G2 continuity or Class A surface.

[JimCarruthers]

Hi,

What sort of errors are you talking about? Getting G2's generally not that tough...it's just that technically getting a G2 transition doesn't necessarily mean it will look "nice." One thing, Rhino uses the angle tolerance setting in the file properties when matching surfaces, you could tighten the default up from 1 degree to 0.1, anything tighter than that is exessive. And be careful that you're not worry about what turns out to be just a render mesh issue.

[EquiNOX]

Perhaps, I may be confused with the circumstance on vertifying if two surface is performed Class A or G2 Continunity. I understand, in order to vertify this, I'f need to use Zebra map. And if a Line between two surfaces isn't blending well or is off, its confirmed that its not Class A or G2 Continunity. So, Here's picture example underneath, I had Network Surface that does tangency match/blend with Sweep2Rails surface, and the zebra map confirmed its little off.
http://i12.photobucket.com/albums/a230/APrather/G2.jpg

So, doesn't that picture tell you that its not G2 Continuity or Class A surface? I am not even sure if its a strict demands to hace Zebra lines to line up correct through two surface.. The only thing I am really concerned is to render out perfect without detecting any surface error. So tell me something, clear me out on things about confirming two surface preferoming class A or G2 Continuity.


And be careful that you're not worry about what turns out to be just a render mesh issue.
I know things about setting up angle to lower tolerance, but I am not really sure I get what you mean about render mesh issue??, can you elaborate?

[JimCarruthers]

Hi,

One note, "class A" does not actually specify any particular continuity, it just means that surfaces are "aesthetically important" and is really just marketing jargon.

Looking at your picture there, it illustrates what I mean about the render/analysis mesh. What I see is a very rough mesh in the Zebra display, such that those surfaces could all be G2(though I suspect some tweaking is probably required, I would guess there's a 'singularity' at the tip of your sweep2 surface and it's verrrry tricky to match to them seamlessly)but I can't tell.

[Snecx]

Try using Zebra command to analyze the surface. Then, adjust mesh and use the following settings:

http://img47.imageshack.us/img47/1615/meshsetting0bm.jpg (http://imageshack.us)

Notice the max distance to edge settings. The value 0.001 produce very fine results for my analysis. Sometimes I use 0.0001 but be careful because such setting will produce very fine (dense) meshes but accurate for analysis. It might take a very long time to mesh if you're doing analysis to a huge part.

The answer to what you want is good quality surfaces with no reflection jumps. You need G2 continuity. Check the simple visualization below. Just make sure your reflection with Zebra analysis does not have that kink at the seam. :)

http://img107.imageshack.us/img107/2663/curvaturevisual7to.th.jpg (http://img107.imageshack.us/my.php?image=curvaturevisual7to.jpg)

@JimCarruthers:
It is for aesthetic reason BUT it is not just a marketing jargon. It really defines the quality of the surfaces. Many car manufacturers have requirement for class A surfaces.

Class A does not specify any particular continuity but a continuity range, which G2 and above.

As far as I know, it is important to learn Class A surfacing. Modern cars nowadays have these quality surfaces in them. Toyota, Honda, Lambo, Ferrari, Audi, etc. Anything below Class A will be ugly as you can clearly see reflection jumps between the seam, something that you really wouldn't want to have on your car body.

Correct me if I'm wrong. I've only learned these through my own studies of high quality surfaces. :)


Edit:

Class A surfaces are not mainly about continuity. Found few more links to share. :)

http://www.design-engine.com/stories/classa_surface/classa.htm
http://en.wikipedia.org/wiki/Class_A_surfaces
http://en.wikipedia.org/wiki/Freeform_surface_modelling

Class A (aesthetic, high quality) surfacing is not the most important aspect for everyone when it comes to modelling. However, aesthetic is the main thing about car body design, therefore Class A is important in this industry.

[EquiNOX]

JimCarruthers: Isn't that what expert want? to have aesthetic surface model? Perhaps for marketing purpose, or perhaps for inspirational feeling. As for me, my goal is to achieve becoming a skilled NURB modeler, It's imperative for me to understand the flow on establishing class A surface.

You are probably correct, there's singularity with those surface, I assume it has to be the curves. because I believe it has mismatch tolerance between the other curve which make a seams or separate boundaries between two surfaces tolerances. That's something I need to understand how to solve this or avoid doing this in future :)

Snecx: The picture of Mesh setting you posted is the exacly same that I have in Rhino. Assumedly, it has something to do with curves. What I just learned... to preform class A surface, curve has to be "one" in order to set single boundary or possibilty matched between two boundaries. which is done by extending curves instead of adding Interpolate curve to connect with the curve that is already developed. Is that correct? (I have attached pics for you to understand what I mean)
The website you attached with explaination of G2 or Class A, I have read that before and I thought it was great. I wish there's more clear explaination such as how to develop good g2 curve, and is able to detect if its a G2 or not. Because basically when I am trying to develop curve most of times I was unsure if it;s actually G2 continuity or not.

[Snecx]

@EquiNOX: Have you tried using a much lower value for the max distance to edge? Your screenshot seems to have jagged mesh.

To detect whether it is G2 or not, it is simple as I've said. If it is G2 continuous across the seam, the zebra stripes will be smooth. If it is not, try and use a higher quality meshing value to create a more accurate mesh. If that doesn't work, try joining the surfaces together. If all fails, the surfaces aren't G2 continuous.

For construction of the curves, you can blend 2 curves to join them together or extend existing curves. There is no absolute way to do it as far as I know. However, extending a curve by adding another curve at the end point, would not be a good choice unless you know how to adjust the points so it'll match the continuity of the existing curves.

Referring to your attached example (the one you are having), it is possible to create the second surface (outlined in red) to be G2 continuous to the first surface. You can send me the 3dm file and I can show you an example of how to do it, based on your set of curves. That would be much easier for both of us because we can deal with the same set of curves which are causing the problem.

Try isolating the curves you have problems with and export it as 3dm format. Let me know. :)

Edit:

I think the best answer to your question on how to make good surfaces with G2 continuity is by trial and error. When the Zebra analysis tells you it is not smooth, you need to figure out why. This is what I did for the past year learning Rhino and messing with the surfaces. :p

I can show you examples to guide you but I fear I am unable to explain things with words well because, I am not good with words. :D

[EquiNOX]

Yep.... Tried everything you suggested.. Here's zipped 3dm file. Let me know you found. Here's link http://h1.ripway.com/EquiNOX/Fix.zip

[Snecx]

http://img477.imageshack.us/img477/5858/clipboard029is.th.jpg (http://img477.imageshack.us/my.php?image=clipboard029is.jpg)

You have a lot of problems with your surfaces. They are way off! Sorry for being rude. :P

I can't continue with these surfaces because the edges aren't made accurately. See the gaps in the edges above.

I would suggest you to resurface those part and make them more accurate based on the wireframe you have created. Always try to zoom in close to see how far are these surfaces apart from each another.

You might need to tighten your other tolerance levels besides the absolute tolerance unit.

[Snecx]

Here's my try. I have rebuilt all the surfaces except for the main one (the biggest one) because I do not have the original curves.

http://img399.imageshack.us/img399/237/clipboard014ty.jpg (http://imageshack.us)

Note that there's a pinch in your original surface. The isoparms at that area seems to be a little dense. I would rebuild the curves again if this is my car, they seem to be a little off, here and there.

Hope this is not too frustrating for you to deal with. Good luck.

[EquiNOX]

Actually, that's what I did last night.... I redo curves, rebuild surface...

http://i12.photobucket.com/albums/a230/APrather/G22.jpg

I hope this looks better now?? Yes I am aware there's pinch at lower part of surface area, but my main object is the flow of the zebra, seem look much better than the first try. you are correct, the isoparm or perhaps control point seem look too dense. I can simply try to reduce number of control points or keeping control point apart from eachothers so it wouldn't create any pinch

I'll admit yes this got me really frustrated... but that doesn't mean I would give up :) The more frustration I collect, the better and skilled I would become later on in future.

[JimCarruthers]

Hi Guys,

"Class A" is indeed marketing-speak(or corproate-speak, same thing,) it does not specify anything except the feature Alias is trying to sell with it. (From the first time anyone ever brought it up)McNeel don't know what it means and as the Wikipedia entry says, stricly speaking it just means that something is aesthetically significant. Technically you could do whatever you want to call "Class A" surfacing with anything that has basic NURBS support, it's just a matter of how painful the process would be.
Your Zebra mesh settings are still too loose, by the way. I also would recomment working in a little more "top down" manner, trying to make larger more general shapes and detailing them as opposed to working kinda panel-by-panel, you're not going to see the forest for the trees.

[EquiNOX]

Your Zebra mesh settings are still too loose, by the way. I also would recomment working in a little more "top down" manner, trying to make larger more general shapes and detailing them as opposed to working kinda panel-by-panel, you're not going to see the forest for the trees.

This is the part I don't really understand what you mean... I am not even sure I get what you mean about zebra map is too lose? Tolerance set were 0.0001 how can it be too lose?

[Snecx]

He means your meshing settings. The max distance to edge, set it to a lower value to get a denser and more accurate mesh for preview. :)

[Snecx]

Hi Guys,

"Class A" is indeed marketing-speak(or corproate-speak, same thing,) it does not specify anything except the feature Alias is trying to sell with it. (From the first time anyone ever brought it up)McNeel don't know what it means and as the Wikipedia entry says, stricly speaking it just means that something is aesthetically significant. Technically you could do whatever you want to call "Class A" surfacing with anything that has basic NURBS support, it's just a matter of how painful the process would be.
Your Zebra mesh settings are still too loose, by the way. I also would recomment working in a little more "top down" manner, trying to make larger more general shapes and detailing them as opposed to working kinda panel-by-panel, you're not going to see the forest for the trees.

Perhaps the "Class A" term is just a selling point but being able to produce aesthetically pleasing and high quality surfaces (Class A) is an important skill, isn't it? The term probably is insignificant but the skills related to it definitely not insignificant.

Sorry I am still new to this but I am attracted to this Class A surfacing term. :D

[EquiNOX]

He means your meshing settings. The max distance to edge, set it to a lower value to get a denser and more accurate mesh for preview. :)

So tolerance of 0.0001, distance to edge isn't dense enough?

[Snecx]

The "Maximum distance, edge to surface" value does not have any relation to the absolute tolerance value. One example of a good and dense mesh for analysis is as follow:

http://img129.imageshack.us/img129/5181/clipboard023kb.th.jpg (http://img129.imageshack.us/my.php?image=clipboard023kb.jpg)

Meshing in Rhino is important to learn because you need to use them for analysis. It is also the reason why I do not prefer simpler alternatives such as "NPower Rhino to Max" translator for starters, besides the extra cost.

[EquiNOX]

Oh that! I never thought I would need it because I have NPower translator for rhino object to export ino 3ds max. The reason is, IGES could develop major pain in neck when some mesh goes wrong with Max. Also it crashes when rendering with MR. Beside, my computer isn't really top brand (2.3 MHZ 1.5GB ram, ATI radeon 9800 vid card).
So... wouldn't it still be necessary to change mesh setting eventhough I just use NPower translator?

[Snecx]

It matters because the settings are changed just for that specific analysis command and not the viewport mesh in Rhino.

Usually in Rhino we work with lower quality mesh so we can speed up the viewport but when we are dealing with analysis, we need something more accurate to deal with. So again I say, don't worry of adjusting the mesh settings because it will NOT affect the viewport mesh setting, just the analysis mesh will be touched.

[EquiNOX]

It matters because the settings are changed just for that specific analysis command and not the viewport mesh in Rhino.

Usually in Rhino we work with lower quality mesh so we can speed up the viewport but when we are dealing with analysis, we need something more accurate to deal with. So again I say, don't worry of adjusting the mesh settings because it will NOT affect the viewport mesh setting, just the analysis mesh will be touched.

Duly noted! I finally notice mesh quality do change when enable for curveature/map anaysis... also when enable in rendering viewport. I been into rhino since few months and never realize it. Something new for me to learn eh?

[VetteMan]

i tihnk lower values in mesh setting doesn't affect the zebra or surface quality over optimal settings. I'm trying to explain that if surface is good, there is no need to extra lower values in mesh settings.
and for your bmw, i think if you find good photo refs, use them. These blueprints aren't give enough information about car (fender lines, profile line, bumps on the body etc.). My advice is: draw lines roughtly on photorefs, these curves aren't for surfacing. Then edit them for blueprints by using curve analyze tools, draw new curves if neccesary. Surface topology can be seen more clear on photorefs than blueprints.

ps: I started the bmw6 too, i think it has tricky surfaces. İf i can get good results, i will aware at you.

[EquiNOX]

Yeah vette, thats right setting lower mesh quality for surface doesnt affect by zebra. I just didn't know until learned that mesh quality changes when emap/zebra or renderviewport enables.

I have been polymodeling cars with MAX for almost 3 years now... I just know I couldn't always trust blueprint, but to rely on picture of BMW 6 or any cars that I am modeling at different point of view more than BP itself. Its just that I am new to nurb modeling which takes up lot of challenge to get curve look identical to real life, avoiding developing seams when render and so on.

I have been hearing about drawing curves or lines on photo references as "blueprint" I have done it once before following by tutorial, but I never can get the accurate size... such as side and rear were off (Scale were weren't right afterall). Also the tutorial I did wasn't very clear at all.

It's good to know are about to do BMW6, perhaps we can share our thoughts if we are struggling at some areas. BMW 6 isnt that hard, I have completed 3/4 parts of body (The back part) which is easy.. the most struggle part I ever encountered was the front fender with Arch (Trying to avoid seams) but rest of those isn't that bad.

[Snecx]

... I'm trying to explain that if surface is good, there is no need to extra lower values in mesh settings. ...

What I meant was the mesh settings for the Zebra analysis, not under the document properties.

It is important for lower values in Zebra mesh settings because it makes a lot of differences.

Default mesh setting: (Zebra stripes appear jagged, continuity could not be identified)
http://img136.imageshack.us/img136/940/clipboard011uw.jpg (http://imageshack.us)

A denser mesh:
http://img158.imageshack.us/img158/904/clipboard020qv.jpg (http://imageshack.us)

The surfaces are good, but analysis can be affected by low quality mesh therefore there is a need for those adjustments on the mesh settings in the Zebra analysis menu.

[VetteMan]

i couldn't explain it because of my bad english.
In first picture max. angle is 20, in second it's 5 and in third it's 1. There are no difference between 5 and 1.
http://img81.imageshack.us/img81/1305/angle6dy.th.jpg (http://img81.imageshack.us/my.php?image=angle6dy.jpg)

[Snecx]

I guess what you're trying to say is that there is an optimal setting for the mesh therefore it is not necessary to go for some very low value?

Yes that's true. Usually Angle of 5 or the preferred setting, max distance, edge to surface of 0.01 is very good for analysis already. :)

[VetteMan]

I guess what you're trying to say is that there is an optimal setting for the mesh therefore it is not necessary to go for some very low value?

yes, thanks for putting right words.:)

[CGTalk Moderation]

This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.