logo
NOTICE:  This is the new PunchCAD forum. You should have received an email with your new password around August 27, 2014. If you did not, or would like it reset, simply use the Lost Password feature, and enter Answer as the security answer.
Welcome Guest! To enable all features please Login or Register.

Notification

Icon
Error

Options
Go to last post Go to first unread
digitalphaser  
#1 Posted : Tuesday, May 28, 2019 3:17:15 PM(UTC)
digitalphaser

Rank: Senior Member

Joined: 4/17/2015(UTC)
Posts: 256
Man
Germany
Location: Berlin

Thanks: 17 times
Was thanked: 72 time(s) in 51 post(s)
Hi,
Finally, I ventured to create a complex object using surface tools.

Thanks to @NeuTechFLA, that he inspired me to this insanity.

As a basis, I took a relatively simple retro car Allard J2X.
Unfortunately the blue print references were not good. Different angles had serious discrepancies.
Therefore, I also looked at photos of real cars. For this purpose, I used reference software PureRef.

Of course in this way it is not possible to create an exact model. But I tried.
digitalphaser attached the following image(s):
5.jpg (219kb) downloaded 9 time(s).

You cannot view/download attachments. Try to login or register.
thanks 4 users thanked digitalphaser for this useful post.
jlm on 5/29/2019(UTC), magicart on 5/29/2019(UTC), Art Smith on 5/29/2019(UTC), flatpack on 6/3/2019(UTC)
digitalphaser  
#2 Posted : Tuesday, May 28, 2019 3:20:37 PM(UTC)
digitalphaser

Rank: Senior Member

Joined: 4/17/2015(UTC)
Posts: 256
Man
Germany
Location: Berlin

Thanks: 17 times
Was thanked: 72 time(s) in 51 post(s)
I spent about 20 hours to create a model. Some elements are repeated several times to find the most appropriate way. Surprisingly, but mainly I used only three tools.

1. Intrepolate Spline
2. Cover Tool
3. Join/Split Surface

For this type of modeling, Cover was the most reliable solution.
Now I do not understand why I need a SharkCAD. viaCAD Pro would be quite enough. ;)
digitalphaser attached the following image(s):
1.jpg (108kb) downloaded 15 time(s).
2.jpg (95kb) downloaded 16 time(s).
3.jpg (103kb) downloaded 14 time(s).
4.jpg (97kb) downloaded 10 time(s).

You cannot view/download attachments. Try to login or register.
NeuTechFLA  
#3 Posted : Tuesday, May 28, 2019 3:22:05 PM(UTC)
NeuTechFLA

Rank: Guest

Joined: 12/14/2017(UTC)
Posts: 263

Thanks: 31 times
Was thanked: 27 time(s) in 12 post(s)
Nice work, digitalphaser!
digitalphaser  
#4 Posted : Tuesday, May 28, 2019 3:22:59 PM(UTC)
digitalphaser

Rank: Senior Member

Joined: 4/17/2015(UTC)
Posts: 256
Man
Germany
Location: Berlin

Thanks: 17 times
Was thanked: 72 time(s) in 51 post(s)
I would like to know how other users model similar objects.

Here, video of my full workflow:

Part 1: https://youtu.be/YSuLkRNcjBU
Part 2: https://youtu.be/iShCk5-VX-U
Part 3: https://youtu.be/Z8HsXaO8x4Y
thanks 1 user thanked digitalphaser for this useful post.
MaiFy on 5/29/2019(UTC)
digitalphaser  
#5 Posted : Tuesday, May 28, 2019 3:26:42 PM(UTC)
digitalphaser

Rank: Senior Member

Joined: 4/17/2015(UTC)
Posts: 256
Man
Germany
Location: Berlin

Thanks: 17 times
Was thanked: 72 time(s) in 51 post(s)
Originally Posted by: NeuTechFLA Go to Quoted Post
Nice work, digitalphaser!


Thank you, NeuTechFLA. Unfortunately, not everything worked out.

What I did not like.

1. The impossibility of smooth joining surfaces. The Match Surface tool helps only on two surfaces. But it is not possible seamlessly to stitch the surfaces of the whole object.

2. There is no clever symmetry tool. Yes, there is a Mirror tool. But it does not give a complete match of the two halves.

3. Join Curve or Segment Tools convert Interpolate Spline to Control Point Spline or to Spline with huge amount of points. It is almost impossible to use to control the surface.

4. Repetitive objects seriously inhibit the computer. I made 50 copies of the tire tread. After that, I could not move the wheel. A little helped Remove Links. But it was still very slow.
digitalphaser attached the following image(s):
6.jpg (72kb) downloaded 9 time(s).
7.jpg (63kb) downloaded 8 time(s).

You cannot view/download attachments. Try to login or register.
NeuTechFLA  
#6 Posted : Tuesday, May 28, 2019 3:41:04 PM(UTC)
NeuTechFLA

Rank: Guest

Joined: 12/14/2017(UTC)
Posts: 263

Thanks: 31 times
Was thanked: 27 time(s) in 12 post(s)
Yes, I found all your dislikes as well on my journey. VC/Shark simply does not have the tools/capability to be used for true automotive surfacing. It can make decent approximations but does not even come close to being complete in that regard. But, Tim Olsen and his developers cannot be held in contempt for functionality the software was never created to do. In reality there are only four or five software in the world today can has this capability. Despite the limitations, your body skin is nice work.
Nick R  
#7 Posted : Tuesday, May 28, 2019 3:59:46 PM(UTC)
Nick R

Rank: Senior Member

Joined: 2/16/2007(UTC)
Posts: 127

Thanks: 1 times
Was thanked: 3 time(s) in 2 post(s)
I've asked Tim to add to Shark's surfacing tools for these kind of issues several times. Hope we will get some new features in this area for the 12 version when it comes.

Nick
digitalphaser  
#8 Posted : Tuesday, May 28, 2019 4:00:53 PM(UTC)
digitalphaser

Rank: Senior Member

Joined: 4/17/2015(UTC)
Posts: 256
Man
Germany
Location: Berlin

Thanks: 17 times
Was thanked: 72 time(s) in 51 post(s)
Originally Posted by: NeuTechFLA Go to Quoted Post
...Despite the limitations, your body skin is nice work.


The front of the body is not bad. Because I started from it and spend more time. With rear part I was already in a hurry. There are visible seams.

In general, this method is applicable not only to automotive. Ordinary plastic canisters for oil or detergents often also have complex streamlined shapes.



digitalphaser  
#9 Posted : Tuesday, May 28, 2019 4:07:30 PM(UTC)
digitalphaser

Rank: Senior Member

Joined: 4/17/2015(UTC)
Posts: 256
Man
Germany
Location: Berlin

Thanks: 17 times
Was thanked: 72 time(s) in 51 post(s)
Originally Posted by: Nick R Go to Quoted Post
I've asked Tim to add to Shark's surfacing tools for these kind of issues several times. Hope we will get some new features in this area for the 12 version when it comes.

Nick


I did not find in 11. version why I should upgrade. Maybe I will find it in v.12. :)
Nick R  
#10 Posted : Wednesday, May 29, 2019 2:23:48 AM(UTC)
Nick R

Rank: Senior Member

Joined: 2/16/2007(UTC)
Posts: 127

Thanks: 1 times
Was thanked: 3 time(s) in 2 post(s)
I meant that new surface should be added in version 12. Versions 10 and 11 haven't added anything significant in this area.

Nick
jlm  
#11 Posted : Wednesday, May 29, 2019 2:24:13 AM(UTC)
jlm

Rank: Senior Member

Joined: 2/18/2007(UTC)
Posts: 1,240
Man
France
Location: Paris & Frankfurt

Thanks: 204 times
Was thanked: 150 time(s) in 95 post(s)
Very nice Work !
I'm not very good in surfaces, I mainly use volume tools.
I wish I could...
murray  
#12 Posted : Wednesday, May 29, 2019 4:22:17 PM(UTC)
murray

Rank: Senior Member

Joined: 9/24/2014(UTC)
Posts: 373
Australia

Thanks: 8 times
Was thanked: 104 time(s) in 85 post(s)
You and NeutechFLA do impressive work with your models. I'd be really interested to see how you'd use the tools to create something original instead of reverse-engineering existing shapes. Have either of you had a look at smcars.net? It's a spin-off site from the Need for Speed game series, which lets players build their own models to import into the game, the "sm" part of the name stands for "self-made". The people who post on the site use a variety of softwares to model, from free subD modellers up to 3DSMax, and their considerations are a bit different to auto body surfacing generally, because they're trying to build convincing models while minimising facet count to keep the game calculations responsive. That aside, there are really talented people there and some of their models that they post WIPs and renders of completed models of are brilliant. It's interesting to me that subD software dominates their efforts. That aside, kudos to both of you. Like jlm, I think and model in solid terms, because I find it easier to imagine how a solid's surface intersections can be resolved at edges and corners than with surfacing tools, and I think that ACIS is more competent at that than it is as a surface modeller, because it originated as a physical modeller rather than a parametric surfacer. I also find that solids can often rebuild with a wider range of solutions to those edge/corner intersections than surfaces. There was a spin-off movie of "the Beverly Hillbillies" made after the actress who'd played Granny had died, in an early scene a visitor to Jed's house admires a wooden carving of Granny in the entryway, in the style of the totemic so-called "drug-store Indian" carvings that used to be shown as an American thing in some representations. Jed tells the visitor that he'd done the carving, when the visitor asks how he responds that he'd just "chiselled off the bits that didn't look like Granny", which is a humorous way of describing what a lot of "subtractive" sculptors (and even gemstone cutters) report, that they can envisage the result in the raw material. Surfacing could be regarded as an "additive" creative process, but the additive/subtractive regard is more ambiguous if you're surfacing against three-view or four-view projections to guide the shaping, because drawings sometimes show edge and corner resolutions that look like they're possible in 2D projection, in reality they must be either discontinous or impossibly self-intersecting, or the projections don't actually agree, but you don't find that out until a conflict presents itself.
NeuTechFLA  
#13 Posted : Wednesday, May 29, 2019 8:21:13 PM(UTC)
NeuTechFLA

Rank: Guest

Joined: 12/14/2017(UTC)
Posts: 263

Thanks: 31 times
Was thanked: 27 time(s) in 12 post(s)
Murray,

The reason I chose to reverse model existing vehicles was multi-fold. 1. It's what I do and have done for almost three decades so I know how a typical body surfacing project should unfold and I used that contrast and comparison to aid in the learning effort. 2. I needed to jump in, head first with a difficult "trial by fire" project to learn SharkCAD's "methods" at the extremes of the functionality Bell Curve. A vehicle's skin is the most difficult, demanding and challenging modeling endeavor on the planet. I knew I would find Shark's cracks very soon and I did. In other words, I knew pretty much that VC/Shark would do all the less demanding type work so I tried to break it. In reality, the two vehicles I created were not actual Class A Automotive body surfacing. I essentially created boundary/spline ruled surfaces/skins so I do not consider mine "real" by any metric. There are different methodologies within the CAD world as well. One you referred to was the "Granny" statue...I.E. carving off everything not wanted, or what I call resultant modeling. The other is developed geometry...I.E. creating something from nothing...similar to what I and digitalphaser did on our vehicles. Developed geometry is essential and required to resolve the conflicts you described above which are absolutely real and a hair pulling challenge at times. Then there is primitive/Sub-D methods made famous by Pixar and ILM. Shark, as most CAD software today, can do all/some of these methods but does not have functionality to do the finer aspects of real Class A surface work "within" the boundaries. Nor does it have any G2, G3 or G4 matching/control. As I stated up the thread, there are only a handful of codes (Icem/SURF, Catia(Suite), NX (old Unigraphics), Alias and a couple lessor ones that do this work properly. I did get duped a bit by the Shark sales and marketing information into thinking Shark had more than it promised. I take that responsibility, but for the money...it's damn content rich and very capable once one gets familiar with its landmines and steers clear.

I will check out the site you mentioned as well. It's always interesting to see the creativity of people.

murray  
#14 Posted : Wednesday, May 29, 2019 8:58:05 PM(UTC)
murray

Rank: Senior Member

Joined: 9/24/2014(UTC)
Posts: 373
Australia

Thanks: 8 times
Was thanked: 104 time(s) in 85 post(s)
Originally Posted by: digitalphaser Go to Quoted Post

4. Repetitive objects seriously inhibit the computer. I made 50 copies of the tire tread. After that, I could not move the wheel. A little helped Remove Links. But it was still very slow.



I built this model of a tyre in the open-source FreeCAD program. It mimiced the pattern of an actual 1950s racing tyre, but with very deliberate semi-circular sipes (that's what they call the grooves in tyres) that would be doable to a similar result if the (scale) model was produced by milling with a ball-head bit, photo-etched or 3D printed/rapid prototyped. I was trying to build it parametrically so that a range of tyre sizes could be generated from using diameter, width and aspect ratio as variables. It's still a work-in-progress, but the initial successful version in one size of tyre blew out to over 720Mb file size, because FreeCAD's history system means it keeps a version of each intermediate construction object in the file, and works through rebuilding each stage when the file is opened. That's a very slow file to work on, but it is workable. The tread pattern is all short segments of helical circular-section lofts and cylinders, that meet and resolve exactly at spheres at their intersections, and as you see from the brief clip of its history, they're propagated as nested arrays of arrays of arrays etc. The equivalent in Shark would be patterns, I can't remember if VCP has them in its tools.

Edited by user Wednesday, May 29, 2019 9:02:54 PM(UTC)  | Reason: Not specified

murray attached the following image(s):
tyre.PNG (200kb) downloaded 6 time(s).

You cannot view/download attachments. Try to login or register.
NeuTechFLA  
#15 Posted : Thursday, May 30, 2019 4:22:16 AM(UTC)
NeuTechFLA

Rank: Guest

Joined: 12/14/2017(UTC)
Posts: 263

Thanks: 31 times
Was thanked: 27 time(s) in 12 post(s)
Nice! Yes indeed, patterns like that have always been CAD's Kryptonite. Even today's high end codes, with their proprietary embedded pattern algorithms, can and do choke on intense/high feature count patterns like that. I checked out FreeCAD last year and found it interesting. It
s like if Catia, Creo, Inventor, NX and SolidWorks had a Love Child! :o) One can see the contributions from each of those developers in it.

How many individual "instances" exist in your model, Murray? 720Mb divided by...?
digitalphaser  
#16 Posted : Thursday, May 30, 2019 1:18:01 PM(UTC)
digitalphaser

Rank: Senior Member

Joined: 4/17/2015(UTC)
Posts: 256
Man
Germany
Location: Berlin

Thanks: 17 times
Was thanked: 72 time(s) in 51 post(s)
Originally Posted by: murray Go to Quoted Post
...I'd be really interested to see how you'd use the tools to create something original instead of reverse-engineering existing shapes...


I often visualize ideas. Sometimes I have either simple drawings or very uncomfortable references and I have to grab the idea of the author and visualize the object in 3D space.
The choice of software is the choice of convenience. It depends on the object, what software I use.

The most common workflow in CG industry is zBrush --> Retopology --> 3DMax(Maja, Blender, Silo etc...)
The creators of hard surfaces props often choose 3D-Coat --> Retopology --> SubD...

SubD has a small number of control points. It is very easy to control and quickly change the geometry. Used computer resource is minimal. Some models do not exceed 100 polygons.
CAD surface modeling is a bit similar to mesh polygonal modeling.

I know only few conceptual designers(game, animation, VR etc...) who use Fusion360 or MOI.

SharkCAD is an engineering software for any practical tasks like woodworking, casting etc... But VC/Shark are rather unpredictable, difficult and clumsy for conceptual design.

Edited by user Thursday, May 30, 2019 1:25:55 PM(UTC)  | Reason: Not specified

digitalphaser attached the following image(s):
wide_brim_hat.jpg (131kb) downloaded 2 time(s).

You cannot view/download attachments. Try to login or register.
digitalphaser  
#17 Posted : Thursday, May 30, 2019 1:24:00 PM(UTC)
digitalphaser

Rank: Senior Member

Joined: 4/17/2015(UTC)
Posts: 256
Man
Germany
Location: Berlin

Thanks: 17 times
Was thanked: 72 time(s) in 51 post(s)
Originally Posted by: murray Go to Quoted Post

I built this model of a tyre in the open-source FreeCAD program. It mimiced the pattern of an actual 1950s racing tyre, but with very deliberate semi-circular sipes (that's what they call the grooves in tyres) that would be doable to a similar result if the (scale) model was produced by milling with a ball-head bit, photo-etched or 3D printed/rapid prototyped. I was trying to build it parametrically so that a range of tyre sizes could be generated from using diameter, width and aspect ratio as variables. It's still a work-in-progress, but the initial successful version in one size of tyre blew out to over 720Mb file size, because FreeCAD's history system means it keeps a version of each intermediate construction object in the file, and works through rebuilding each stage when the file is opened. That's a very slow file to work on, but it is workable. The tread pattern is all short segments of helical circular-section lofts and cylinders, that meet and resolve exactly at spheres at their intersections, and as you see from the brief clip of its history, they're propagated as nested arrays of arrays of arrays etc. The equivalent in Shark would be patterns, I can't remember if VCP has them in its tools.

A few years ago I was doing a visualization of a case for portable speakers. This case had a zipper. When I created teeth pattern(about 200 Path Duplicate), I could not open this file anymore. It was absolute K.O. for Shark. :)


murray  
#18 Posted : Thursday, May 30, 2019 7:08:01 PM(UTC)
murray

Rank: Senior Member

Joined: 9/24/2014(UTC)
Posts: 373
Australia

Thanks: 8 times
Was thanked: 104 time(s) in 85 post(s)
For that tyre, the pattern has 80 radial repetitions, so there are 800 helical loft sections and another 800 mirror copies of those, 1600 cylinders, 2400 spheres and 160 cutouts around the side edges! I'd never done the head count of instances for it before and I'm impressed by that. It did stall a couple of times during the modelling when edges didn't stitch and it lost manifold integrity, but there's a "make solid" tool in FreeCAD that's like ACIS' stitching and healing that was able to make it good.
FreeCAD outputs the tyre as a 200Mb STEP file that takes up to ten minutes to open on this PC (old low spec 1.4GHz Celeron notebook, essentially a netbook but with 8Gb RAM) in both Shark and TurboCAD, and then I tried Shark PowerPack's unfold and TurboCAD's unfold on the tyre surface, with the result in the pic. PowerPack can struggle with trimmed analytic surfaces, while TC doesn't unfold general ruled surfaces, strictly analytics, but it does that very well.

Edited by user Thursday, May 30, 2019 11:50:01 PM(UTC)  | Reason: Not specified

murray attached the following image(s):
unfolding.png (227kb) downloaded 6 time(s).

You cannot view/download attachments. Try to login or register.
NeuTechFLA  
#19 Posted : Friday, May 31, 2019 6:57:45 AM(UTC)
NeuTechFLA

Rank: Guest

Joined: 12/14/2017(UTC)
Posts: 263

Thanks: 31 times
Was thanked: 27 time(s) in 12 post(s)
Impressive that you were able to "unfold" it. Now the challenge...create the same model but instead of a ruled cylinder (OD) tread add a crown or fullness (both directions) to the tread. In other words, "pressurize" the tread surface so it bulges outward a bit. If you can get Shark or TurboCad to unfold that set of surfaces, then there is indeed some power lurking under the bonnet. :o)
murray  
#20 Posted : Friday, May 31, 2019 3:36:46 PM(UTC)
murray

Rank: Senior Member

Joined: 9/24/2014(UTC)
Posts: 373
Australia

Thanks: 8 times
Was thanked: 104 time(s) in 85 post(s)
I think that I could bulge the tread surface graphically in FC, Shark and TC, but it be difficult and long-winded for me and that would be me doing it, not automation. TC's surface deformation by pressure loading wouldn't touch it, nor would Shark's. I'm pretty sure that the surfaces that I built could be distorted parametrically using VC/Shark's and TC's surfaces from law, but I'm not capable of that myself.
Unwrapping a compound curvature calls for surface expansion rather than development and that can't be done exactly by anything, but one of the prolific FreeCAD programmers has built a FreeCAD system that does expand compound NURBS with quantifiable "fudge factor", he's done it as a system for laying out patterns for shoemaking leather, and even Autodesk bought up Delcam to get that capacity and others.

Edited by user Friday, May 31, 2019 3:44:27 PM(UTC)  | Reason: Not specified

Users browsing this topic
Guest
Forum Jump  
You cannot post new topics in this forum.
You cannot reply to topics in this forum.
You cannot delete your posts in this forum.
You cannot edit your posts in this forum.
You cannot create polls in this forum.
You cannot vote in polls in this forum.