PDA

View Full Version : Direction of polygon normal.

PICTRIX
09-15-2007, 08:30 PM
As for LW9.3 modelers, the computational method of the polygon normal seems to have been changed.

It is a thing that understands from the normal guideline display.

Did the normal come to be decided in clockwise order of vertex?

This is a very joyous thing.

However, why that it is not reflected in the preview and rendering.

Need not the direction of the normal and FACE be corresponding?

50403

meshpig
09-16-2007, 01:32 AM
Pictrix

?????????????????????????????????
:) M

meshpig
09-16-2007, 02:01 AM
PT ...sorry,

-Couldn't find the Japanese character set in time for the 5 minute edit!!

Not sure what you mean...

PICTRIX
09-16-2007, 03:18 AM
Not sure what you mean...

Hi meshpig ,

My intention?
The polygon normal is suitable for you.
However, you cannot see the polygon face.
Do you think this to be normal?

dpont
09-16-2007, 03:59 AM
As for LW9.3 modelers, the computational method of the polygon normal seems to have been changed.

I don't think so,
in 9.3 polynormal is still calculated with
first, second and last point of polygon,
and I think that in your poly the first
point is the displaced one, so if you
create a triangle with it as first point
and second + last (clockwise) you
will see that this triangle is flipped.

This doesn't occur if you choose
another point (not displaced) for

Denis.

PICTRIX
09-16-2007, 04:36 AM
Is my LW abnormal?

http://www.pictrix.jp/normal.mov

What do you think?

I certainly saw UFO !!:D
Besides this, there is evidence !!

What do you think that it becomes it when the polygon of the movie is done in the bevel?

MooseDog
09-16-2007, 07:55 AM
Is my LW abnormal?

confirmed. simply repeated the steps in your demo and got the same results.

the statistics panel does not show any non-planar ploys either.

at this point i'll leave things to you guys who understand what's going on under the hood. good luck:thumbsup:

dpont
09-16-2007, 08:16 AM
Not strange and not new,
unplanar polygon is not the unique situation
for triangulating quad or ngons, if you repeat this demo but starting by the upper point this doesn't happend.

A polynormal is calculated with only three points.

Denis.

PICTRIX
09-16-2007, 09:01 AM
confirmed. simply repeated the steps in your demo and got the same results.

the statistics panel does not show any non-planar ploys either.

at this point i'll leave things to you guys who understand what's going on under the hood. good luck:thumbsup:
Hi MooseDog ,
Thank you for confirming it.

Hi dpont ,
Okay....
Normality is calculated with the 1st, the 2nd, and the last vertex.
I do not care still.
It is a guess that I wrote.

However, the thing with different result is important.

Please go as shown in my movie.
Preview and a normal guide turn to opposite as the face is the same.
The problem is not the direction of the face.
It is a direction of a normal guide.
It is new at all.

PICTRIX
09-16-2007, 09:19 AM
Not strange and not new,
unplanar polygon is not the unique situation
for triangulating quad or ngons, if you repeat this demo but starting by the upper point this doesn't happend.

A polynormal is calculated with only three points.

Denis.
Vertex that I moved in the movie is the first point.
You should have seen the face turn back by the moved thing.
The face might be calculated by three points.
As well as before.
However, a normal guide might have seen the thing kept for the same.
A normal guide is corresponding in the direction acquired from mesh edit.
Here is a problem.

dpont
09-16-2007, 09:49 AM
You have right, polynormal guide is wrong
in preview for this poly in 9.3.
You should report it as a bug
LW-Bugs (at) newtek.com

Denis.

PICTRIX
09-16-2007, 10:10 AM
I think that this normal guide is correct though I do not know a normal
computational method.
It is a thing that the consistency of a clockwise policy of vertex is given.

The thing that others are left is regrettable.

Silkrooster
09-16-2007, 05:03 PM
I don't think the normal guides are correct as it looked like you created both polys in a clockwise direction and each poly pointed in opposite directions.
Silk

PICTRIX
09-16-2007, 06:40 PM
I don't think the normal guides are correct as it looked like you created both polys in a clockwise direction and each poly pointed in opposite directions.
Silk
Hi Silkrooster ,
I am intentionally making the first vertex concave.
Moreover, it is important that there is a possibility of becoming concave without intending it, too.

http://www.pictrix.jp/normal-2.mov

I think that the display of a normal guide is not a bug.
It is thought that it is a new specification. I want to think so.

The polygon made clockwise shows the same normality regardless of the position of the first vertex.

I want you to apply this new specification to the preview and the rendering of the face.

GregMalick
09-16-2007, 09:56 PM
How can a surface display opposite the direction of its normal? :stumped:
Looks like a strange bug in Modeler's OGL display.

PICTRIX
09-17-2007, 07:25 AM
How can a surface display opposite the direction of its normal? :stumped:
Looks like a strange bug in Modeler's OGL display.
Aloha Greg ,

Welcome to AREA 93. heheh

You also think that the display of a normal guide is an error.

The thing with a wrong the idea is proven.
The movie of the anatomy of the alien is here. :D

Concave make the polygon with LW93 the first vertex.
To externals of normal.mov that I presented.
Select the polygon.
A normal guide has expanded in the direction opposite to surface.
When following LScript is executed, evidence without this guide because of the error is understood.
The content of LScript acquires the polygon normal, and moves the polygon in the direction.
You will understand. :)

In which direction did the polygon move?
Is it a direction of the guide display?
Is it a direction of seen surface?

And, you must think what happened.

// normal move
//
//

@version 2.4
@warnings
@script modeler

main
{
selmode(USER);
editbegin();

wk= polycount();
polcnt= wk[ 1 ];

for( i = 1; i <= polcnt; i++ ){

pol = polygons[ i ];
polinfo = polyinfo( pol );
nml = polynormal( pol );

for ( j = 2; j <= polypointcount( pol ) + 1; j++ ){
pnt = polinfo[ j ];
pos = pointinfo( pnt );
v = nml + pos;
pointmove( pnt, v );
}
}
editend();
}

GregMalick
09-17-2007, 09:25 AM
I don't think anything is wrong in your code.

My guess is that the OGL code in Modeler is wrong.
Notice that the normal points up when you select it.
Try saving the object and rendering it in Layout.
I think it will render fine.
I think FPrime will display and render it properly too.

There is a good chance that Layout will even display it properly, that Layout OGL code may not have the same bug. But that is only chance.

BTW, another bugs are:
1. the normal isn't displayed for all the polys. Try selecting them one at a time.
2. the normals are being displayed in odd places.

Oh.. save the object before any flipping and bring it into v9.2
The normal is pointing downward.

so I take back what I said. It's not OGL.
LW thinks the normal is downward.

Also, you can make the upper left poly point downward by pushing the "stretch" too far - beyond the edge.

So now I think the stretch code is messed up and some test for moving beyond the right edge is in error.

That's enough analysis for now....

GregMalick
09-17-2007, 09:54 AM
BTW, this bug exists back to LW7.5

PICTRIX
09-17-2007, 05:49 PM
Did the polygon that the first dents move in the direction?

Okay.

It is the following evidence.
The 1st, the 2nd, and the last vertex are and there is a straight line polygon.

The polygon normal had not been obtained before 9.2.

In 9.3, the polygon normal guide is displayed.
The polygon status displays the error.
However, the calculated direction of the normal of the polygon is obtained.
If above-mentioned LSCript is started, it moves in the direction.

What I want to say

The modeler renewed the processing of a normal calculation.
The guide display of the shown polygon normal is not a bug.
It is a result of a new calculation.
All normals are calculated clockwise of the polygon.

However, even neither the display, rendering nor the status of the face do a new calculation method.

I want you to go early in a new computational method by all processing though it will not be a bug.

It is suitable for the same direction even if the first vertex is at an
arbitrary position.

It will be easy to do modeling.

Even rendering can be done.

PICTRIX
09-18-2007, 01:25 AM
It is the worst. :devil:

I had dreamt the imagination up to now.
It apologizes to you.

The modeler seems to calculate the calculation of the polygon normal with the 1st, 2nd, and 3rd vertex.
And, when three points are straight lines or it dents, a normal calculation has been generated an error.

The polygon normal guide is not displayed.
Moreover, the tool of the bevel etc. doesn't move.
There is a thing that the modeler crashes, too.
It is a bug. :devil:

GregMalick
09-18-2007, 02:08 AM
Like I said, it's not simply that the points are in a straight line.
If stretch the middle poly and you watch the poly on the left, the normal will not be a problem when the points are in a straight line.

The poly normal on the left flips down when you stretch past the edge.
That's why I think the problem is in detecting the edge for the poly on the right.

This is all guesswork since we can't see the code.

But it is definitely a bug. And one that has existed since v7.5 8~

PICTRIX
09-18-2007, 02:13 AM
Like I said, it's not simply that the points are in a straight line.
If stretch the middle poly and you watch the poly on the left, the normal will not be a problem when the points are in a straight line.

The poly normal on the left flips down when you stretch past the edge.
That's why I think the problem is in detecting the edge for the poly on the right.

This is all guesswork since we can't see the code.

But it is definitely a bug. And one that has existed since v7.5 8~
The error is acceptable for the straight line.
It is a problem that calculated vertex has been changed.
How when the second vertex dents?

Then, the face is seen or is not seen or, it doesn't exist.
It is a problem that the modeler tool cannot be used.

GregMalick
09-18-2007, 02:18 AM