Advanced Modeling with T-Splines Webinar – Archive Now Available
For those of you who made it to my webinar hosted by T-Splines on the 26th of April and want to review the material, or for those of you who missed it, the archive of the webinar is now live on the T-Splines website. Click here to view. If you’d like to download the file that I used for the presentation, you can download it here.
I’ve been really thrilled with the positive feedback I’ve received from people about the webcast. I’m glad that so many people found it useful and informative. I spent some time tweaking the model after the webcast, just to see how well I could smooth out the top middle section of the t-tail. That further tweaked model is found on the layer called Step 9. I left Step 8 as it was at the end of the webcast. I would say that it’s about 98% there. It’s well past the point that I would need it to be for manufacture – in my experience the small disturbances in Zebra/Emap would disappear in the primer stage with even the smallest amount of block sanding. That being said, my goal is to always strive for perfection in my models. I spoke with Matt today, and he made a suggestion that ended up opening up a whole new door for me with T-Splines. If you look at the points across the trouble section, you can see there are some rather unusual manipulations. However, in smooth mode you can see that the T-Spline isocurve is very nice and smooth. The trick to doing this, and not spending forever doing it, is to utilize the new/improved tsMatch command. The concept is an extension of the BlendCrv trick I outlined at the end of the webcast, but this is just….well I couldn’t believe how cool this trick is. So here it is. I’ve intentionally made this isocurve that spans the center section look bad:
Sure, we could point edit it until it looks nice, but there must be a better way right? Let’s start by creating something that we want the isocurve to look like. Run ExtractIsocurve, and extract the isocurve on both sides of the line of symmetry:
Now split those curves at the point where the hit the mid section where we’re having problems, and delete the inner sections. Now your curves will look like this:
You can guess where this is going next – I’m going to blend those two curves. Just make it pretty, whatever setting makes you happy. Don’t join your blend to the input curves.
Now wouldn’t it be wonderful if we could get our T-Spline isocurve to follow that nice curve? Okay, no problem! Run the tsMatch command. Notice there is an option called “Midsurface.” In fact it’s the only option when the command is run. Click that. It asks you to select a T-Spline border edge loop to change. Ignore that “border edge loop” part, just click on your bendy T-Spline isocurves.
Hit enter, then it will ask for the curve to match it to. Give it your nice blend curve and hit enter. You can see where it has highlighted the bad isocurve, and where it will be pulling it into the nice blend curve. I’ve just been using the default settings so far with great results, so go ahead and hit enter.
Ta da! Can you believe that? I couldn’t either. Now, this is good. I mean really good. But here’s how it would be even more amazing. Part of why this worked so well for me the first time was because my T-Spline isocurves were already pretty darn close to this state just from point editing. By doing this, since you’re working with a degree 3 surface, making this isocurve nice and tidy will inevitably make the adjacent ones a bit wavy. You can keep going back and forth a bit, and get it perfect. I spoke with Matt at T-Splines today and suggested that they implement a “Multiple Matches” option, which is something that the Rhino MatchSrf command does have. He seemed to think it was immanently possible, but will get back to me later this week. I wouldn’t expect this to show up next week, but if it’s possible I know that the folks at T-Splines understand just how powerful this type of functionality can be to really perfect a surface.
You may also notice that some new edges have been inserted into the blend between the vertical and horizontal surfaces. This is a workaround to a very subtle issue I ran into on this project. The way T-Splines is implemented right now, if you have a crease that fades out within the 2 face region of a star point – mind you, not going through the star point, but just in the region of the star point – then the crease ends up carrying on further than you would expect. It travels to the edge of the star point 2 face region, and then fades out over 3 edges. So, if you pull up on one of the control points that are along the line of symmetry on the top of the horizontal surface, here is what you get:
That’s the crease from the rudder wrapping itself around the trailing edge. In practice, I’ve found that if my isocurve in this region is nice and smooth, the crease is for all intents and purposes not there. However, Matt has assured me that they have researched the issue and that it is indeed something that can and will be solved in future releases. There are two solutions right now. The first being, just use the trick I showed you above, and the crease will not be visible. Or, if for some reason you cannot abide that, just subdivide (I used exact in this case) the face just behind the star point:
This will create enough faces between your star point and your crease so that this is no longer an issue.
Personally, I’d rather just use the BlendCrv trick shown above, because I never want to make my T-Spline any heavier than it possibly needs to be. So until it’s resolved, there’s two perfectly suitable ways of mitigating it.
I’ve had some requests to show how I made the nice pointy tips on my model. I’m happy to share that, I’ll be posting in the next day or so a little how-to.