Hi, ill try to answer your questions, Im working as Class-A surfacer in automotive industry, particular in WV concern, so my knowledge comes from concern standarts
How true is this and what other software is commonly used?
Alias is widely used, so as Icem Surf and even Catia surfacing mods like generative shape design or ISD (icem shape design)
for visualisation 3dmax, alias or any other kind of quality rendering software, physical based unbiased softwares are giving great results too
Do poly modeling programs play any significant part in the pre-vis stage or are they used by companies in the advertising stage or when producing photorealistic renders?
According to low control over surface quality, they are not much used, best renders and vizus are from class-a surfaces
Or is it entirely a nurbs field?
Mainly Bezier to be precise (base part of nurbs surfaces)
When it comes to the manufacturing stage, are the parts usually re-modeled using engineering programs such as Pro-E?
Engineering stuff like plates and whole construction, electric cables and so on are done in catia, visible surfaces (class-a) are never re-modeled, you dont want to degrade their quality.
You mention later, talking about 3ds Max, that ‘according to low control over surface quality, they are not much used’. So what purpose is it used for visualisation? Do you mean that Bezier/Nurbs models are modeled in Alias and then taken into 3ds Max purely to be rendered?
What are plates? Is the ‘whole construction’ the mechanics/underlying structure of the car?
Thanks for the help.
I have a lot more experience with poly modeling than I do with Nurbs. I enjoy modeling cars, so I’m interested in knowing what employment opportunities are out there for poly modeling cars. As far as I’m aware, it is done a lot in the gaming industry and I think it is used in advertising and film. Am I right here and are there any other common uses?
Yes, thats the way it works in my company. Bezier/Nurbs models dont need necesarily to be modeled in alias, it can be any relevant software.
Plates was bad expresion, we usually call it simpy “construction” and yes, its underlying structrure of car
If you realy want to do automotive poly modeling profesionaly, you should think about game industry or advertising agency. Im not much experienced in polymodeling fields, you should look for sites like CGSociety or other forum/web community interested in polymodeling and its uses.
Khalid was pretty spot on, the only bit I’ll add is I’ve heard in some studios they will work in Maya using a hybrid of Polys/Sub-D’s/Nurbs for quickly fleshing out concepts in 3D form before jumping into more detailed surfacing. Polys in general are more flexible and more forgiving, and Maya has pretty clean data exchange with Alias, so I’ve heard it being used though haven’t seen it first hand.
As far as what employment opportunities their are? It’s pretty tight and you won’t find someone willing to hire an auto designer based only around 3D. The non-NURBS techniques out there seem to just be a fill in for the designers to quickly sketch 3D ideas - there is still an expectation for you to be a full on car designer from sketching forward before that. Other then that film or ads is probably the main area, and even then that’s a lot more about being able to not just model but render and animate as well.
there is hope…
there is alot to be said about polys.
Yes advertising for sure. google cgcars
i have heard that in Europe some OEMs are using polys early in the design process as you have mentioned. equivalent to sketching. still has to be NURBS at the end though. its a lot easier to reverse engineer a poly mesh.
I particularly hear about it for Interiors. look for the “bentled” channel on youtube. “mudbox_to_alias” or “ip mud”
often the cad data is not available outside the OEM. its alot easier to cheat or recreate the look of the vehicle in poly’s . for film and motion, close enough is good enough. you can get away with a lot with fast camera edits and motion blur.
Poly’s are a level of security for distributing files, This protects the Intellectual property of the Design.
-Level of detail for mesh density and rendering times that come into effect.
you might be able to find examples on the Chrysler website. they use to have a public area that you could down load their Vehicle files.
Quicker and easier to work with. A lot of times, NURBS will also NOT translate in a friendly fashion to polygons. The nature of taking a mathmatically perfect curve and trying to break it into linear segments means you’ll wind up with gaps, or super heavy meshes when intersecting parts meet.
When building natively with polys or Sub D’s, you can be a lot sloppier in getting the overall shape without worrying as much about the perfection of your surfaces, because it will never be seen in real life. Have you ever watched a CGI scene and said “hey that car looks like the reflection on the rear door doesn’t quite line up”? VS have you ever gone to a car show and seen a restored car and say “ew…that door doesn’t line up”.
Plus polys only need to represent the visible surfaces. CAD data needs to represent all of the internal parts that hold everything together which when converted directly winds up being MUCH more data.
I hear people say this. I’ve never modeled a car in both poly and NURBS so I can’t say. Have you? I just can’t believe that it would be quicker pushing lots of vertices around rather than joining curves. Surely a dedicated tool such as Alias would be faster.
Why does it need to be translated? I hear it can be easier if texturing (uv mapping). I’ve never experienced any gaps when translating NURBS to polys. From what I gathered it is an issue if animating/deforming the model.
I have, and it generally is faster to do polys (especially if it’s your job) than it is to do it with Nurbs, because it is by nature less accurate. The other reason is NURBS modelling (especially for something like a car) requires a lot of understanding of surfacing strategy. With polygons, if you don’t like a part of a model you can delete a chunk off, add some new polys, and radically change the impact of the smoothed surfaces when you’re done.
This was a poly mesh for a Sub-D model I did back in college.
You can see the final model doesn’t look bad but if you look at the reflections you can see where the jankiness occurs in the surface. That has more to do with my crummy base mesh, but if I were to throw it in an environment where the highlight line wasn’t so crisp, you wouldn’t be able to notice.
You always need to translate between NURBS->Polys, the question ends up being “What happens when I do this”. Some software is very good about it, and some software (especially some of the more popular animation packages like Softimage or 3DS) tend to not digest the geometry as well.
But to get a good looking poly model it also involves a good understanding of surfacing strategy. The image you posted is a low-poly gaming model but when modeling a higher detailed, high poly car, the larger number of vertices can become very awkward.
This is just an image I googled but it shows what I mean. And this isn’t a very complex car neither.
Trying to keep curves looking good involves lining up all those vertices in the correct position. Then if an alteration needs to be made it has to be all done again …and again …(can you tell I’m bored of moving vertices ). How can that be easier than applying sweeps/lofts?
Well the example I provided was actually a Sub-D base mesh, but the point is relevant to your example as well.
A model like the one you show is typically taken and then has a “Smooth” function thrown on it (creating more geometry from the simplified mesh). In poly modelling, I can take a row of edges and extrude it rather quickly. No matter what I do, when I apply that “smooth” function to a continuous mesh, I’m going to get a reasonable looking shape.
With NURBS, you can build your base surfaces fairly quickly, but when you start looking at some complex transitions where you have multiple surfaces coming together, it becomes rather tricky to get it right and do it in a way that is managable. You need to trim away, align new curves, and hope that everything comes together right.
Both can be time consuming, but the advantage of polys in a case like that is that if you want to make a point where 5 things intersect, you can do so rather quickly and don’t need to worry about breaking tangency or continuity, as soon as you smooth your object - bam, continuity is back.
If you look on CGcars or some of these forums you can see some of the works these guys bang out. I think if anything it has to do with the fact that it’s a lot easier to figure out poly modelling since theres only 3 real elements (vertex, edge, and a face). Theres no real worrying about things like curvature continuity, adding fillets (the smooth will do it automatically if the edges are lined up right) etc.
With that said, I do know people who focused specifically on getting good at modelling cars in NURBS, and they can do it pretty fast as well, albiet not to automotive standards.
Point taken on board about NURBS. As I said, I’ve never modeled a car with them so I can’t say …yet, but I think that you’re greatly underestimating the difficulties when modeling high poly vehicles . Give it a go and let me know
Not underestimating it, I know it’s difficult, but I also know the toolsets and overall workflow is much simpler than the nurbs approach by comparison. Check out CG-cars or SMcars - just about every 17 year old with a stolen copy of 3D Studio and enough time can crank out a pretty decent car these days. I would love to tackle a NURBS car if I had any free time left.
But you haven’t actually modeled a high poly vehicle. The image I’ve attached is one of my projects in progress. It’s more high poly than usual because there are lots of angles going into curved surfaces so the corners and pinching need to be controlled. It gets to the point where this is not easily manageable and any alterations further on involve a LOT of backtracking. I have to question in situations like this if this is the easiest or quickest way to do it.
Yes I know CGCars and SMC. With the time, effort and dedication, a car can be made using any 3d software but that’s not the point.
Just busting balls - My point was not to trivialize either method but if you look at the above example (which is admittedly extremely complex, and possibly moreso than it needs to be you can see that creating a Class-A surface model also involves a lot of complexity. If you imagine tweaking a point on say the front bumper just to slightly adjust the curvature, there winds up being a lot of interdependency between the patches.
Only thing I would say is finish your poly car, then give Sub D’s a shot (the benefit there being you can keep your mesh lower quality and used creases to create some of the more complex curved areas, and then jump to NURBS.
Then you can say which one is easiest.
Edit: Just found this link: Gallardo done in Solidworks, much cleaner surfacing though not to the same standards that an Automotive modeller would build to.
I was not comparing a sub-d model to a Class-A surface model. I was comparing the two to the same standard ie. using NURBS purely for visualisation purposes. So your surfaces don’t have to be Class-A perfect, in the same way that sub-d’s surfaces are far from perfect.
It is modeled in sub-d. There’s no way I’d attempt that with classic modeling. For the reasons I mentioned above, to get a true representation of that form, the geometry is dense.
I’ll definately brush up on my NURBS and do it. I use to model in NURBS (Rhino/Solidworks) so even though I’ve never modeled a car, I have modeled products. If there’s one thing I’ve learnt from poly modeling it’s, when it comes to complex ‘product type forms’, mixing continuous curved surfaces with detailed ‘cuts’, it’s a long process that really doesn’t lend itself well to alterations. There’s a good reason why NURBS is used in Product Design and it’s not just for manufacturing reasons.