Hi Jason, thanks for testing that - yeah that confirms that the dense default meshing was the main thing responsible for that higher level of memory consumption as expected.

MoI is not really fully tuned up yet to handle higher detailed models, some of the things that you are seeing like apparent freezes while crunching away during exporting, and this higher default memory usage are basically due to that.

But I guess the good news for you is that it should improve from here! ;)

There actually isn't really too much stuff left to do, a few key things should tune things up quite a bit, and changing the density setting to be lower improves a chunk of that already.


> Unfortunately, being a trademark product I doubt I can get
> this model out. I can make an inquiry.

No that's ok, just if it happened to be easy for you to share it it looked like a good test subject because things like lots of little fastener pieces are the things that I want to have reduced mesh density on.

Thanks,
- Michael

In light of this though, I think once we get an object manager it just shows that under the hood MOI is capable of larger assemblies and isn't just for small products. I'm looking at tools like SpaceClaim and Innovate which cost far more but yet yield similar in-viewport performance.

Any timeline when we can start playing with an object manager? Once this appears, I'll gladly boast about the capability in MOI for dealing with larger assemblies; imported or not.

Hi Jason, just some other notes and information for you on your first message there. This is all just to explain some current priorities and goals, stuff like that. Not everything will align precisely with your own priorities I'm sure, but some should!


> My leading request is a native 64bit version, but I'm sure
> it's been asked before.

This one unfortunately I don't think will be likely anytime soon, mostly because more than one bifurcated versions of the program greatly increases the testing burden. I currently just do not feel confident that I could handle the additional time required to manage multiple simultaneous versions. Right now I basically need to focus just on one version to make sure that I can get it to a very high level of quality.


> re: layout builder or UI rearrangement

It will probably also be quite a while before I can focus on making a full sort of drag/drop type UI editing mechanism.

However, the UI is actually extremely malleable at a kind of lower level than that, because all the UI is defined by HTML files that are under the \ui subfolder in MoI's main installation folder.

You can actually edit the UI definition files to change things around if you want. It is certainly not as convenient as a high level drag/drop type mechanism would be but if you want to pursue it here are some links of some discussions about it:
http://moi3d.com/forum/index.php?webtag=MOI&msg=1173.1
http://moi3d.com/forum/index.php?webtag=MOI&msg=844.1
http://moi3d.com/forum/index.php?webtag=MOI&msg=1290.1


> 2) Enhanced GL settings. I would find it being advantageous for MOI
> to use the extended GL for additional lights, reflections and shadows.

Yeah, this one is fairly high up on the priority list.


> 3) Zebra striping: does MOI support this yet or any other surface
> analysis? Zebra striping would be a great initial addition.

This kind of more advanced analysis stuff is kind of a lower priority currently though. I expect to get to these functions eventually but there are a lot more things that are more sort of "core" to get done first, like a few example are object organization tools and better history editing tools. For MoI it has not really been a huge priority to focus on stuff like "surface styling" type things yet, the focus is more on making simple things work easily first and there is still quite a bit to cover there first.


> 5) Scene manager: I know it's coming but it's a very much needed tool to
> support assemblies from other CAD tools.
> 6) Pervasive Layering: a layering system where any feature can be layered.

Yeah basically these areas are coming up next. There has been some recent discussion here:
http://moi3d.com/forum/index.php?webtag=MOI&msg=1944.1


> Selection:
> 1) A poly modeling technique showing through, but when I'm sculpting by
> editing points in the UI is there a way to 'paint' a selection of points
> rather than just picking?

You can use a rectangular selection area also. You can hold down shift+ctrl when you click to force a rectangular window selection to trigger if there is not a good empty space to click to make it happen.

In general surface control point manipulation is not really a fully fleshed out area in MoI yet. It's not a current high priority because I figure that typically many of the things that classically you might want to use control point manipulation to do are actually better done these days in a subd modeler system instead. So the focus for MoI has been on more of the things that are really different than subd, like more stuff driven off of 2d profiles and booleans, etc...


> Tools:
> 1) Convert curve: to be able to convert a curve between through-points or control-points

I would like to add an option for converting a particular selected point between being a "corner" or being "smooth", I'm not sure if that would cover what you are wanting here?


> Exporting & Importing:
> input for target number of polys so you can set baselines for mesh quality.

Unfortunately this would be quite tough to implement. Under normal conditions the number of polygons generated depends largely on the curvature of the shapes involved. If I tried to work with a limited "budget" of polygons it would be pretty hard for me to spread that budget out evenly, it is more likely that some areas would get nicely formed and denser and others be just really jaggedy.

If you have a particular target number that you need, you'll probably need to generate a dense mesh out of MoI and then use some decimation type tools in a polygon-focused program to reduce the mesh down to your desired number of polys.

I don't think that it will be very feasible for me to incorporate that directly into MoI's exporter.


> 2) <..> However I did note some graphics issues where the interior
> surface was showing solid exterior lines.

Yeah, that is a fairly typical display artifact - edges that are displayed have to be pulled closer towards the eye point or else they would frequently be partially submerged in the surfaces that are displayed. One side effect of this "pull towards eye" mechanism is that if you have a model with thin walls to it, the edges from the far side of the thin wall can get pulled too far and become visible.

It does not mean anything is wrong with your actual model data, it is purely a display artifact. I have some ideas that may help to reduce it in the future when I get a chance to work on it some more. But currently it is just something that you have to ignore.


Hope this helps give you some additional info, thanks very much for the feedback!

- Michael

Hi Jason,

You wrote:
> also the export variations. Notice both were exporting to Wavefront but
> notice the archaic meshing Pro/E still uses. BLeh.

For an even bigger contrast try setting the Output: option to Output: N-gons.

One of MoI's signature capabilities is that it can export NURBS surface data to a polygon mesh structure without being forced to generate triangles in the result, even around trimmed or booleaned areas. As far as I know there is no other CAD or NURBS based program available anywhere that is able to do this. (let me know if you know of one).

Of course, your target application must support more than 3 or 4 sided polygons to take advantage of this - what is the target application that you are exporting the mesh data into?

- Michael

Strange to believe, but, my 3D efforts on various competition 3D sites resulted in my now owning BOTH the latest versions of Lightwave and SoftimageXSI.(Not bad for an old 76yr wrinckly chicken!)

Michael.

Your concept in NOT going down those @#$%^&* working sytems is spot on!

RETAIN YOUR MoI IDEALS!

Please!

Brian

{PS) TWO versions of MoI
1. Ohhhh! thats lovely!
2. Did I REALLY need this!

Hi Jason,

> In light of this though, I think once we get an object
> manager it just shows that under the hood MOI is
> capable of larger assemblies and isn't just for small products.

Yeah, sort of the fundamental mechanisms are well optimized and will be able to handle it.

Basically it needs some targeted reduction in density / quality to make it come together well for larger things.

The initial focus on small projects is kind of just getting the fundamentals in place. It will help a lot for the interface in the long run to have this initial focus on making it easy to do simple things. If that is not a priority in the beginning it does not really tend to evolve that way later on...


> Any timeline when we can start playing with an object manager?

I've only just started to work on some of the pieces for it. It will probably take a couple of months and some pieces will span across a few beta releases.

- Michael

Brian, I don't understand what you meant by this:

>Your concept in NOT going down those @#$%^&* working sytems is spot on!
>RETAIN YOUR MoI IDEALS!

Michael,

I very much appreciate the responses and your openess to the product development.

In response to:
>In general surface control point manipulation is not really a fully fleshed out area in MoI yet. It's not a current high >priority because I figure that typically many of the things that classically you might want to use control point >manipulation to do are actually better done these days in a subd modeler system instead.

From my point of view MOI is a great NURBS immersion application that subd modelers would gravitate to and having some similar workflows between the two camps would be great. It isn't so coincident that what we do in a subd modeler mimics the NURBS workflow in that at the subd level you are manipulating either the smoothed cage or raw polygons which is synonymous with curves and patches.

>I would like to add an option for converting a particular selected point between being a "corner" or being "smooth", I'm >not sure if that would cover what you are wanting here?

Hadn't thought of that, that would be good too! Any thoughts of supporting different styles of curves like bezier, cubic etc?

Nurbs are not the little daughters of Bezier's Curves?
Mr Pierre Bezier was a very cool French engineer very pragmatic astute clever ingenious! And with some sens of humor :)

Hi Jason,

> From my point of view MOI is a great NURBS immersion
> application that subd modelers would gravitate to and
> having some similar workflows between the two camps
> would be great.

I guess I'm not quite sure what the value add would be if the workflows were very similar...

I mean if a polygon modeler wanted to have the same kind of workflow that they already have, why not just use the tool that they already have?

To me it seems like it provides more value to those people if they have an additional tool that works differently from what they currently have, and is better at different kinds of tasks, etc... That way the additional tool expands their overall toolset in new areas instead of just going over the same well treaded ground that they have already covered.

But still, in the future I would like to fill out the "surface point squishing" toolset in MoI some more. It is just more that the initial focus and priorities have been on more of the construction and 2D profile driven areas instead.


> Any thoughts of supporting different styles of curves like bezier, cubic etc?

Not particularly at this point - actually the standard Freeform draw curve tool in MoI will make a cubic curve already.

The sort of classic style of Bezier curves tend to be somewhat problematic, they are really a set of smaller 4 point curves that are chained together to make them look like a longer curve, but you don't get very good smoothness between each of the pieces. This is not too noticeable in 2D only outlines but once you have surfaces and things like reflections and highlights on the surfaces the lack of smoothness becomes more noticeable.

NURBS were essentially invented to solve this problem and provide a type of curve that can have more than 4 points in it which will ensure a greater degree of smoothness interior to the curve. So adding Bezier curves would be kind of like going backwards in technology and bringing back problems that are solved with NURBS.

By the way Pilou, this does not at all mean that Pierre Bezier's work is not important, in fact in many ways it provides the foundation for NURBS. NURBS is kind of an extension that helps to solve some particular kinds of problems.

- Michael

Jason
About my "odd quote"

If you have worked with either Lightwave or XSI, the joy of having a simple UI, without also, adding scrillions of complex often illogically named dropdowns/plugins/keyboard shortcuts is why MoI is such a pleasure.

In many ways Michaels Rhino background, were it also grew like topsy, gave him the incentive to try minimising that in his MoI, very inspired, effort.

I was one of the first to get the then quite brilliant Hexagon poly modelling app. I have not used it since MoI!

Brian

I was beta with Hex and I pushed the team to have the surfacing tools like Amapi and they complimented each other. Hexagon had many of the patching tools we have in XSI, which will operate for polys or nurbs ;)

Michael, I hope to document the workflow that I was eluding to in my subd comment. And to answer your question that I missed, I tend to use Cinema 4D for most of my rendering (along with finalRender) with spurts of XSI.

Hi Jason - Cinema4D can handle MoI's N-gon output very well.

On MoI's Meshing options dialog, you should be able to set the Output: option to Output:N-gons when exporting an OBJ file to be read into Cinema4D.

If you give that a try with that Dive suit part, I think you should see a pretty strikingly clear wireframe inside of Cinema4D for the meshed translated data.

- Michael

The problem that occurs often with nurbs is that you can almost create the shape you want with nurbs but not quite. So being able to pull a few points to fix or tweak the not quite right part would indeed be really useful. At the moment it is not possible to tweak as most surfaces have way too many points to make it a practical tweaking mechanism.

However I can see Michaels view on this as well.

Hi Kevin, do you have an example of the kind of thing that you are talking about?

Normally the stuff that will work well with NURBS are objects that are much more completely defined by profile curves and do not require tweaks in a little localized area after they have been surfaced.

I mean other than combining components together. In NURBS modeling you can add a detail to a little spot like a knob or something by booleaning a structure into place there rather than moving points around.


There are quite a huge number of objects that are suited to this process... Probably a bunch that you are looking at right this moment like your computer monitor, keyboard, computer case, etc...

But if you are not able to define each component very completely by a set of 2D profiles, then it is likely that a subd type modeler which is very focused on that kind of 3D point cage manipulation would be a better choice for that particular kind of model.

For things that are 2D profile driven, doing those kinds of shapes with NURBS will make the model come together very much more quickly and accurately than trying to push points around in 3D though.

If you have some examples of things that you're having problems constructing with NURBS it would help to give you some more information on a better approach or whether it is indeed a case that should be done with subd instead.

- Michael

Michael

Here are two examples that I could solve if point pulling were available.

1.
Unable to get a straight line for the windshield (see my last post of this thread).
http://moi3d.com/forum/messages.php?webtag=MOI&msg=2020.1
2.
Unable to get a smooth transition at the wingtip from the leading to trailing edge, you get a bit of bump or flat spot where the curves meet at the wingtip. Point pulling could solve this. The bump is not as noticeable on this flat wingtip type but on WWII fighter or a lot of jet aircraft the wingtip is a little chunkier and rounder where the problem manifests itself.
http://moi3d.com/forum/messages.php?webtag=MOI&msg=1762.10&highlight=yes

Other places where it would be useful is wing fillets. It is far easier to define the shape with curves and the create the surface and then use point pulling to tweak it to final shape which you might have to do after trimming if it doesn't looks quite right.

There was also a thread recently where a user was trying to create a curved surface on top of a guiter. That would be easy to solve with point pulling.

Nurbs by its nature seems to create tensions in the surface when lofted or swept which often warps the nurbs surface to bumpy shapes. Your scaling rail option (congrats on this feature) address the problem to some extent but there are many instances when that option isn't available or applicable and bit of point pulling could solve it. This is the main reason I gave up on Rhino a few years ago so point pulling would set MoI apart if it were available.

BTW I think you should give priority to layering and object organisation as for me that is the most needed thing in the software.
The other was construction plane which you've now added to the latest beta (cool !).

Hi Kevin, sorry I missed part of your last message on the canopy thing (did you possibly edit in some more details after your initial post?), I'll head back to that thread to see what was happening there. But basically if you want to get a perfectly flat side view profile, that should be possible by cutting with a line in the side view. That will generate a perfectly flat side profile a lot easier than any point pulling would. Maybe I don't understand what you want there.


> There was also a thread recently where a user was trying to
> create a curved surface on top of a guiter. That would be easy
> to solve with point pulling.

I guess I should have clarified a bit more - as you saw from some of solutions to that, there was some point manipulation happening but it was applied to curves first and once the curves were set up there were plenty of curved surfaces created there...

But first I'll see if I can understand what you were looking for on that windshield.

- Michael

Hi Kevin, I posted some more details on your windshield thread here:
http://moi3d.com/forum/index.php?webtag=MOI&msg=2020.11

Maybe I don't understand what you are trying to get there, but as far as I can tell that is a very good candidate for a totally non-point-pulling type approach because if you want a completely flat side profile cutting the larger sheet with a line will give that to you very easily...

The key thing is to allow the cutting operation to generate that flat edge and to build the base shape to be a larger and more simple sheet, rather than trying to draw all the edges first and trying to surface around all of them only in a "piece by piece" type fashion.

If you try to surface around holes it is way more difficult, the natural way to approach these things with NURBS is to build larger sheets and then cut pieces instead.

- Michael

Hi Kevin, just one other note, you wrote:

> Nurbs by its nature seems to create tensions in the
> surface when lofted or swept which often warps the
> nurbs surface to bumpy shapes.

That kind of thing is going to happen to you a lot more often if you try to build directly to every single 3D edge curves of your final result (like what you were showing on your initial approach for that windshield), rather than building larger and more simple surfaces and using trimming and booleans to form the final 3D edges.

I know that it is a bit odd because other styles of modeling do work more directly on placing all edge loops more directly.... But that approach is not going to work well with NURBS, it is not the natural workflow for NURBS.

For NURBS to work well you need to generate broader surfaces that match a more 4-sided topology, then cut away the areas you don't need to form the final edges.

- Michael

Hi Michael

I better clarify a few things
>Hi Kevin, sorry I missed part of your last message on the canopy thing (did you possibly edit in some more details after your initial post?)
I did, but I didn't worry too much about it as I doubt anybody could help me with this one, (see my response below).

>But basically if you want to get a perfectly flat side view profile, but if you want to get a perfectly flat side view profile that should be possible by cutting with a line in the >side view.

Yes that is the only way to approach the problem with nurbs it seems and I fully understand that. The problem that I have is that I'm trying to model the windscreen so it looks like the real aircraft (F9F Panther). That means the flat windscreen shape has to look correct compared to the real aircraft when viewed from the front as well as the side. If I simply cut a line though my existing surface my windshield shape does not look correct. I then have to try and figure out what shape my curves (that I two rail swept) need to be in order for the windscreen shape to look correct when I cut a line in the side view with a straight line. Also trying to figure out exactly where that staight line needs to be in the Z axis would be a trial and error process. This is a geometric puzzle. I hope my explaination makes sense to you and you understand why I was approaching the modelling of this object the way that I did originally. If I was modelling a fantasy spaceship or aircraft of course none of these things would matter.

I'll try and post up a better wingtip example for you as it will clarify better what I was talking about.