Full Version: Testing v3 for vehicle modelling

Show messages:  1-10  11-30  31-50  51-70  71-90  …  111-116

. wrong thread...

---

Why does this always seem to happen with me and Nurbs .... Time to start again me thinks with a new strategy, again










Any ideas are welcomed, and needed!

Martin

Image Attachments:
Image2.jpg  Image4.jpg 

---

Hi Martin, you mean that those surfaces are not meeting up?

How are you creating them, by something like a sweep? If you're doing a sweep make sure that you're not doing something like only selecting a single profile just on one side of the sweep - if you want a sweep to match up in between 2 different shapes then you need to select both of those shapes as profiles for the sweep so the sweep has not just a start profile but an end profile as well.

When you have both a start and an end profile for the sweep, the sweep will gradually morph between the 2 shapes through the traversal of the sweep.

If you give it only a single profile on one side, then the other end of the sweep is generated purely by movement of that one starting profile along the sweep rails, and that may result in some other kind of shape at the end of the sweep.

If you're still having a problem with it or the above explanation does not make any sense to you maybe you could post an example single surface that's not working like you expected so that I can look at it and give you some advice.

- Michael

---

Thanks for the offer of help Michael, I have this same problem with rhino too, which is why i came to MoI, so i know the problem is with me and methodology - No formal training so i must be missing something with the theory of it all :)

If you go back to the start of this thread you will see the two surfaces that have been zoomed in on on my last post, that they are sitting nicely aligned - The surfaces are either sweep or network surfaces, but they are fine. Once i start to slice up the surfaces (trim) for where the doors should be and to round off creases etc, then for whatever reason these surfaces start to move around as if by magic, I have tried ShrinkTrimmedSrf but it don't make a difference. The problem is I don't know when it happened or how, but because all the other trims, blends and fillets are done, and because i use surface edges rather then new curves where possible, the original alignments are gone.

On grab-04.jpg the right had trim line for the door has shifted the remaining surface segment from alignment without me noticing, but that one is easily repaired with control points or by deleting and using the new edge that got inserted during the trim process, the other surface - image4.jpg also shows how it has altered the surface curvature some how too.

The area shown in image2.jpg is perfectly fine in grab04.jpg despite having the trim going through part of the surface, but then some how during the process of;

Trim
Offset
Loft edges that need fillet
fillet corresponding surface
delete offset surface
join all surfaces created to keep everything as a group

Somewhere during that process it's changed it's position as if some sort of elasticity that held it in place has gone. I know it must be my technique. I don't single side sweep as i know it creates problems with continuity and accuracy, when i try doing vehicles I only use sweep, loft or network as you don't get many planar surfaces in vehicles as you already know. Is there a book you would recommend for me to read that covers Nurbs modelling rather than the maths, the theories and algorithms are way too heavy for my feeble brain ...

Many thanks

Martin

---

Finally got round to trying this vehicle again (been busy organising my wedding) - So this time i have not joined any surfaces and it's done as a patch work quilt thing. As can be seen I just have free-form surfaces with no depth or mass to them. If there are any experienced car modellers here, when should I trim ? should i Join before trimming or after ? Should fillets and Blends be done before trimming ?

With thanks ... Martin

Image Attachments:
Grab-05.jpg 

---

Hi Martin, congrats on your upcoming wedding!

re: when to trim - that would likely depend on what particular piece you are planning on trimming.

Usually you woudl want to use trimming fairly early in the process, but you usually do trimming when you have modeled larger extended pieces to then remove some material from the single large piece to make the final boundaries. When you model in a kind of patchwork way where you actually have constructed individual surfaces that go around different contours and holes then usually that is something that handles features separately from doing trimming.

Usually though you will get a higher quality result buildling a smaller number of larger surface pieces and then trimming those to produce some of the edges in the final result rather than doing a patchwork type process. The patchwork process will make for a complex topology of a lot of interior edges in the final result and you will likely have a lot of pieces that come to shallow angles to one another - either of these things tends to make filleting difficult.

- Michael

---

Thanks Michael - Can Moi take a number of sub surfaces and join them up so as to be able to do a Rebuild on the joined surfaces into a new single surface ? Or would be better off duplicating the edges as curves - joining them and then rebuilding those for new sweeps and networks ?

Martin

---

Hi Martin,

> Thanks Michael - Can Moi take a number of sub surfaces and
> join them up so as to be able to do a Rebuild on the joined
> surfaces into a new single surface ?

No, MoI does not yet have any mechanism for doing that, that is something that I want to add in the future though.

For now the main way to avoid making fragmented sub surfaces is to do the rebuild on your starting curves, so that you have longer initial starting curves that you use for creating the surfaces initially.

- Michael

---

Thanks Michael. Guess that means another re-start and shift in logic then ... I thought it went too easy when i started this ... :)

Martin

---

Hi Martin, well not necessarily, depending on what you want to do next...

If you are already satisfied with how your model turned out with the patchwork method then that could be ok to keep it as-is.

But you mentioned filleting in particular - you could run into difficulties doing filleting with that current model. I'm not really sure which particular areas of the model you plan to fillet though, could you maybe give some more specific information on what the next steps are that you were planning to do with it?

Just in general vehicle modeling is one of the more challenging and advanced types of projects to undertake, by the way.

- Michael

---

Thanks again Michael - The image below points to the areas to be trimmed. Those trimmed surfaces will then need to have chamfers or fillets added to them to create some depth and volume. I won't be attempting the interior, just need to get the outside looking right.






The image below has all of what needs to be rolled edges highlighted by MoI selection. These lines need to be rounded off rather than coming to a sharp edge. The other edges don't really need to be rounded as they are fairly accurate with the way that they just alter how light bounces across the surfaces.




I hope you can understand what it is i am trying to get across - Surfaces that are timmed then get the trimmed edge either offset or extruded as to have another surface to fillet. I choose fillets as they are more rounded than the blend which has that "s" type of shape to it and just don't work right in this situation. I have not found a way in MoI to extract what Rhino calls an isocurve which would have been more suitable for doing a subtle blend of the connecting sub surfaces to create the more massive surface.

Michael - "Just in general vehicle modeling is one of the more challenging and advanced types of projects to undertake, by the way."

Oh yes, I have been trying to master this art-form for years. I can do things like bottles, pistons, nuts n bolts and all that kind of modelling with no problems, that's why I can't grasp why I can't do vehicles. The only one i have managed to complete is an old 1950's bedford van as shown below.



Many thanks for the offer of help Michael - It's appreciated.


Martin

Image Attachments:
RoundedEdges-001.jpg  sample-01.jpg  Trimming-Lines-001.jpg 

---

Hi Martin, so trimming those marked areas should not be a problem you can do the trim at any point I would think.

re: filleting - you will definitely have significant problems filleting the areas that you show there with this kind of fragmented "lots of separate sub patch" type modeling technique.

It's difficult in general for the filleter to handle cases where 2 surfaces come together at only a slight crease where they are like within 5 to 10 degrees of being smooth with one another but not actually totally smooth. That kind of only slight sharpness usually means that the fillets that would be generated are very small in size particularly in corner juncture areas where multiple fillets are colliding together. The whole process of intersecting and cutting back fillets that meet at a juncture point tends to get confused when shallow angle corners since the fillets are close to having overlapping surface areas instead of having a sharp intersection line between them, and it also causes problems when the corner patch itself is extremely thin or slivery.

Also having more complex topology with multiple edges coming together at one juncture point also makes it hard for the corner aptch to be generated, so like for example this area here in your model is unlikely to be filletable:



Also areas like this where you only want to smooth some edges and not all of them can also make it difficult for the area at the juncture to be fillable:



Usually the best way to generate smoothness in areas like that is to build them out of larger surfaces instead of using fragmented sub patches. You can also build a surface to be extended and broader than you actually need the final result to be and then trim away some parts of it - it can often be easier to shape a more simplified 4-sided extended surface rather than trying to build surfaces that all directly hug every outline of your final result.

Also although you mentioned you don't like Blend , it may be better route for connecting up some areas of a model like this rather than fillet.

You're basically into an area of quite advanced surface modeling doing this kind of stuff, and there are some kinds of advanced surfacing tools that MoI simply does not have yet, like a "MatchSrf" tool for editing one surface to make it smooth to another one at an untrimmed edge. You may get good results taking your model into Rhino and using Rhino's MatchSrf or Patch tools to edit some areas of the model to make some smoother areas.

I do plan on working on adding more of these advanced surfacing tools in MoI v3 but for now using Rhino alongside of MoI may be your best bet for doing advanced surface "trim and patch" type modeling.

- Michael

Image Attachments:
car_fillet1.jpg  car_fillet2.jpg 

---

Hi Martin,

> that's why I can't grasp why I can't do vehicles.

Vehicles are a fundamentally more difficult shape to create - they are more towards the "organic" side of shaping with a lot of subtle variation in their shapes and not as fully defined just by 2D profile curves like other kinds of models.

They're kind of in a gray area in fact where you may be better off using tools that are more focused on organic character modeling to them like sub-d modeling tools.

- Michael

---

Thanks fort he reply Michael,

> They're kind of in a gray area in fact where you may be better off using tools that are more focused on organic character modeling to them like sub-d modeling tools.

You know what - That's got to be the best description of the scenario anyone has ever offered. I have tried my hand with Sub D but I find it way more difficult than Nurbs. The surfaces generated by Nurbs are wonderfully smooth and give pretty much instant gratification, sadly i struggle with poly modelling and the amount of time that needs to be invested in that technique makes it somewhat prohibitive to me - Instant results and the "honest" feedback that Nurbs delivers is far superior.

But i will give this lambo a try in Rhino and see what happens with an import to MoI and then Export to Obj to try in Blender Cycles. It's a shame I don't have more time to play with MoI as I would love to work on an exporting script for an external rendering engine such as LuxRender.

Martin

---

Hi Martin,
Can you post some of the sample surfaces that you want to see this on? You can just break out one area with the curves highlighted that you would like to see smoothed out.. Since you are already doing patch modeling, I may be able to show you a method to get the curvature you want.. (Be sure to include one of the complex juncture areas Michael is speaking about... The front fendfer area would be easy and not present any problems, but I dont want to lead you with this, if it cant do the other areas you need done)

---

Thanks for offering to take a look BurrMan, Here is the exported 3DM file. On the photo you will see the angular "seams", for want of a better word. Although these areas are sharp compared to most other cars, these are still like rolled seams. Feel free to play with it and see what you can come up with, as I am positive the error is with my modelling technique and not MoI's toolset. There are more fancy commands in Rhino, but they are still using the basic infrastructure that MoI has, more like refined scripts.

The area round where the door will fit needs to have more volume, as said in previous posts, I normally offset an edge and then loft them together to get surface which can then be Filleted.

Hope this all makes sense, it would be cool to learn how to do this part of modelling vehicles as it's always where i goof it up.

Martin

Attachments:
Merge-Fillet-Test.3dm

Image Attachments:
2008-Lamborghini-Reventon-019.jpg 

---

Part of your problem is the way you have built these surfaces. They will not join together, because there are edges that are not shared.

If you are going to make a patchwork type object, you have to make sure that each edge is joined to it's neighbours. Use the join command after you create each piece. There is a script to show open edges which you also need to use, otherwise you will overlook some edges that have not joined properly. When you get an eror you have to deal with it there and then. Going back later is not always possible.

---

Thanks for your reply SteveMacc, most of the surfaces will marry up, but the naked edges are not what the problem is, or I assume not - All of the surfaces need to have a fillet or blend applied first, and then they would be joined up after that process. If I join tem up now, the fillet wont work as the surfaces are joined - pop them apart again and it will work. If there was an Isocurv extraction facility then it would be as easy as trimming and then either blending or doing a 3 point arc and a twin rail sweep.

Martin

---

Hi Martin, you can fillet 2 separate unjoined surfaces but when you do that it performs a surface-surface fillet operation which is a little bit different than an edge-based fillet which is more usually used and which requires things to be joined.

Surface-surface fillets can indeed construct fillet surfaces in some kinds of difficult situations where the edge based one can fail, but one reason why is that unlike the edge-based one it does not try to construct corner patches where multiple fillets are coming together. So if you do surface-surface filleting you will usually have to do a lot of more low level trimming and manual construction of juncture areas.

re: Isoparm extraction, that is on my list of stuff to add, for the time being if you need isoparm extraction you would need to use Rhino in combination with MoI to do that particular operation though.

- Michael

---

I think you may run into problems with edges like this:



Personally, I would try to align surfaces as I built them rather than relay on trims and fillets later.

Image Attachments:
06-05-2012 22-07-49.jpg 

---

Show messages:  1-10  11-30  31-50  51-70  71-90  …  111-116