Hi Brian, yeah the naked edges that Jonah shows are areas where your different surfaces don't touch closely enough to be joined. There are some spots where they have a bit too much of a gap between them, and other spots where they are pushed a bit too far in the other direction and overlapping each other too much.

For example, here is one such area:



Usually the best way to solve this is to get the curve framework all tightly positioned first, before creating surfaces from them.

Do you still have the curves that you used to construct this object? If you have those, please post them, editing those would be probably the easiest way to tune it up.

- Michael

Hi Brian, I just went ahead and took some of those edges and drew in some new curves to tune this up for you. Take the attached Nose02_tuned.zip, select all the curves, and run Construct / Network to build one surface through them.

- Michael

Thanks Jonah and Michael.

Michael.
1. How would I ever able to "pre determine" the "faults" within my work that you show?
Do I allways have to find out what does not work and then go back to fine tune analyse?

2. Your zip file looks brilliant---but, sorry, how the @3@ do I utilise it effectively?
Do I sellect all and just go "networK"?

Mumble, mumble---it's a brian day?

Answering some of my own questions!----
(2)--yes.
BUT that adds the questions of the detail of the whys and hows of the specific "modifications" you made to my original file?


Brian

Hi Brian,

> 1. How would I ever able to "pre determine" the "faults" within my
> work that you show?

Well, any time that you just sort of free-form position 2 pieces nearby each other without a snap, that will tend cause this kind of problem.

So unless you have snapped things together, it is best to assume that these faults are present.

The way to avoid them is to use snaps to ensure curves are locked in and touching each other before building surfaces from them.


> Do I allways have to find out what does not work and then
> go back to fine tune analyse?

No, it is much more efficient to use snaps from the beginning.


> BUT that adds the questions of the detail of the whys and
> hows of the specific "modifications" you made to my original file?

Well, I did quite a bunch of little edits. I went through and duplicated some edge curves from your original file so that I would have a shape closely matching it, then I dragged pieces that looked like they should be locked together with "end" snap so that they were exactly locked together, I used Edit/Trim to snip off a few ends that were kind of shooting over an axis line so that the curve stopped right on the axis line, I turned on control points and used Transform/Align to line points up that were wandering around a bit.

I did all of this to half of the curves, then used mirror to make the other half.

To smooth out the kink between a few joined pieces, I used Edit/Trim to cut pieces back from the kink area to make some space there, and then used Construct/Blend to put in a smooth piece. Several surfacing commands such as Network can build one surface if you have a smooth curve, but will build a creased multi-surface output if your curve framework also has a crease (sharp corner) in it. Also another way to smooth out a crease is to use Join and then turn on control points and select and delete the point where 2 segments touch one another. I did that in a couple of spots as well.

So that was a lot of little edits kind of wrangling things, aligning bits that were wandering around instead of all in one plane, smoothing out junctures between pieces, trimming off excess bits that were sticking out beyond an axis line, ...

- Michael

"any time that you just sort of free-form position 2 pieces nearby each other without a snap"

Sorry, right there, I am lost.(Thinking back, I can not think of a logical step I took incorrectly from what I have learnt/done previously?)

Could you please diagramatically detail the difference betwen the two. What I did wrong and what I should have done!

Brian

Hi Brian, do you have any older versions of this file that you could post? It might help me give you some more details about what stage had the problem.

It's a bit difficult to tell exactly what went wrong by looking just at the final result. Ideally I would watch you as you were constructing it, but of course that is difficult to do.

Here is some illustrations of the kind of thing that I'm talking about though.

Say for instance you have these 2 curves and you want to position them next to one another:




One way you might try to do it (this is the bad way) might be to grab just at some spot interior to the curve like this:



And then drag it upwards like this:



Continuing upwards until it looks like the ends are pretty close to each other, and then releasing it there:




That's the bad way - the position of these 2 objects are just "eyeballed", there wasn't any end-to-end snapping kicking in which would ensure that the ends are definitely locked together. They just look pretty close but looking close from a zoomed out position is not the same thing as locked together - it will cause things like gaps between surfaces that are constructed from these curves later on.

So instead of eyeballing it, you want to use snaps that target the specific points you want to lock together. So for example in this case, you want to snap end-to-end.

So when you start to drag the bottom curve, instead of grabbing it somewhere near the middle of the curve, grab it nearby the end, like this:



When you click down and start to drag from that position, you will see an "end" snap highlighted on the curve, like this:



That means that you grabbed a snap point and you are going to be specifically placing the end of the curve somewhere. Now you can move to the end of the other curve, and it will show an end snap there at the target location as well:



That's how you know you have placed 2 curves exactly snapped together precisely end-to-end with no gaps between them. Then when you create surfaces from those curves, those surfaces will touch precisely as well instead of having empty space between the surfaces.

- Michael

Thanks Michael.
I think I went through very single thing I could originally.
I mean everything-- as you described---I had hoped I was learning!
It's late in the day any I am really frustrated,

Maybe I need to wait till Friday and start again from scratch?
(Or Monday week!)

However could you, please, doubly confirm that this was you response to that sentence I queried.(I am having a thick day maybe)
Brian

Hi Brian, that last one with the diagrams/snapshots was meant to illustrate this sentence that you had quoted:

"any time that you just sort of free-form position 2 pieces nearby each other without a snap"


In the example I show 2 curves, and the idea is to position them so that they touch end to end.

The first method is the "free-form" or "eyeballing" type method. This is the method that does not use an End object snap. It is the bad method - it is the way that can cause problems because it is just moving things with the mouse until objects just sort of look like they are close to each other. But if you zoomed in on them you would see that there is some space between them. That is a lot different then snapping them together to ensure that they are actually exactly touching end to end.

The second method that uses End object snap is the way to position 2 curves in such a way that they definitely touch each other instead of just being pretty close to touching.

Just moving things nearby each other is not generally good enough, especially if you want to start creating surfaces from those curves. The resulting surfaces will have gaps between them like the screenshot I showed earlier here: http://moi3d.com/forum/index.php?webtag=MOI&msg=1604.4

If you use the second method that uses the "End" snap, that will avoid that kind of gaps-between-pieces problem.

- Michael

Thanks Michael.
At about 3am this morning I thought of something that may have been involved. Later today I will try and emulate that oringinal path and compare with your advice.
Busy day and its only 10 am. XP SP3, Carrara6.2 update---what next?
Brian

Michael
Not a very helpful couple of hours. I wont bore you with a lot of the other stuff.
I think, in the main, the problem relates somewhat (maybe/perhaps/inpart/also!!!) to the relationship between MoI and Carrara.

Here is that file you made exported as 3dm and obj from MoI.
!

Anyway, A shows problems in Carrara doing what I wanted to do.
B shows the same file as exported from MoI as an obj file.

MMmmmmnnmmm!

Brian

Hi Brian, I'm sorry I'm not really following you here.

Is your shot A showing problems in Carrara trying to do displacement type editing?

Maybe Carrara needs to have a more finely diced up mesh with smaller polygons to do a good job of that. You might try using the "Divide larger than" setting in MoI when exporting to divide polygons up into smaller pieces.

For example if you try putting in 0.1 as the "Divide larger than value", that will force polygons to be divided until they are no larger than 0.1 units, that type of dense mesh may work better for displacement type operations inside of Carrara.

- Michael

Michael the wierd thing in A is that there are different mirroring responses (soft sellection poly movement) on either side--left fine, right daggy!

Hi Brian, does it seem to behave better with a denser mesh?

If it is a problem with Carrara's movement tools, I'm not really sure what I could do about that. You might need to ask about that in a Carrara forum to find out more about that specifically.

But if you find out that Carrara would prefer to operate on a certain style of mesh like lots of little polygons (instead of a mix of some larger and some smaller sized polygons), then there are settings that you can adjust at export time to create something like that.

- Michael

Hi Brian,

>>the wierd thing in A is that there are different mirroring responses

I do see the same problem. But that is down to the 3dm importer in Carrara, it is not good.

Personally, on that model, I would output an obj file with angle of 3, and set the output as quads/triangles.


- manz

Hi Brian. I'm not sure what's the point to model this in Nurbs if you plan to sculpt in polygons anyway. You could have modeled a nose in subd in just a few minutes. And the toplogy would be vastly superior to what you achieved by all this extra work. Unless the purpose was just for study???

jonah

Hi Brian, ok now I understand that you are comparing importing .3dm files versus importing .obj files into Carrara.

When you import a .3dm file directly into Carrara, you are relying on Carrara's meshing code to convert NURBS surfaces into polygonal facets. It does not look like it did a very good job of that conversion in this case.

When you import an .obj file into Carrara, you are instead relying on MoI's meshing code to convert surfaces into polygons. MoI's conversion mechanism tends to work better, that's why you get better results when doing it that way.

- Michael

Sorry I was away. Yes, I was just reporting another queer thing.
Jonah. I was looking at an experimental exercise about taking a MoI file into the Carrara Vertex room and using the Displacement Paint.

With SR3 and the Carrara updates etc I wont be able to follow thus up with examples for a while.

Brian

Michael.
The attached shows the import adjustment settings available when importing 3DM files into Carrara.

Can you (or anyone) make any suggestions on where these settings should be set to improve the quality of the resulting files in Carrara please.

(I really don't know what the settings are designed to do/adjust! ---NO CLUE!)

Brian\

ps I presume I am right in that no adjustment to export settings for 3DM can actually be done within MoI?

Hi Brian,

The main setting to change is the "max angle", as that will control the amount of polygons produced over a curved surface.

As example:

Importing a sphere with default settings (20) will give a mesh:-



Changing the "max angel" down to 3 will produce a mesh as:-



I would still advise you to export directly from MoI (to obj) and make needed setting for export there (angle etc), it is certainly better (I find)

Maybe we need to start a thread with examples of exporting?


- manz

I agree, unless you're having problems with OBJs exported from MOI, you should just stick with MOI OBJ exports.

What's the advantage or your reasoning for working with 3DM imports? Unless this is just a test to see how Carrara behaves working with .3DMs?


The only reason I've ever attempted to load 3DM (in XSI I believe), was to try and keep the model in the NURBS domain - if it's going to be converted into polygons anyway, you're then stuck as Michael said with the quality of the target app's importer.

Frankly I'd rather export from MOI and point any problems to Michael (who has a way better history of fixing his code) than now depending on DAZ to fix any issues with their mesher / code...

-Will