I have seen few videotutorials about:"What to do -IF- Fillet fails"
Thanks to you guys to make these videos,but i say:"Fillet should work" ;)
Time ago Michael said to add another solid library to improve Fillet tool
Fillet is highest priority because is the most basic-essential tool that everybody use:from basic modeling to advanced
It's time to solve this



@Burr : """4 way Blend.... I think we discussed that integrityware's kernel can do it..."""

4way Blend (obviously G2 ) means class-A- surfacing,means to be close to Alias Studio
What happened to VSR?...T-Splines? If so,it will happen to MOI too? :) :)


Is it feasible?
Integrity's Kernel expanded to do this+a different kernel to do better fillets
more selling options:
basic like now 300$
advanced with 4way Blend 600$ (just thoughts..)
We already discussed about all these things more times:What is on your Radar Michael ?

Hi Mauro, well for filleting it's a difficult case. A strong fillet engine isn't made by executing a single algorithm well, it's done by recognizing and handling a large number of special cases and corner configurations. That's something that takes a long sustained effort like 30 man years of effort before it's very good.

I can fix some individual bugs in the current filleter, and in fact I did fix a couple of them near the end of v4 (at a time cost of several weeks of focused work for each one). But since I don't want to work for 30 years on just that one area it's not really feasible for me to make a complete new fillet engine myself, it's something that I need to license from an existing library.

I have already spent several months doing in depth investigating of 2 different libraries so far, one of them did not pan out because although it had a strong filleter it was very sensitive to slight deviations in geometry like say a 0.1 degree deviation off of tangent on the seam of a closed surface. That made it difficult to guarantee a totally reliable data transfer into it. The other library was promising but seems to have fizzled out and isn't dedicating the large time investment needed to get filleting in particular up to a high standard.

There is another one that I would like to investigate too though, but since it is a fairly time consuming area of work I'm not sure when that will happen.

re: Blending tools - no unfortunately the IntegrityWare kernel does not have a very reliable mechanism for that. That's something that could also come from another library but it hasn't really been a big priority because the type of "patch filling" workflow where it would be used is just an overall delicate modeling strategy that has a very high learning curve. It's something for very advanced highly technical users, which is not overall a primary target area for MoI.

In general sub-d modeling has a better workflow for difficult surfacing than patch-by-patch NURBS modeling, and so that will probably be more of a priority. In fact the fundamental basis for this is already in place in MoI v4 with the SubD to NURBS conversion function that is working right now.

If you need improvements in these areas then a practical way to do that is to use other CAD tools in combination with MoI. It isn't very practical to expect that a $300 tool made by one person is going to completely replace every other CAD tool and do everything especially in niche areas for very advanced users.

Having said that I would like to make improvements in these areas but it is difficult to know the timeframe. The primary focus for MoI is more on things that I am able to actually innovate in like having quick fluid workflow.

- Michael

3+ years left on the t-splines patent. (?), US
But there are many more, more recent associated patents...

- Brian

Thanks Michael for quick answer :)
SubD importer looks very promising
I'm exploring it, as in my recent post

M

Hi Mauro, basically with the "fill in patch by patch" strategy it's a dying art form even in programs with multi-sided G2 continuity blending because it's so sensitive. Just having G2 continuity by itself doesn't do enough, even if a surface is G2 to another right at an edge, if the surface curvature changes too much in a localized area around the boundary it will leave some visible evidence of the patch topology in the model.

The other thing that happens is corners can be difficult, it's easy for the blend to have competition between different shapes right in corner areas. It's also difficult to make shaping adjustments with it.

It takes a great amount of experience to deal with these problems, so it's a very high learning curve type of modeling strategy.

In general it's a strategy that has been dropped by most people doing complex surfacing in favor of Sub-d modeling. It's a trend that has been going on for a long time, sub-d modeling was actually invented as an easier to use organic surface modeling strategy specifically because doing character models with NURBS patch by patch modeling is so difficult. "Geri's Game" back in 1997 was the big turning point for this in the "DCC" / animation-oriented modeling world.

So it's not an area that I'm particularly excited to put a lot of effort into. It can be useful in limited circumstances but it's not really a winning primary modeling strategy for complex surfaces.

You get better quality for things that are supposed to be smooth by making larger extended surfaces that get cut back rather than trying to fill things in patch by patch. If it's difficult to make out of one larger surface then Sub-d is better, it's more predictable and controllable. It has a learning curve too though, higher than 2D profile driven NURBS modeling but not as high as patch-by-patch surfacing type NURBS modeling.

2D profile driven NURBS modeling has one of the lowest learning curves so it's generally been an area I've been more focused on.

The Sub-d modeling method is a possible workflow now in MoI v4 with the converter which is currently the highest quality converter available. You construct your sub-d control cage in a sub-d modeler, not in Moi but it generally makes more sense for me to work on introducing sub-d modeling methods in MoI for advanced surfacing rather than patch-by-patch NURBS surfacing methods.

If I am able to, I would like to have more types of NURBS continuity tools in MoI too though, they can be useful it's just not a high priority because of all that I described above.

- Michael

Thx for this in depth answer and yeah makes sense :)

Other simple things in terms of qol would be, as I mentioned in a different thread, a simple UI color changer inside the MOI prefs.

Another thing I would like to see if possible is smart edge selection so that in this case the program would be able to select all corner edges inside this e.g.

I would like to have a realitme subD-modeling like Rhino7 have now.... i think il will upgrade my RhinoV5 to V7.
And different linestyles (hidden, centerline, centerpoints) and width's
And a Block / grouping feature would be very nice

Hi micheal

How long will we have to wait to see a set of Sub-d modeling tools in Moi? .... I was deciding to buy a polygon modeler but at this point I'm waiting for Moi with sub-D :)

Hi Tony, I expect it will be quite a while before a lot of direct Sub-d modeling tools are available in MoI. When they start out they will be focused on pretty light duty simple shaping. I don't expect that they will be a replacement for a dedicated poly modeler anytime soon.

The SubD converter that is in MoI version 4 should enable you to have a good workflow using a separate dedicated poly modeling program to construct the SubD control cage and then importing that into MoI.

So if you want to do some sub-d modeling I recommend to go with a dedicated modeling program for that now.

- Michael

Michael, i would like to know which polygonal modeler at this time being that most 'similar/compatible' with MoI's spirit?

Sorry my English.

Hi Elang, a few ones you could try:

https://www.rocket3f.com/
https://www.nevercenter.com/silo/
http://www.wings3d.com/
https://metaseq.net/en/index.html

These ones are all focused just on modeling similar to MoI. There are a lot of other ones that are part of a larger package doing animation and rendering also but they will be more complicated (and usually more expensive except Blender) since they have a lot more stuff in them.

- Michael

1) Analysis tools for curves, surfaces, objects build from many surfaces.
2) More easy way to search for naked edges and other errors.

Circular selection tool

Hi Michael,

first let me say congrats for the release of v4 and the new site design!

If possible, I would like to see in v5 zebra stripes analysis for continuity
and a crease edge function for the subd importer, if this can be somehow
solved. And yes, a grouping mechanism in the parts list.

Best regards
Stefan

I would like to have a simple material editor and a better objekt manager.

Hi, Michael.
We all want something here, but what's your plan?
When is the next version expected?
What will be the changes?
Or will there be a break for analysis?

Hi WN,

> Hi, Michael.
> We all want something here, but what's your plan?
> When is the next version expected?
> What will be the changes?
> Or will there be a break for analysis?

I don't have any detailed plan as of yet.

- Michael

1. waiting for a long time - fillet, chamfer
2 . better selection (lasso, hotkeys for 1,2,3,4,5: switch modes: blender approach is great 1,2,3 for poly, tab - object mode)
3. ogl improvements: reflections, mb other shaders, lines that are closed together ~ that makes buggy view, mb glass with transparancy
4. Bézier and better curve manipulation - SAI as example
5. Hotkeys: scroll up and down (adjusting radius: as designer i don't need to be precise)
6. scripting : points
7. instances

Hi,
- a Pivot tool like in polymodeling software would be verry helpfull.
The gizmo in Zbrush works verry nice and is simple to use.
- Some of the addons or plug ins that are around should be integrated in Moi.
this might be not so exiting but usefull. Because many especially new user dont know much about them and how to install them.
It took me quite a while to find some of them.

Apart from the fact that class A surfacing is a black art, it isn't a magical thing that the tools do for you. Even Alias with it's tons of surfacing tools requires a tremendous amount of effort from the user to help the software do its thing. I don't thing this should be an area that MoI should be focused at. OK, maybe some diagnostic shading/materials and a bit more robust fillet/blending tools are welcome.

On the other hand I believe that MoI lacks a proper industry standard layer system which will help with organization, selection and geometry visualization. It kinda works as it is with the styles, but I really wish there was a treeview control with drag & drop support and the concept of an "active" layer. In fact I experimented by using some javascript treeview widgets and treated styles as layers, but I ended-up abandoning my effort because it was a hacky way and nowhere near as robust as I'd wanted it to be.

The other thing would be more of MoI's internals exposed to scripting and proper documentation for it. The community has already done wonders with so little to work with. It's what every major software has done because it allows anyone to have a custom workflow, it lets you do any non-standard stuff you can imagine, and frees the developer from having to fulfill every little request. Not to mention that a powerful API will attract even more knowledgeable people to do even more complex plugins.

My 2c.