Mini T-Splines tutorials...

[PhilProcarioJr]

I have run into a lot of people that struggle with T-Splines so I decided to put together some mini tutorials.

What would you like to see?

Comments and feedback welcome.

How to create a sphere with the proper topology using T-Splines....

Here are the curvature and zebra maps to show between the two. I would like to point out that the sphere with poles is perfect but as soon as you do any modifications to one of the poles your T-Spline mesh with turn to garbage.

Cheers

[PhilProcarioJr · Accepted solution]

Putting a Dome on a cylinder with T-Splines.

One thing to note in the video I move the plane up to 4.5" but after merging the geometry you will still need to move the point for the exact height. I just used a sketch and placed a point 4.5" so I could make the correct dome height.

[TrippyLighting]

Awesome stuff!

Would the quad ball primitive not do the same thing ?

[PhilProcarioJr]

@TrippyLighting

That's funny...how did I miss that...oh well live and learn that will teach me to do work right when I wake up in the morning...lol

Ok then I should have said remove the other sphere so people will quit using it.....

[jeff_strater]

That's a nice method for capping a cylinder, @PhilProcarioJr.  I always tend to use Fill Hole, mostly in the "reduced star" mode, but I think your method ends up with better geometry.  It is limited, as you say, to cylinders with 8 divisions, but that is not that big of a limitation, IMO.

Jeff

[TrippyLighting]

Yes, a little manual work goes a long way with t-Splines.

Phil's advice is spot on, as usual!

It's not really limited to 8 sided circles, but it is a lot less work and results in cleaner geometry.

[Fonzis]

Hi @PhilProcarioJr, hi all

First of all big compliments for the skeleton works in the other post.
If you start make videos like you announced i think this will be a big step in understanding better t-spline and them complexity.
I am sure the condivision of your know how will make me, and also many others, understand relevant things in t-spline, so thank you very very much.

Since my Fusion works better i was experimenting to connect t-spline bodys with each other, but i am encountering always the problems you describe in this video. Remove and reconstruct zones with triangles that appear when bridging or converting curvy surfaces to t-spline body... exactly like the sphere, same issue, and i can confirm its a bloody mess.

Mostly test to see how it fits turn wrong and i must undo to try other ways for union. Till yet i didn't find a way to correctly unite or split the cells and have continuos looking surfaces. Reconstructing somehow makes lose the continuity elsewhere. In a way or another, as you say if triangles are around many other cells become crazy and make strange twists.

While looking for a solution and understand more about unions i found this video. It explains a lot ok, https://www.youtube.com/watch?v=WV62TqZ_x2g but i still have no clue how unite correctly because i surely never have this 1-1-1-1 edge factor like he has in his cell subdivision. How to influence the subdivision? is there really a way to manipulate subdivision? Understand how unite unpair edged cells or critical zones where quads must diramate in more directions or curves, would be also very important to learn for me. It would be big help if you could explain how understand the cell subdivision and the union problem with complex curved surfaces.

[catot]

@PhilProcarioJr

That would be great, I think a lot of users finds your tutorials & feedback to be very useful. (I know I do).

What I would like to see:

- How to analyse the curvatures, identify bad areas, and cleaning up for smooth transitions.

- How to avoid bad curvature (high-spots/pinching) during modelling. (triangular faces)

Small tip (maybe you know this already):

You can avoid a lot of the clicking in the first screencast by turning on "Select through" under Select\Selection Filters, then you don't have to manually Select all the faces which are on the back side for the subdivides.

@Fonzis

You can specify the number of faces for the subdivide, I have done this several times, but it looks like it can be a little unstable in some cases.

[Fonzis]

@catot yes i know very well the subdivision tools, i subdivided till my ears started to take fire lol, i was talking about another kind of subdivision if i understood well the video that i posted, kinda subdivide an entire body to same sized quads at once before make the union... understand how prepare the cells for the union.
I mean when i add more edges and subd. rings or add loops its always good to use make uniform, but the uniformity in those places where the triangles are is forgotten or has big problems.
The guy from Rhino also says that as time the union is good, no worth make more loops or intersections near the curves but he says also that to prevent isocurve drift more loops that guide the intersection outside are good. Till there i am ok, but the dilemma remain also for me those 2 things that you mention, how make a proper transition in the complex union and how prevent the triangles.... with theory i think we are on the right way to understand, but a good example could bring the light we need in this dark tunnel 🙂

[TrippyLighting]

@Fonzis The light at the end of the tunnel, in particular as it pertains to your model is to completely re-model it in T-Splines.

Getting proper transition is enough work with a T-Spline modeled from scratch. Your model was recovered from an exported .stl if I understand it correctly. It has too many edges because the topology created by those semi-automatic tools is far from ideal. This is also what makes it difficult to "join" these different pieces of your mesh. If you continue with the approach you are taking with your current model you will come to a dead end.

When I say "re-model" I don't mean re-topologize either. I really mean re-model from scratch. That might be more work but I guarantee that the results are going to be far superior to what you are doing at the moment and that mesh will be much more editable and customizable to different hands that the mess you currently have on your hands (pun intended ;-).

Also, what you'll learn simply by practicing modeling with T-Splines will invaluable!

[Fonzis]

Hi @TrippyLighting

i am doing? lets say tryng.. because i still did nothing  lol

In reality at begin the intention was only to prevent redo the splines in fusion because they are corrupted if coming from dwg and lost intersections, i just was searching for the fastest way to become it as solid with continuity. Then the stuff became more interesting as i discovered tools like loft t spline and convert surf to t spline, asap i also want to test better soft modification tools.
Yes you are right, that it will be a dead end and needs be reconstructed there, and thats what i gonna do. The intention is like the original, create loft surfaces from splines that  in fusion are better as they remain editable, at same time matching the spline construct like normal guided lofts... i have good feeling for the results i already am testing

The mesh way was seen as a fast option where i hoped obtain the fform as entire and already united t-spline body. It was not possible ok, so... buried and amen.
Stl you can forget, since abandoned the mesh experiments i am back in t spline on the dwg. Before the problem was the performance of fusion itself and work in t spline was impossible for me.
Since i am again able work in t spline environement, i did what you see from the dwg that has lot surfaces where the transition are good, so i have convert directly surfaces into t-splines tryng adjust them and unite.
Redo it from scratch i will start asap, but before i need answers to a couple of questions to understand from begin the correctly behave of splines btween the planes when i loft, then i can start new project in t spline environement. If it becomes to dark i hope you are around 😉

Anyway this takes nothing away to the importance of how resolve the union complexity, because i/we need connections on curvy stuff generally. For example me i need it to can adapt later parts/accessories that follow the bodys shape and as you say also adapt to other hands. I keep this file we talk about as frankenstein version, gonna use it for experiments such as the unions.
Many other complex projects and simulations with stress test i would like to do in fusion, most of them i will surely recreate or bring in t spline environement.

[PhilProcarioJr]

@Fonzis

First thing I can tell you about your T-Splines is they are way too dense. When I work with T-Splines I keep the control cage VERY VERY SIMPLE. Almost all of my T-Spline meshes you see that look denser are after all the forms are finished and I am ready for my final surface. So I take my control cage and subdivide it and that's what you end up seeing in my posts. In the screen shots you have posted here I would at very least delete every other loop (maybe more) in both directions. There is no way your going to work with a T-Spline model with this much topology, nor should you want to. The one and only time would be if you want to add surface texture details and that's a whole other can of worms. Start stripping down this model to the bare basics needed for the forms. Trust me you will be happy that you did and you will be able to finish this model.

@catot

"Small tip (maybe you know this already):

You can avoid a lot of the clicking in the first screencast by turning on "Select through" under Select\Selection Filters, then you don't have to manually Select all the faces which are on the back side for the subdivides."

I have found this to be very unreliable in my work...lost 2 hours of work one time because it missed faces......but I am aware of it and I do use it from time to time. 99% of the time I don't want to select through to the back side and I don't want to forget that it's on.

As to your requests for topics I noted them and will try to make some soon. 

@Fonzis

"The guy from Rhino also says that as time the union is good, no worth make more loops or intersections near the curves but he says also that to prevent isocurve drift more loops that guide the intersection outside are good."

I don't think you are understanding everything he is saying. I have not watched the video but I see two things wrong with what your doing and what he said. You do add more loops to stop the isocurve drift, BUT that comes AFTER you have a completed form. Keep it simple and block out your forms and connect all your pieces then and only then do you refine your forms.

"The intention is like the original, create loft surfaces from splines that  in fusion are better as they remain editable, at same time matching the spline construct like normal guided lofts..."

Mistake number two, create the simple forms NOT using your splines to extrude from but rather to conform your T-Splines to after creation.....this is another huge mistake I see people making. I will try to make some videos showing what I mean.

Cheers

[PhilProcarioJr]

Something else I would like to point out about the "ISOparm drift".... This problem is almost impossible to solve manually and maintain CAD accuracy. This is EXACTLY the reason I use Topogun and retopologize my work. In other 3d apps this is referred to as surface tension. When surface tension happens you get ISOparm drift, the more the tension the more the drift. This is the runners broken leg in T-Spline CAD work. Anyways there is a tool in Topogun called the brush tool with surface relax.

[video]

https://vimeo.com/41330297

When you use this tool it literally removes all surface tension provided you have enough edge loops in your model. This is yet another reason I said you need more then one tool in your toolbox. Would I like to be able to do it all in Fusion...of course but lets be honest, are all the tools you will ever need going to be added to Fusion? The answer is no.

[Fonzis]

Hi @PhilProcarioJr

Is a bit confusing yes but we come more near to the point, there where i have doubts in the realization, i wanna keep as good as i can the shape, but take away rings means lose this precise shape that i try to mantain, thats why i try keeping more loops instead take them away.
If you say add it later ok, but how? as more rings or loops are deleted and added later as more he loses the initial shape from native conversion and adapts the rings how he wants? no?

"The intention is like original......

Not using my splines? i think there is some missunderstanding, perhaps i didnt explain right. Sure i gonna use my splines.
Whit that i mean like in acad i wanna create loft sections made of splines, and other splines as guide to loft it, like i am used to do when i need precise 3d that follows referenced curves. With the difference to acad that instead lose the associativity because is simple surfaces and acad is limited in the continuity, here i have t spline surfaces when i loft and 1000 tools more as simple associativity. At least thats what i expect with a real new project made from skratch.
Take in consideration that to obtain the precision of the guide curve or the loft spline itself, i make follow fit points of the spline tru the references of measures. If its not a organic or ultracomplicated curvy shape, 99,9% of time continuity works fine in acad with this method.
So i really wonder how can that approach be wrong? in reality i know it as one of most valid methods to create precise 3ds, have also t splines with it in my opinion makes it to the best method that exists atm if we talk about create.
What i surely can say, many videos if not most of them was copiators and surely not creators, the tools was applied to copy and not to create.
I am always open for new methods, so if you see anything wrong its surely not a bad idea if you can indicate better ways for precise 3d creations combined with the flexibility of t-spline .

[PhilProcarioJr]

@Fonzis

I understand everything your saying and your using "Surface Information" to base your logic on, but see here is the problem. Surface modeling takes your CAD splines and rails and interprets the surface between them with a complex mathematical formula. T-Splines do not interpret using the same mathematical formula as Surfaces do so you will get two totally different results. So until you change the way you think about building models with T-Spline your not going to make progress. Right now your trying to remove a bolt with a screw driver. Hopefully I will get some time soon to make some videos on this.

"Is a bit confusing yes but we come more near to the point, there where i have doubts in the realization, i wanna keep as good as i can the shape, but take away rings means lose this precise shape that i try to mantain, thats why i try keeping more loops instead take them away.
If you say add it later ok, but how? as more rings or loops are deleted and added later as more he loses the initial shape from native conversion and adapts the rings how he wants? no?"

Did you see the skeleton I am making? I can promise you it is as accurate as you can get with a solid model and I used all the methods I am describing to you. Your just not understanding what I am saying yet. 

To be honest I'm not sure your ever going to be happy with T-Splines in your workflow and the work it will take for the accuracy you want. The surface tools and T-Spline tools need more work to do what you expect them to do.

[Fonzis]

@PhilProcarioJr

Sorry but now you made me to horny, cant wait till you made the vid 😄 now i dunno if is better wait or make a little experiment and loft a normal surface  to compare it with a t spline loft?... i wanna know what is the difference and how is the cell generation behave, can i set the density myself in the command? does it auto-generate the density when i follow guides?  and also important, can i loft from surf edges of t spline body to next loft spline like i expect?

I agree what you say at the end, but for some organic or semiorganic of my projects atm i see only in t spline  the functions that i really think to need.

[Fonzis]

never expected it is so easy when splines are there correctly.
I made this test and used the rails that he see as uncorrupted.... i would say this test announces me positive results coming, specially if i compare it to a conversion from a surface. I mean even if i have less quads he maintains a lot more better the shape with rails.

I am tryng to make those splines that was imported as open to closed ones like they was. I think if this works i will not need witchcraft to loft it directly in t spline environement like i was imagining 🙂

[PhilProcarioJr]

@Fonzis

I have to be honest I don't understand why your using T-Splines....the way your using T-Splines you might as well just use surface modeling...

You will not be able to edit them with any kind of control....and stitching them together is still going to be a hassle...

[Fonzis]

Is to later can adapt eventual accessories or other parts and integrate them to the existing shape with bridge union for example. Also because the possibility to make more accurate modifications to adapt it to other hands, with all those quads i am thinking can work on specific zones by using soft modding tools. With normal surfaces i can only play around at the spline intersections if the loft is and remain associative.

Normal or t spline both version need first a good spline construct, plz correct me if i think wrong, so if lofts will not be to many and not to complicated it will no be big deal make both version of them.
Thats where i see the advantages of t spline, but i would like to know anyway how would you make this because beside some thinking mistakes i have also doubts about wich one is the right way to go with security.

Lets say in fusion with lofting the normal surface way i come to the same point as in acad it became impossible. Can i be sure with normal surfaces loft here in fusion it becomes good continuity? then ok i must not worry to later be forced make complex unions.  I can use occasionally t spline and steal what i need when i must adapt.
My doubts that have bring me to eliminate normal surface way come from my convincion i need anyway t spline to close critical zone and will be forced to convert and unite even if i make normal surf lofts.
Am i making a mistake? and do you suggest me to go the normal surf way and thats it because fusion will do it where acad failed, no worrys?

[PhilProcarioJr]

@Fonzis

"Is to later can adapt eventual accessories or other parts and integrate them to the existing shape with bridge union for example. Also because the possibility to make more accurate modifications to adapt it to other hands, with all those quads i am thinking can work on specific zones by using soft modding tools. With normal surfaces i can only play around at the spline intersections if the loft is and remain associative."

I don't understand this either from a manufacturing stand point....If you want different interchangeable parts you wouldn't want to merge the surfaces. You would keep the parts that can be changed out separate from the frame.

The point is you still have way to many subdivisions....As far as which route you should take, it all depends on what your wanting to do with your model and how you are wanting to work with it. If you want parametrics or not, then you say you want complete accuracy to your curves but then say you want to use soft modifications to your model which then deviates from the spline accuracy you claim to want. This is why I say I really don't know how to help you because you don't seem to have a clear goal for this model. 

"My doubts that have bring me to eliminate normal surface way come from my convincion i need anyway t spline to close critical zone and will be forced to convert and unite even if i make normal surf lofts."

You will still have to convert the T-Splines to surfaces and work with the surfaces anyways so I'm confused here too.