Non-Planar Polygons = BAD


#1

I would like to make sure that people know, who may be starting out using Lightwave, more specifically modeling, that it is never acceptable to have non-planar polygons on a model that will be in a final, rendered shot coming out of Layout (or any other package for that matter).

The most important reason for not having non-planar polygons is simple. At times, during rendering, the render engine may not know which direction the normal of a non-planar polygon is facing. Because of this “confusion” the renderer may have the normal facing one direction on one frame, and facing the opposite direction on another. What you see in the final rendered animation is a flickering polygon. Not good. Rendering = Time, Time = Money, Re-renders = More Money.

Having non-planar polygons, even ones obscured by other geometry, are not acceptable.

If you ever want to work in a production environment, as a modeler, your supervisor will check for non-planar polygons. It is “Modeling 101”. There are times when things may happen and you get a model that has non-planars because you are in a rush or for some other reason - but rest assured having them in your models will get you a visit or email from your supervisor. Having them all the time will reflect poorly on you.

Having clean models is essential in a production environment. Whether that production is at a large facility, or in your basement - the theories still apply.

This public service announcement has been brought to you by the letter “T” and the number “13”


#2

Also, it might be wise to add for some of the newer people to LW, that you can check whether your model has any non-panlar polygons in it but simply hitting “w” on your keyboard, and bringing up the polygon statistics.


#3

But surely subpatching kinda cancels this out? As a subpatched poly can’t be non-planar… or can it? :curious:


#4

nope leigh…you’re right:)


#5

No, but you don’t subpatch all models :wink:


#6

Originally posted by Leigh
But surely subpatching kinda cancels this out? As a subpatched poly can’t be non-planar… or can it? :curious:

You are correct.

My original post had an introduction explaining why I decided to make this post, but I removed that portion and just decided to go with the main text body. Contained in the first part was the explaination that this applies only to standard polygon modeling, meaning non-subd modeling.

Sorry for the mix-up. And thank you for catching that for me.


#7

Though sometimes if you create geometry by spline patching for example you can get large groups of polys that are nonplanar but will work fine because they are kind of all nonplanar together.
I think it has something to do with the average of the vertex normals changing in a consistant way along a surface.


#8

Originally posted by SplineGod
Though sometimes if you create geometry by spline patching for example you can get large groups of polys that are nonplanar but will work fine because they are kind of all nonplanar together.
I think it has something to do with the average of the vertex normals changing in a consistant way along a surface.

It has more to do with luck, probably.

Sooner or later you’ll get caught using non-planars and look like a dope. (“why is this model flashing??”)

If you work alone and have plenty of time to rerender, go ahead and use non-planars.

If you want to work in a real facility that has other artists and a pipeline where you hand off your model to someone else, learn to build it right the first time.

Dave Adams
I,ROBOT/Digital Domain


#9

Ya know what - I suped commercials before. If I’m staring down the barrel of a 12 hour deadline and that there’s either a million bucks or a lawsuit on the other side, ya better not tell me “maybe it will render ok”.


#10

Originally posted by SplineGod
Though sometimes if you create geometry by spline patching for example you can get large groups of polys that are nonplanar but will work fine because they are kind of all nonplanar together.
I think it has something to do with the average of the vertex normals changing in a consistant way along a surface.

The simplest way to avoid non-planar polygons is to triple everything containing more than 3 points. This will avoid render errors and completely remove non-planar polygons from you model.

If you care to read more about polygons, non-planar polys, and flatness I recommend reading:

http://www.newtek.com/freestuff/lwhelp/content/021_points_polys/003_pointpoly_polygons.htm

An important note on that page is:

“If a polygon is non-planar, but within the Flatness Limit, it may still cause rendering errors.”

So keep that in mind when/if you ever adjust your “Flatness Limit.”

I have never heard of what you are referring to Larry, and honestly speaking am very suspect. You may want to check your original source because it sounds like hogwash…

Every production I have been involved with, it is stressed to all modelers that non-planars are totally and completely unacceptable. Actually, here at DD it has never been mentioned because it is assumed at this point you know to stay clear of them…unless you are working with a sub-d cage. But that is beyond the original scope of my post.


#11

Originally posted by alanchan
Ya know what - I suped commercials before. If I’m staring down the barrel of a 12 hour deadline and that there’s either a million bucks or a lawsuit on the other side, ya better not tell me “maybe it will render ok”.

Murphy’s Law applies to Non-Planars.:blush:

RING!

“hello…um”

THIS IS YOUR SUPERVISOR

“Geez, it’s 4am!”

I KNOW. YOUR RENDERS ARE SCREWED UP"

“But I test rendered frame 11!”

GET DRESSED AND COME IN SO I CAN SLAP YOU.


#12

Originally posted by LWD
[B]It has more to do with luck, probably.

Sooner or later you’ll get caught using non-planars and look like a dope. (“why is this model flashing??”)

If you work alone and have plenty of time to rerender, go ahead and use non-planars.

If you want to work in a real facility that has other artists and a pipeline where you hand off your model to someone else, learn to build it right the first time.

Dave Adams
I,ROBOT/Digital Domain [/B]

Nothing to do with luck. Its experience. Ive never worked at studio or on a project where someone hands off a model without doing test renders first. I guess I must have been doing it right the first time since I cant recall ever having a model come back due to nonplanars. :slight_smile:


#13

Originally posted by SplineGod
Nothing to do with luck. Its experience. Ive never worked at studio or on a project where someone hands off a model without doing test renders first. I guess I must have been doing it right the first time since I cant recall ever having a model come back due to nonplanars. :slight_smile:

If you are putting non-planar polys in your non-subdivision surface models (straight polygons) then it is luck.


#14

Originally posted by SplineGod
Nothing to do with luck. Its experience. Ive never worked at studio or on a project where someone hands off a model without doing test renders first. I guess I must have been doing it right the first time since I cant recall ever having a model come back due to nonplanars. :slight_smile:

I’m glad you have been lucky, but don’t personalize everything; this isn’t the Story of Larry’s NonPlanars. These are general rules people need to learn to follow.

Let’s not turn this into another Splinegod thread, let’s keep it for everybody.:slight_smile:


#15

I have an experience with non-planar object, when I did my interiors. I used boolean, and of course the sides will be more than 4. for the unsmoothed material it work ok, but when i smooth it it became realy2 bad…

ps: is this non-planar thing?


#16

Smoothing is based on shared vertices. Sometimes with nonplanars or areas where you have lots of polys sharing a vertex or two in some area can give shading errors.
This can usually be fixed by tweaking the geometry, adjusting the smoothing angle or cutting/pasting polys (which inhibits smoothing because the vertices are no longer shared along that edge). A lot of times youll see this on the faces of extruded letters and tripling may not work so well. Cutting and pasting polys can help solve it. Also it can help where you have unwanted smoothing on a surface and turning down the smoothing angle doesnt help either.


#17

Originally posted by xvampire
[B]I have an experience with non-planar object, when I did my interiors. I used boolean, and of course the sides will be more than 4. for the unsmoothed material it work ok, but when i smooth it it became realy2 bad…

ps: is this non-planar thing? [/B]

Yes, but more correctly is referred to as a “smoothing error” which was caused by a number of things I am sure - non-planar polygons had a hand in this no doubt.

While there are times when using the boolean tools have to be used, a general rule of thumb is to use one of the “Solid Drill” tools (usually Stencil) and then build the geometry out (or in) from there. Booleans are tend to be very sloppy and you are (usually) at the mercy of the program to determine how the geometry is cut. Many times “Stencil” will help control this.


#18

I have an experience with non-planar object, when I did my interiors. I used boolean, and of course the sides will be more than 4. for the unsmoothed material it work ok, but when i smooth it it became realy2 bad…

select the poly’s with the smoothing error…hit x, hit v…error should be solved unless I misunderstood what you said:)


#19

Originally posted by LWD
[B]I’m glad you have been lucky, but don’t personalize everything; this isn’t the Story of Larry’s NonPlanars. These are general rules people need to learn to follow.

Let’s not turn this into another Splinegod thread, let’s keep it for everybody.:slight_smile: [/B]

Dave last time I checked this was a public thread in a public forum.
Randy posted some comments about nonplanars and I do fall into the category of “everybody”. I dont see the difficulty in keeping to a topic and leaving the personal comments out of it.
I would suggest carefully rereading the 4 guidelines that the moderators have posted up and that can be easily seen just as you post. :slight_smile:


#20

This thread seems like an information pissing contest.