Hdk Basics

[hoknamahn]

Hey, guys!

How to compute the point normals? I try to calculate the point normals thus

nml_index = holder_gdp->addPointAttrib("N", 

[George]

Try just calling "holder_gpd->normal();"

George.

[hoknamahn]

Thanks, George, it works.

But I can't understand why I get two different results in this case

holder_gdp->normal();
gdp->merge(*holder_gdp);

and

gdp->merge(*holder_gdp);
gdp->normal();

In the first case I give an incorrect result. But in second I see what I want to see.

Any ideas?

[George]

Well in the first case you're computing the normals on the geometry that doesn't include anything in gdp, so the normals in gdp will be set to a default (I think (0, 0, 0)) -- which will result is ugly blackness. In the second case you're computing the normals on the entire gdp.

George.

[hoknamahn]

No, in the first case holder_gdp contains the polygonal geometry and in the second case too (first and second holder_gdp are same). And all normals in the first case have values approximately {1, -1, 0.15}. Strange...

[hoknamahn]

I try to add "Cd" point attribute

illum_index = gdp->addPointAttrib("Cd",  sizeof(UT_Vector3), GB_ATTRIB_VECTOR, &result);

Shader sees this attribute, but I don't see it in viewport. How to visualise Cd in viewport?

[George]

illum_index = gdp->addPointAttrib("Cd",  sizeof(UT_Vector3), GB_ATTRIB_VECTOR, &result);

19313[/snapback]

Try changing GB_ATTRIB_VECTOR to GB_ATTRIB_FLOAT.

George.

[hoknamahn]

Thanks!

[hoknamahn]

I need to find a group of the polygons for which the current point is the common (or to find the neighbour points). Whether there is a ready function or it is necessary to create my own algorithm? The first that occurs - search of all polygons and check of concurrence of the points with current point. But it's a slow way.

And if to look at the structure of geo-detail it is possible to see, that between a point and a primitive there is a link (suspect, that is not bidirectional link), that is primitive contains indexes on points which enter into it. But not vice versa. Right? Whereas to be?

[sibarrick]

I need to find a group of the polygons for which the current point is the common (or to find the neighbour points). Whether there is a ready function or it is necessary to create my own algorithm? The first that occurs - search of all polygons and check of concurrence of the points with current point. But it's a slow way.

And if to look at the structure of geo-detail it is possible to see, that between a point and a primitive there is a link (suspect, that is not bidirectional link), that is primitive contains indexes on points which enter into it. But not vice versa. Right? Whereas to be?

19414[/snapback]

try

GU_PrimGroupClosure.h

and

GEO_Closure.h

[hoknamahn]

Thanks, Simon!

[sibarrick]

Let me know if you have any luck with them I can't get findPolysUsingEdge to return anything other than an empty list.

[George]

Let me know if you have any luck with them I can't get findPolysUsingEdge to return anything other than an empty list.

19435[/snapback]

That's odd. findPolysUsingEdge() needs a valid GB_PointRefArray (or a GEO_PointRefArray) in order to figure out the connections. (Of course the polys must be fused also). Are you specifying a particular group when building the point-ref array? or maybe not specifying a gdp?

George.

[sibarrick]

That's odd.  findPolysUsingEdge() needs a valid GB_PointRefArray (or a GEO_PointRefArray) in order to figure out the connections.  (Of course the polys must be fused also).  Are you specifying a particular group when building the point-ref array?  or maybe not specifying a gdp?

George.

19439[/snapback]

Ah maybe that's the problem I thought all you needed to supply was the two points pt0 and pt1 which represent the edge and then primlist got returned by the function. If prefarr was empty I thought it would consider everything. So are you saying the prefarr needs to be filled with a group of points to consider when looking for the polys, even if you are searrching the whole gdp? That was the next thing I was going to try.

I'm specifying a gdp when initialising my GEO_Closure object, I assume that's correct, since it worked for the edgeDelete method.

[hoknamahn]

Let me know if you have any luck with them I can't get findPolysUsingEdge to return anything other than an empty list.

I use findPolysUsingPoint() but I have a same problem.

Constructor of GB_PointRefArray expects the GEO_Detail and GB_Group.

So, what is it GB_Group? And how to initialize it?

My brain boils

[sibarrick]

I got this working in the end.

It's a bit of a hack at the moment as I'm still figuring it out, but maybe it will help a little

const GEO_Point            *pt0,*pt1;
const GB_PointGroup        *pointGroup = parsePointGroups("*",gdp);
GB_PointRefArray           prefarr(gdp,pointGroup);
UT_IntArray                primlist(2);
GEO_Closure                connectedPrims(*gdp);

connectedPrims.findPolysUsingEdge(*pt0, *pt1, primlist, prefarr);

obviously you need to set pt0 and pt1 but then primlist comes back with useful things in it.

Still looking at this though so I sure there are far better ways to do this.

[George]

That's looking great -- very close! The first thing is to get rid of the pointGroup, if you supply the gdp to the prefarr with a NULL for the GB_Group, then you get everything anyway. The reason GB_PointRefArray takes a GB_Group and not a GB_PointGroup is because it can accept either a primitive group or a point group. If one or the other is supplied, only those are considered in the ref array -- if NULL is supplied, then it uses the entire geometry. Building a point-ref array is pretty expensive, so the groups give you a way to only build for a subsection of the geometry that you're interested in. A lot of GB/GEO/GU methods behave in a similar way.

The other thing to note is that GEO_Closure gives you two options for findPolysUsingEdge(), one gives you back an integer list of prim ids, the other gives you a primitive group. If you don't need a group (to pass to other parts of the code), it's probably quicker to use an int array. Creating and destroying groups can be wasteful.

Hope that helps.

George.

[hoknamahn]

Thanks for explanation, guys!

Whether I can trust result which returns prim->calcArea() or it is better to write my own function?

[sibarrick]

That's looking great -- very close!  The first thing is to get rid of the pointGroup, if you supply the gdp to the prefarr with a NULL for the GB_Group, then you get everything anyway. 

19455[/snapback]

When you say a NULL do you mean, like this, where you don't specify it or that you give it a null pointer.

GB_PointRefArray           prefarr(gdp);

I tried it this way and it didn't work, which was why I ended up trying to create a point group, didn't try giving it a null pointer cos I thought that was *bad* . I think ultimately I'll need one anyway since the user will probably need to supply a group to work on.

[Mario Marengo]

When you say a NULL do you mean, like this, where you don't specify it or that you give it a null pointer.

GB_PointRefArray           prefarr(gdp);

As far as I know, a null pointer is simply the value 0 (zero). So I'm thinking he means something like this:

prefarr(gdp,0);