Okay, I tried something... If you hold down the left mouse key to grab and move an object in perspective view, you can simultaneously hold down the mouse-pan button, go to the opposite side of the screen (with the object now moved to the opposite side of the screen) and pan the view without dropping the object.
Then you can mooooove your object back over and continue.

So I know what I'm thinking about might work, it would just have to know you're going "over there" too without having to click and hold the mouse pan button.

Hi Mike,

> So, to clarify. (in theory), the object and cursor would stay in the intended
> view (by which the mouse key was held down) and the viewport would pan
> in that relative direction - instead of crossing over into the next view and
> jumping the object somewhere else.

Yes, but that "jumping" part can be a useful feature that would then be eliminated with the method that you're describing here.

Currently if you want to grab say this back rectangle here and drag it upwards in elevation, you can grab the rectangle in the Top view here:



Then all in one motion without letting up the mouse you can move into the Front view and be adjusting it in elevation here:



That can be a convenient way to adjust elevation - if things worked the way you are describing this kind of multi-viewport dragging would not be possible anymore.


> well, I don't know, is bouncing around with the same object from one view
> to another a common workflow practice? :-/

Basically the split view exists in order to do things generally similar to this like selecting things in the top view and then making adjustments in a different view for elevation.

There is similar functionality in drawing as well - when you draw a curve or a polyline for example you can switch to a different view and place a couple of points in elevation so that you can more easily get a full 3D shape all in one go rather than the drawing being 100% locked to only the first viewport that you clicked the first point in.

It also seems like it would be kind of weird if object dragging was restricted to only be locked to the first view you clicked in but other commands like transforms (move/copy/array/etc) or drawing commands behaved differently and did not lock things or auto pan in the same way.

- Michael

So basically since MoI is not just 100% focused on 2D drawing but also manipulating things in 3D it can sometimes be hard to exactly replicate 2D workflow like you're asking about in cases where it would cause some degradation in being able to do 3D multi-viewport manipulations.

Aside from dragging, just the basic act of drawing it can be really useful to be able to pick the center point for a circle in one view and then pick the next point in the front view to produce a circle that has its center in some spot that was not easy to pick in the front view but still is oriented flat to the front view because of the second pick going there.

Basically this kind of combination of some picks in one view and further picks in another view can help out in placing things directly in a 3D position in many kinds of situations, the sort of "lock all mouse activity to the first view you clicked in" type mechanism like you are describing would have a side effect of removing all of that current multi view placement abilities.

- Michael

Well... thanks for the explanation. I knew there'd be some good wisdom in the way it was working now.

But at least we have the forum to hammer these ideas out. ;-)


To tell you the truth, the more and more I use MoI, the more I'm preferring the use of the Transform tool set.
As the actions I require become more concise, so - the need for exact placement.

Hi,

It's about quads export again. I know it would be a lot of work to do it in MOI but what about an external program that would do a great job at it in an as automated way as possible. I know about 3D-Coat and TopoGun a little. The later being mostly manual and 3DCoat as an automated means but the result often (if not always) requires additional editing.


Edit: Though I've found quite a lot of research papers on the subject, there seems to be a lot less application implementing these kind of algo.


Thanks,
Felix

Not sure why you want quads only export. NGons work fine in most applications, even Blender with the latest release. Unless you want to animate. But why would you character model in Nurbs? You are better of using a polygon modeller for these sort of things

Hi Felix, re: all quads - as far as I know 3D Coat's automated conversion tool would be your best bet. I'm not aware of any other automated processes right now except maybe in Finite Element Analysis software which does analysis of object structures for strength and such. I think some of them have some quad meshing mechanisms in them but that software tends to be extremely expensive.

> The later being mostly manual and 3DCoat as an automated means
> but the result often (if not always) requires additional editing.

Getting a nice "all quad" structure involves making good choices about edge flow topology which is something that comes from experience and judgment. It's very difficult to make a totally automated process to do operations that require judgement like that.

Usually like Steve says if you need an "all quad" output you should be creating it that way from the start in a polygon modeler - using MoI for that kind of specific target output is basically not using the right tool for the job at hand.

- Michael

Michael,

I search the net for this but as you say the subject is vast and most research on the subject are still just that. And yes, one of the export cmd of 3D-Coat give pretty good result.

@Steve:

if I was a decent subD modeler I'd do just what you suggested but I'm not. What I'm trying to do for now is I try to create a pretty basic NURBS version of the model in MOI, export it as a dense quads and tris to either Blender or 3D-Coat and I try my best to add the details I want. It's seems to be easier with 3D-Coat, at least for me.
Here is an example.











I should have mentioned these models are intended for CNC.

Thanks,
Felix

Hi Felix - if you're adding details in 3D Coat and you're using its voxel editing mechanism, it doesn't actually make any difference if you have all quads or all triangles - when you import your polygon model into 3D coat for editing using the voxel toolset it gets converted from polygons into a voxel structure at that time, and it does not matter if you have all quads in the original polygon structure or not.

You would only want to go to "all quads" if you were going to be using a sub-d modeling toolset to do further work on the object.

Sub-d smoothing is the thing that particularly likes quad topology - voxelization does not care about it.

- Michael

Hi Felix, also is that a ZBrush screenshot you're showing there?

You don't need to have all quads to go to zbrush either - you can send a finely diced up mesh and then turn off the "smoothing" modifier inside of Zbrush and it will then not try to use the mesh as a sub-d cage and instead just displace the polygons directly. See here for an example:
http://moi3d.com/forum/index.php?webtag=MOI&msg=804.26

Zbrush has a setting in its subdivide method for whether to use sub-d smoothing during the subdivision or not - you want to turn that off.


Are you really sure that you actually need "all quads" in your output? Can you maybe give some more detailed information about what processes you use on the mesh after you export it from MoI? Because if you want to use those brush based deformation tools on it, that should be possible to use without needing special sub-d oriented "all quad" topology.

Have you possibly seen some tips from people saying that all quads is an absolute necessity for everything? The thing is, that only actually applies to one particular workflow of doing actual sub-d modeling, it does not apply to other situations.

- Michael

Michael,

I agree with you for both 3D-Coat and ZBrush quads aren't necessary. For Blender though it's another story and I assume it could be the case for many other programs. There are also another kind of program that convert the mesh to a bitmap, these don't need quads either but editing these in the program is somewhat limited, at least those I've tried.

I know it not your problem but I have a hard time learning new stuff and it's the reason I'm using as much as possible programs I've used a long time ago like Blender. It's kind of the bicycle thing, ounce you learn it you don't forget it.

I tried the trial version of ZBrush and I'm useless with it, it's way to different then anything I've used and when it ended I had not made any significant progress with it so I didn't buy it.

The image I provided is a screen capture from 3D-Coat (trial version). This one seems more user friendly to me at least and maybe I'll buy it. The main reasons would be that quads aren't necessary and hopefuly, I'll be able to learn enough to use it for my needs.

Thanks,
Felix

Hi Felix, what particular tools are you using in Blender - you can do things like sculpt mode in Blender on triangle meshes and not only quads as well.

If you're doing sub-d modeling, then yes you need to have a sub-d friendly mesh structure for that and typically that kind of structure just does not get created for you automatically, you will be looking at doing retopology for that.

- Michael

>Hi Felix, re: all quads - as far as I know 3D Coat's automated conversion tool would be your best bet. I'm not aware of >any other automated processes right now except maybe in Finite Element Analysis software which does analysis of object >structures for strength and such. I think some of them have some quad meshing mechanisms in them but that software >tends to be extremely expensive.

Hi Michael and Felix,

here's a free Windows software which produces quad meshes. Input format is .ply

http://www.it77.de/studienarbeit/index.htm

Regards
Stefan

<< here's a free Windows software which produces quad meshes. Input format is .ply

so meshlab will be your save life :)
http://meshlab.sourceforge.net/

.

Michael,

the problem with Blender is that yes I can sculpt and or edit an imported STL for example but to get a nice and smooth model the mesh as to be quite dense and unfortunatly even with an 8 core cpu and 6 gig off memory Blender become quite sluggish, basically unsusable.

Also, yes retopo would be necessary to do something useful with Blender. Fortunatly for me, programs that translate the mesh to something else, either voxels and or bitmap are not as fussy about the mesh structure and size.

Though I can understand the major difference between Sub-D and Nurbs I would have thought the transition between the 2 would have been made much easier by now. Unfortunatly for me, my goal is to machine the model I would create. The mesh needs to be dense enough to be machined smoothly, especially the small parts. I understand also the quality of machined model is dependent on the CAM software and the macine itself.

Thanks,
Felix

Hi Felix,

> Though I can understand the major difference between Sub-D and Nurbs
> I would have thought the transition between the 2 would have been made
> much easier by now.

The problem is that sub-d requires a particular kind of topology and edge flow which is usually implemented by a human using their judgement and experience on how to arrange all the elements.

It's quite difficult to replicate processes that require judgement like that since they're not like checking off a list of simple yes/no type decisions like a computer algorithm is fundamentally based off of.

Retopologizing toolsets are probably what you would need to be focusing on if you need to do this kind of conversion.

- Michael

Michael,

yes again, retopo is the route one as to take now and there are good tools for doing just that out there. You'll probably agree with me that an even better route would be to start directly our work with a SubD program and bypass completely the retopo and or conversion phase. IMHO it would be much more efficient.

Out of curiosity, to your knowledge, is the route from SubD to NURBS somewhat more efficient? I just love using MOI and I'd like for example to be able to add a part made as a mesh from the start, say the "flowers" I made earlier to a mirror frame made in MOI. I already know I could convert the mesh to a heightmap and use ZSurf to get a Nurbs surface. Though it would probably work fine, it's not what I would call an efficient route.

Thanks,
Felix

Hi Felix,

> Out of curiosity, to your knowledge, is the route from SubD to NURBS
> somewhat more efficient?

It is, but going in that direction is a relatively new area of tech and it's not very common yet.

But there are 2 tools that can move things in that direction - either the T-Splines plug-in for Rhino (http://www.tsplines.com/) or the Power SubD-NURBS plugin for Modo (http://www.luxology.com/store/Power_SubD-NURBS/) can convert a SubD control cage to smoothed NURBS surfaces.

- Michael

Michael,

thanks for the info. TSplines seems awsome at first glance, it looks like it brings the best of both WORLD into a single uniform environment. I kind of like that approach. As for Power SubD-Nurbs, it seems more like a mean to translate SubD to Nurbs which is nice as well. Different approach and workflow, I guess it's a question of personal preferences.

Thanks again,
Felix