 | | | Manhanon and plugins | Manhanon and plugins 2004-03-22 - By Richard R. "Remnar"
Back
First I want to thank those who've welcomed me back. It's good to be back
and I hope to work hard in contributing to the community here.
Well I have a story to tell here that will turn into a truespace project. I
have been sick with a severe viral infection for almost a week, and it seems
im not the only one. Its funny that while I am sick I get the strangest
ideas and dreams. Well this dream was about a silvery disk shaped city
really high up in the sky. It had 3 layers, each one having its own
function. The top layer was for dining, the middle for administration and
recreation and the bottom was for rooms. When i woke up I felt really
emotional about it, but I knew I would get over it. But I did spend some
time thinking about how it could be done. Maybe I have a knack for
mechanical engineering heh. Well anyway, this idea as vague as I put it
should be a great project to work on for a while. I want to get everything
down to the last detail I can add including some ideas that could possiblity
make such a rediculous idea work (and probably cost billions heh). Most of
my ideas will be like writing a fictional novel. Even though it will be
fiction I am sure it will get plenty of attention. I was looking for
inspiration for a truespace project and Ive found it, in a sci-fi dream
heheheh. Oh I almost forgot. I am calling this project Manhannon: Paradise
In The Heavens.
As for the plugins Im developing, I have completely redone them with great
improvements in memory management and stability. Some features that require
a beefy computer even work faster and without interface problems. I am still
looking for a few testers who have time at least once a week. I need one
tester who has a beefy computer (384 megs of ram, pentium IV etc) to test a
specific feature. I try to do a fraction, about 30% of the calculations on
this old clunker and it just cant handle it. I can do 20% and get some
decent results and I will post them soon. Oh the feature I am talking about
is from the Scatter plugin. its called Image Map Particles. So if you think
your computer can handle 100,000 geometry based particles or more, i need
you! Really this feature will make your machine crawl unless you got the ram
and processing power. One thing that comes to mind about Image Map
particles, is that it is useful even on a low percentage. Scatter is also
useful for light arrays, donut sprinkles, and obscurities. The operating
system that I know this plugin works on is windows 98 and XP pro. Anything
else is questionable. When i get all the base features of Scatter working to
every ones liking, I might add animation features and possibly other
requested features. As for the price of Scatter, I am not sure where to
begin. It is a rather complex plugin, but easy to learn and use. The
interface is similar but smaller than true particles. I also want to
implement trueparticles in scatter for the Ts 6.6 release. I have a working
release for 4.3 and 5.2. I am still hammering out some memory handling
routines for metaballs.
I had much frustration working with the metaball api in Ts 5.2. Those of you
who know the API (Application Programming (or is it Protocol?) Interface),
you pass a tsxMBALLOBJ variable to the tsxMetaballIsPrimitive function, it
supposed to return 2 things. 1: the tsxBool function result, and 2: the
tsxMBALLOBJ** parameter. Now if the variable being passed to the metaball
parameter is NULL, it will always remain null even if the function returns
TRUE. This was driving me nuts cuz it would not do this in Ts 4.3 API. So
after much trial and error, I found out that the variable needs to be
initialized with a block of memory. Usually this is automatic but for some
odd reason, the compiler automatically puts NULL into that variable. Now you
know why I was so frustrated. So I solved the problem by putting a single
block of memory of type tsxMBALLOBJ* into the variable like this
tsxMBALLOBJ* mball = (tsxMBALLOBJ*)tsxMalloc(sizeof(tsxMBALLOBJ*));
Now i thought i needed to free this if the function was true or not, however
that is not so (cuz it creates an access violation). So you need to test the
function. If its true, you dont need to free the varaible, otherwise you do
with tsxFree.
I also have solved the problem with vertex normals getting its Z position
offset in Ts 5.2 (dont know if its a problem in later versions). Well i
wrote a lot of matrix and vector functions. One function sets the T
(transform) column in a matrix. To fix this offset problem, you get the axes
position of the object and put it into the T column of the object matrix.
Heres a bit of code for you to see for yourself.
void mx_sett(CtsxTxmx3f* mx, CtsxVector3f t)
{
*mx.matrix[0][3] = t.x; // this can also be mx->matrix
*mx.matrix[1][3] = t.y;
*mx.matrix[2][3] = t.z;
}
void fSeeNvxsOffset(void)
{
tsxPOLYHEDRON* pPolyh = (tsxPOLYHEDRON*)tsxGetCurrentSelection();
if (!pPolyh) return;
CtsxTxmx3f mx;
tsxPolyhGetVxTxmx(pPolyh, &mx);
CtsxVector3f axesPos;
tsxGNodeGetAxesPosition((tsxGNODE*)pPolyh, &axesPos);
//mx_sett(&mx, axesPos); // remove the remark tag (//) to fix the offset
int nbrVxs = tsxPolyhGetNbrVxs(pPolyh);
CtsxVector3f* nvxs = tsxPolyhGetVxNormalsArray(pPolyh);
tsxGNODE** objs = (tsxGNODE**)tsxMalloc(nbrVxs * sizeof(tsxGNODE*));
for (int i = 0; i < nbrVxs; i++) {
CtsxVector3f v = nvxs[i];
tsxTransformVect3f(v, v, mx);
objs[i] = (tsxGNODE*)tsxCreateCube(1, 0.5f, 0.5f, 0.5f);
tsxSceneAddObject((tsxSOBJ*)objs[i], e_tsxTRUE);
tsxGNodeSetLocation(objs[i], &v);
}
for (int i = nbrVxs - 1; i >= 0; i--) {
tsxGroupAtCurrobj(objs[i]);
}
tsxFree(objs);
tsxSceneRefresh();
}
The correct result you should see (if you used a cube as the current object
for example) is the grouped objects within and centered of the cube or
whatever you used. You can see the offset (slightly below the center of the
current object) if you remark the mx_sett line.
Sorry for the long email, but I hope it was informative and helpful. Dont
know where else to post techy stuff. I dont own a current version of
truespace yet so I cant ask for technical help. I like the new Truespace
6.6. And btw, even though truespace maybe still lacking in some areas, its
the easiest application to use. Just try to paint a face in other
applications (much more difficult). Also truespace is like a clay modeler.
Your not limited to only 3 point faces like in cinema3d. Plus you dont have
to memorize 1000s of features to understand how to do the basics. And the
render speed is faster than most apps too (depending on what you use).
Cinema3d has terrible banding, 3dsmax looks cheap unless you get a high
priced render plugin. Sure I could tinker with 3dsmax and then tell Caligari
to put all those animation and some modeling features into truespace, but i
think they would really have to rewrite the whole program. An example of
something nice would be stack modeling. Cant be done even with a plugin.
They would have to rewrite the modeling engine of truepace to implement that
feature. A feature of stack modeling would be like you would perform a
boolean operation of object A to object B. Well you dont like where they
are, so you move them around. You dont work with 1 object, but 2 objecs
within the stack to move them around. I really dont see how this can be done
even in a plugin. Do you programmers got any ideas? Well a work around would
be to calculate the boolean during rendering. But how do we access the
mechanics of real time rendering? Ill check the api a bit further. Well if
we could move boolean stacks around we could do some nifty animations like
bomb fuses, etc.. Well I better stop before i write a book heh.
Richard R. "Remnar"
|
|
|