I can see all
the steps from the trillions
of 0 s and 1 s electical impulses stored
in silicon to typing a word.

Sorry practical progarmming does not work that way. All the kind of plugin writing you will ever going to do to any viable application commercial will be without didly squat of the code behind it. So you need to improve the deduction capability further. Much about programming is understanding black boxes you have no control over, otherwise you can not operate. Because you can NOT think in silicon anymore. Those comppanies that did are goners or make stuff for very integrated circuits.

Sorry.

Besides you dont really really want to read the sourcecode for any bigger application than notepad. I mean far simpler application than maya have 10,000,000 lines of code. You cant even begin to understand how much job it is to understand all that.

Cause eventually in Maya 3421 theres going to be this button called "MOVIE" with a handfull of attributes.

mayas going to die much much before that   Besides IF and when that button arrives (and it will), we need to rethink capitalism entirely.So you have bigger problems by then than just recreating the code.

We dont recite this mantra because its waste of time. But because ist waste of time AND its trivial. Incidentally we described each line of code you need to write just didnt write the code.

I suppose you just need about 10 years of practical programming understanding to begin to understand even the basic concepts for code.  (one of my favorite code bloggers put it something like this: When you begin to code your like a infant as you dont know any common sense about code. So you need to relearn the world, granted your probably twice as fast as a infant is at learning. Which might explain why people with 5 years worth of university studies in computer science behave like adolecents when it comes to code. After all you dont give productin tools to 10 year olds)

PS: both me and bateman are up for hire. The comppany i work for charges a comission at 2000 euros per day (with a minimum of 1 day). If your willing to pay ill happily write the code for you.

[stephen_weber]

I KNOW and thank you for the fact that you gave me all the steps. I even trust that fact. I suspose I picked such a topic so that I can have a known output to fight to get .
I have programmed on my own since 1982. I haven't programmed as a career. I have lots of languages that I have worked with. And c++ , I will admit isn't one that I have mastered. To me all that pointers and referencing boils back to peeking and poking addresses in an old Apple IIe. And I like that you want capitalism to ... evolve.

I suspose I could have gone to college in 1985 and studied computer science. But without the internet or hindsight , all society wanted hackers to become was cubical workers protecting and accounting for assets. (I know that is a simplification). I know exactly the type of person you are assuming I am. And I apologize for that .

I am currently working on the artistic side of Maya as a student. And I have modeled and animated and worked on animation, and texturing. I believe I understand Maya as the tool. I have now been absorbing Mel, and its more powerful allies Python and C++. But I also have to give thought to the possibilities of Renderman's shading language and Houdini's procedural programming setup. Not to mention HTML/CMS/ and all that. So If I appear slow on the uptake it is because I have only a bit of time to focus on this .. But it's been a slow day for me. I've only been called a genius once today by one of my instructors for a hotkey I wrote for him.

I don't have 2000 euros so I'll give you a gem. Use it wisely and it will take you far in the computer word. When the conversation boils down to people taking with letter abreviations Oh the RAM is this and the .. you know what I mean. Offer this one TLA....

TLA translates to Three Letter Acronym.

[joojaa]

Your slightly missing the point.

From maya programers perspective its a bit superfluous. You should concentrate on how the node networks work. The how and what gets called when something just works is uninteresting for you.

[stephen_weber]

So I almost have a simple plane created. It just isn't visible yet. I have it attached to the initial shading group so I don't think it is a light linking issue. Just not sure. As far as how node networks work,, hmm . I am concentrating on node networks that I have used regularly for a few years. And it is me learning the nuiances of this API.
So to interest you in my uninteresting work.. I am coding in python. (Badly of course, no undo or error checking).

import maya.OpenMaya as OM

dgModifier=OM.MDGModifier()
MESH=OM.MFnMesh()
FACECOUNTS=OM.MIntArray()
FACECONNECTS=OM.MIntArray()
FPA=OM.MFloatPointArray()
THISOBJECT=OM.MObject()
NV=4
NF=0
XPOINTS=[-1,1]
ZPOINTS=[-1,1]

for x in XPOINTS:
for z in ZPOINTS:
POINT=OM.MFloatPoint(x,0,z)
FPA.append(POINT)

TRANSFORM=MESH.create(NV,NF,FPA,FACECOUNTS,FACECONNECTS,THISOBJECT)
MESH.updateSurface()

dgModifier.renameNode(TRANSFORM,"MY_GEOMETRY")
dgModifier.doIt()

this_command ="sets -e -fe initialShadingGroup "
namepart=MESH.name()
this_command+=namepart
dgModifier.commandToExecute(this_command)
dgModifier.doIt()

DAGNODE=OM.MFnDagNode(TRANSFORM)

next_command="select "
next_command+=DAGNODE.name()
dgModifier.commandToExecute(next_command)
dgModifier.doIt()

The nodes appear and the manipulator is holding space...

I have some idea that this should work somehow soon, as I found a few code examples that were of this type of design. Such as polyPrimitiveCmd.cpp...
And I think I just need to append in the right INTARRAY values for FACECOUNTS, and FACECONNECTS

[in] polygonCounts array of vertex counts for each polygon. SO FACECOUNTS=[4]
[in] polygonConnects array of vertex connections for each polygon.
so FACECONNECTS=[0,1,2,3]
This gives me a parameter out of range which I don't follow yet.