PDA

View Full Version : Bug Report: Changing Smoothing Angle doesn't take effect during current session



PabloMack
08-22-2014, 11:19 AM
I am using LW 11.6.1. This feature seems to have broken in Modeler since version 9X or 10X at least. When I check or uncheck smoothing or change the angle in the texture panel it does not take effect during the current session in the preview. I have to save, exit then reload the model to see the change in the preview.

lertola2
08-22-2014, 02:03 PM
That does not happen for me. Changes to the smoothing angle update quickly here. Mac or PC? Could it be a graphics card issue?

spherical
08-22-2014, 04:30 PM
Works as expected here, too.
OK. "Preview". Which Preview?

PabloMack
08-22-2014, 08:09 PM
That does not happen for me. Changes to the smoothing angle update quickly here. Mac or PC? Could it be a graphics card issue?

PC. This same graphics card and system worked with all previous versions.

PabloMack
08-22-2014, 08:15 PM
Works as expected here, too.
OK. "Preview". Which Preview?

I open quad view and it is the same for all preview windows. I make a change and I see no change in any of the preview windows. But if I save it and reload it I see the change in all preview windows. It's all or none.

spherical
08-22-2014, 08:46 PM
OK. We gotta get on the same terminology. Viewports going clockwise: Top, Perspective, Right, Back. (Which are named wrong but that's been covered previously elsewhere.) If those are your Previews, I changed every setting I could find that would have anything to do with model display, and a few that don't, and cannot make it fail. I was going to ask what your GL settings are but nothing that I did with them made any difference. I'd guess that a good step would be to trash Modeler's config and generate a new one.

That said, if this has been broken for you since v9, then there's decidedly something else wrong or users would be complaining all over the place; especially having been that long.

PabloMack
08-23-2014, 08:27 AM
Of course you are right. We do need to get our terminology straight.

All viewports show the same degree of smoothing regardless of setting and type of view. I just changed the angle of smoothness for another texture and the viewports updated (accurately) immediately. I then changed the texture I was seeing the problem with to 100 degrees (from 30 degrees) and then all viewports updated immediately. But then I changed the smoothing angle back to 30 degrees and all viewports stayed looking as though the smoothing angle was still set to 100 degrees. It is inconsistent. It is difficult to trace down a cause when the problem is sporadic like this.

I am not so sure that the types of views are named wrong. I have found during my interactions on different forums that artists and engineers often think differently about the same subject. While an artist wants to look at the "front" of his 3D object, an engineer (more so than artists do) vicariously puts himself in the place of his avatar and looks forward as his character is doing. While an artist wants the default view of his object to be facing him as though he were looking in a mirror (i.e. turned around to face backwards), an engineer is mentally placing himself into the body of his object so it must be, by default, looking in the same direction he is looking so that their right arms correspond (engineer and 3D character), left arms correspond etc. This means that an engineer prefers to see the "back" of his 3D character by default because they are both looking "forward" in the same (+Z) direction. Other software packages have the character looking in the negative direction which has detrimental mathematical consequences and causes major problems in engineering. I think the naming convention and coordinate system in LW reflects the way engineers think and so I hear a lot of complaining from those with more of an artistic point of view. Of course, this is all my opinion but from experience having a great deal of interaction with lots of engineers and artists.

I stipulate that both Americans and the British used a "right hand rule" when deciding which side of the road to drive on. Americans applied the right hand rule by driving on the right side of the road. The consequences are that you now have to put the driver on the left side of the car so that he can be closer to the center of the highway in order to better judge how close his car is to oncoming traffic. The British applied the right hand rule by placing the driver on the right side of the car. So the consequences of this is that the car must drive on the left side of the road for reasons I outlined above. Now I ask you "Which one is correct?"

MrFurious
08-23-2014, 10:49 AM
I have the same issue where smoothing angle doesn't update. Close then re-open and it's as it should be. Happens sometimes but not always and I've yet to isolate a cause

spherical
08-23-2014, 05:12 PM
All viewports show the same degree of smoothing regardless of setting and type of view. I just changed the angle of smoothness for another texture and the viewports updated (accurately) immediately. I then changed the texture I was seeing the problem with to 100 degrees (from 30 degrees) and then all viewports updated immediately. But then I changed the smoothing angle back to 30 degrees and all viewports stayed looking as though the smoothing angle was still set to 100 degrees. It is inconsistent. It is difficult to trace down a cause when the problem is sporadic like this.

Well, that's just weird. Why on Earth would it stop? I'm leaning toward lertola2's assessment. Which video card and are the drivers up-to-date? There have been a number of changes in the OGL code in recent versions, so a driver may now be incompatible.


I am not so sure that the types of views are named wrong. I have found during my interactions on different forums that artists and engineers often think differently about the same subject. While an artist wants to look at the "front" of his 3D object, an engineer (more so than artists do) vicariously puts himself in the place of his avatar and looks forward as his character is doing.

Not the engineers I work with. I'm an artist and an engineer. Work in both worlds. Here's an example. I've obliquely rotated a pentagon and then selected two polys, so as to determine what we're looking at and which way things are orientated.

123846

Standard orthographic projection, unless using the German convention which rotates each view the other way (the object falls over in the direction of the view instead of "skidding out feet first", as it were, and this tripped me up frequently when working with the Can-Am 917 Porsche engineers), you have Top, Front, Right as the three standard views. They are arranged thus:

T
F R

If LW's Top view is looking down at the top, and the Right view is looking at the right side, "Back" should be labeled "Front". If "Back" is correct, then "Top" should be "Bottom" and "Right" should be "Left". The labels aren't consistent with each other.

PabloMack
08-24-2014, 07:48 AM
If LW's Top view is looking down at the top, and the Right view is looking at the right side, "Back" should be labeled "Front". If "Back" is correct, then "Top" should be "Bottom" and "Right" should be "Left". The labels aren't consistent with each other.

Which way do you face your model's? If you face them forward (i.e. looking in the +Z direction as you the modeler are also looking) then you will be looking at the back of the model in the port (appropriately) labeled "back". It seems to me that most artists build their models facing in the -Z direction (i.e. backwards). That is logically backwards but artists want to be looking at the front of their models by default when the application comes up so to them the view is labeled wrong. My models always face forward (i.e. in the same direction I am looking so that we are both facing in the +Z direction). That way my right arm corresponds to the model's right arm and my left arm corresponds to the model's left arm and so on. I am vicariously looking through the eyes of my model instead of looking at him face on (turned around backwards to face me). Of course this way of doing it only benefits you for bilaterally symmetrical animatable models and doesn't apply to such things as vases which really have no front and back. But, arguably, you have much more challenging work ahead of you with models that have to be rigged and animated and most of those are bilaterally symmetrical. So it makes sense to me to have a coordinate system and modeling convention that is most suited to meeting those challenges. i.e. the model has its own local Z dimension that is positive in the direction it is facing and that by default is looking in the same direction that the modeler is looking. It is trivial to turn the model along with its local coordinate system around to look at it from the front during the modeling process to see it face-on.

When you face your model backwards then it causes further problems because "left" becomes "right" and "right becomes "left" for the model. But artists often miss this because they interpret "right" to mean the artist's right and not the model's right. This comes from the modeler failing to realize that these terms are to be applied to the model and not to the artist who is doing the modeling. Indeed, the artist can turn the model around to look at the model from any orientation and angle so an artist's relative directions become irrelevant and not useful for these naming conventions. The model's right arm is still the model's right arm regardless of how the artist is looking at it. The model's top does not become its bottom just because you are looking at it up-side-down at the moment.

PabloMack
08-24-2014, 10:17 AM
I have the same issue where smoothing angle doesn't update. Close then re-open and it's as it should be. Happens sometimes but not always and I've yet to isolate a cause

Thanks. Now I know I'm not (totally) crazy.

ncr100
08-24-2014, 02:30 PM
Is the bug suitable for FogBugz ? (http://fogbugz.newtek.com)

spherical
08-24-2014, 04:49 PM
It is trivial to turn the model along with its local coordinate system around to look at it from the front during the modeling process to see it face-on.
<SNIP>
When you face your model backwards then it causes further problems because "left" becomes "right" and "right becomes "left" for the model. But artists often miss this because they interpret "right" to mean the artist's right and not the model's right. This comes from the modeler failing to realize that these terms are to be applied to the model and not to the artist who is doing the modeling.

Again, I come from an engineering background, where Orthographic Projection is, and has been, a standard. I don't have any problem working a model by looking it in the face, as if it were, well, another person that I am dressing. I can keep the orientation consistent in my mind, as I flip the model around in my brain. Never have had any issue with that and the model's "Left" and Right" are absolutely no problem for me to keep separated from mine. I don't have to make myself be the model to be able to understand it and prefer not looking at the back of its head.

Object rotation between ortho views follows a convention. I didn't invent this; I work within it. You are using First Angle Projection. I use Third Angle Projection because, for the engineer (in the US and Canada anyway), it is more logical and less confusing; especially when working in 2D drawings, because the projection lines do not have to cross the entire part, and obviously each other, in order to be represented on the opposite face of the unfolded box that ortho projection is. In simple parts, this may not be a problem, but complex ones can be confusing with all the crisscrossing projection lines. Why use 1st Angle Projection in a 3rd Angle World? (http://technologyinterface.nmsu.edu/Fall08/10_Melto/)

You may not come from a 2D heritage or may not have an engineering background. As you say, with 3D applications, one may now switch between the two projection types whenever one wishes. We can just agree to disagree on how things are labelled because, in the long run, they're essentially irrelevant; unless you are trying to explain the part to someone else.

PabloMack
08-24-2014, 08:04 PM
Spherical: I apologize if I came across too preachy. I do not come from a drafting background. So the terms "1st Angle Projection" and "3rd Angle Projection" are new to me. Thank you for teaching me something I was unfamiliar with. My background is Biology/Chemistry->Computer Science->Electronics so my engineering experience is light on mechanical but heavy on computer science and electronics. In CAD I deal with electronic schematics and 2D board layouts. The 3rd D is just layers so I don't deal with orthographic projections except in my CG work which is not fully professional. It is not so much in the modeling but it is in the animating that putting myself into the model that helps me visualize how to animate a bilaterally symmetrical model. In my modeling and animating I use the coordinate system a lot to computer offsets and such so it helps that the model is facing in the +Z direction and not the -Z direction. I had an electrical engineering professor who got on his soap box about how his life has been made more complicated because someone decided that electrons have negative instead of positive charges. Because of this, the mathematics of electron flow are backwards. No one ever calculates how protons flow through wires and electron boilers don't emit protons or even positrons. And my professor was right. That convention should have been thrown out a long time ago. I feel the same way about the mathematician's right-hand-rule. But then b:+ch:ng about it won't change anything but it is good to know that there are bad standards. As far as 1st Angle Projections vs. 3rd Angle Projections go, I guess I don't too much care. As far as I understand it, they don't change how you label the views. They just determine how you organize the views.

spherical
08-24-2014, 11:16 PM
No problem. That's why it's good to discuss things.


As far as I understand it, they don't change how you label the views. They just determine how you organize the views.

Yes. First puts the image of the part on the inside of the box; you have to look through the part to see the projection. Third puts the image of the part on the outside of the box; you look at the box in the same way that you look at the part if the box weren't there. I prefer to look at the object as if it were before my eyes. Especially if using a 3D mouse. It's all a matter of what you are comfortable with; until you need to expose another person to the part and their convention differs.

That's what I dealt with working on the Can-Am cars. Making matters, shall we say, more interesting, the Chief Powerplant Engineer spoke only a smattering of social English and I spoke near zero German. It was an extremely interesting evening in the bar after we had won Mid-Ohio when he successfully explained to me how the fuel injection pump for the turbocharged engine worked; using completely non-technical terms, a lot of hand gestures and a growing pile of napkins. After a while, we had a small crowd onlooking, because we got so into it. Pretty fun.

Anyway, back to the projections in LW; to me, it looks like they've mixed the two types. If you look at the First Angle examples and you place yourself in the position of the part, looking along the Z axis to the box surface, you are still looking at the Front view; it's just behind the part, as it should be in First Angle. Here's where it's screwed up.

IOW, with your explanation of how you work (you are the same as the part—looking toward the back) which is fine, but at the same time you aren't looking toward the top in the Top view. It doesn't matter whether the view is before or behind the part, you're looking at the top, not toward that direction. Translating this to a tumbling Perspective view, where the part is being looked at first surface on all sides, I can't wrap my head around thinking that I'm looking through it to the other side. Therefore, Third Angle Projection is a better fit. Open MeshLab, Rhino or AutoCAD and they have the viewports as I use and expect to see them.

MODO, however, is even more strange. Load up the Hippo and it has you looking down upon its top, up upon its bottom and at its front, but Right looks at the left side of the object and Left looks at the right side.

Why NewTek/Lux chose these mixed methods I don't know. So, except for instances such as this, I ignore what they think the label should be and it all works out.

Thanks for the discussion. :)

PabloMack
08-25-2014, 07:58 AM
I guess where I am missing something is that LW seems totally consistent as far as view names are concerned. But where their arrangement is involved, you are much more familiar with those conventions. I have made up an image showing all six views. But the uploader seems to be not working. So I will just describe it. The view labeled "Top" shows the top of the animal. The view labeled "Bottom" shows the bottom of the animal. The view labeled "Right" shows the right side of the animal while the view labeled "Left" shows the left side of the animal. The view labeled "Back" shows the back side of the animal and view labeled "Front" shows the front side of the animal. All of these names are consistent. Certainly this is the way the developers of LW intended modeling to be done in Modeler. But so many artists face their models backward and it screws up the naming convention. Beyond this, you can select any view you want in any port in any combination. But if you want "Right" to be looking to the right instead of looking at the right side of the model then it will be wrong. In fact *all* names will be consistently backwards for you but the names are all consistent as long as you face your models forward.

But if you faced the animal in the -Z direction then four of the names will be wrong. This is because front and back are switched as well as left and right are switched.

spherical
08-25-2014, 06:43 PM
That may very well be true but it all comes down to the basic fact that you happen to like being the actor, where I prefer being the director. When I direct someone in a film, I don't have them facing away from me when I'm talking to them or trying to show them something. We'll just have to agree to disagree on this. :) The other application examples that I found (didn't look very long—found enough) supported the convention that I subscribe to. MODO doesn't in an odd way that doesn't even match LightWave; which is a bit surprising in that it sprung from LightWave in the first place and a lot of basic stuff like this were carried over. All of the character/sculpting applications I have load models in facing me.

Just loaded up Rebel Hill's T-Rex. It faces me in both the "Back" and Perspective viewports. All of his rigs face me, as do any I've seen from any other riggers. I don't expect to have to turn a model 180 as soon as it loads into Layout before I can begin working with it there. Facing off to the horizon doesn't make a lot of sense... to me. Yes, it is the user's choice to turn the model around and have it facing away if desired. Just feels awkward... to me. That the huge consistency between other applications and users exists, indicates to me that when the model is facing the "wrong way", the viewports are labelled to match. This is not saying that you are wrong. There is no right or wrong, when it comes to preferences in how to model. It's just... different.

I guess we've beaten this to death and fully understand each other's point of view (pun intended), so can return to trying to figure out why the smoothing stops working all on its own; which is a far bigger problem than how a viewport identifier is spelled. :D