FBX Import

[Kombowz]

I was wondering if there will ever be an FBX importer that can import the mesh and bones of an FBX file.

 

I currently use a program called Fuse combined with Mixamo Rigged character models that export into FBX and would love to animate them using A:M.

 

I love A:M, but there's been no progress on any plugins for it in forever...

 

I can understand if it's never going to happen, I just need to know so I can use something else that can handle more modern file formats.

 

The only reason why I'm asking is because I would rather not use something else... I love A:M...

[pixelplucker]

Big problem is converting sub division surfaces into usable splines, there is an obj importer and an external program called Troher that can sort of translate the quads into splines with some success. Problem comes in with non quad or nsided faces. Because AM is spline based there is a lot more information embedded into each cp which is something that isn't done with polygon models or sub division surfaces that may have information tagged to the base control shape. Recently they added the ability to trace over high poly objects with splines like retopology which works nicely and will give the purest AM model with the least amount of translation problems.

 

I used to think that the parallel theories between poly objects and AM models could make for a descent convert until I dug deeper and found that it is horrifically complex. It might be possible but not sure at what cost.

 

Alternatively BVH is pretty successful when bound to an AM Skeleton, there are some tutorials around on the forum.

[Rodney]

Hey Kombowz, its great to see you again.

 

I love A:M, but there's been no progress on any plugins for it in forever...

 

 

Good news... that's not entirely correct.

While there isn't currently a direct FBX importer/exporter several plugins have been improved and several others added in the recent past.

And while there isn't a direct path for FBX to move back and forth it's a fairly trivial matter to move back and forth using secondary utilities.

 

A lot depends on your workflow.

If you build incompatible constructs in another program they don't miraculously become compatible in A:M.

 

I was wondering if there will ever be an FBX importer that can import the mesh and bones of an FBX file.

 

In time there will be but someone has to actually program and release said converter.

Because this is a valuable workflow it stands to reason initial successes will likely be 'close hold' and proprietary for like.. forever.

In other words, a programmer usually is hired for such work because the solution has financial value. Other than diving in for the shear challenge of it all why else might a programmer release such code.

 

A constraint in this is that the average A:M User has few reasons to require such conversion. Therefore it depends on those who require the capability to invest (time/talent/money) in the solution. But there then is a key to opening doors in that where others can demonstrate the benefits of the FBX format in A:M (as a standalone all-in-one program) the usefulness of the format is shown.

Big problem is converting sub division surfaces into usable splines

 

SubD's to splines in not a 'big problem' if you subdivide to quads before exporting (or perhaps more properly stated *before importing into A:M*).

Having said that, continuity of splines can be problematic if the modeler isn't conscientious about how they are modeling so care should be taken to ensure smooth and continuous edges and loops.

 

The real issue at hand for most folks is as stated in Kombows post; that of transferring bones and motion/animation.

I haven't delved into this enough to offer much of an opinion but have seen enough to know there is a solution.

I don't have a lot of incentive to explore in that direction because I have little personal need for that solution.

The degree of success you will see will be directly proportional to your own interest in the solution.

Since you love A:M I have high hopes.

 

One thing you could do to further your success is fully document your current approach (i.e. what does success look like on both sides of the chasm), then a bridge or at least a better understanding can be built between the two.

 

 

*It's well established that A:M works best with quads so ignoring this fact will lead to frustration.

[Kombowz]

I've done some BVH work with A:M before like this:

 

MOCAP in A:M

 

Would be nice to have the rigging done with Mixamo or something like it.

 

Rigging makes me a little suicidal, LOL.

 

Maybe what I should have asked is this then...

 

Is there a way to get a BONED FBX into A:M with it's weights etc done already. I kind of need this to happen as an indie developer because I don't have the man hours to put into it with everything else that needs to be done...

 

As far as I know there's no import option to bring in Bones and Weights into A:M. This is the biggest problem I'm faced with.

 

BVH import into A:M was very easy for me, it's the rigging before that that's killing me.

[Rodney]

I've done some BVH work with A:M before like this:

MOCAP in A:M

 

I like!

Rigging makes me a little suicidal, LOL.

 

 

We tend to avoid painful processes don't we. That may be why I haven't delved deeply into rigging myself?

I keep thinking 'this is the year I will master rigging' but thus far have always yielded to other priorities.

 

Would be nice to have the rigging done with Mixamo or something like it.

 

 

I took a quick look at Mixamo but don't know enough to comment on it.

It seems to me though that it's mostly a matter of workflow.

On the one hand , *if* you could rig like Mixamo in A:M I assume you wouldn't have much need for Mixamo.

If on the other you could export directly from Mixamo to A:M you'd have a more direct workflow.

At issue then is the incompatibility between the two rigging approaches.

BVH provides a solution but my initial foray into that area met with blank stares.

Folks seemed content with starting from scratch every time which doesn't make a lick of sense to me so I shelved that pending additional R&D.

 

My holy grail of rigging (and likely disconnect with reality) is a set of images from a book published in early 1900s that outlines 5 points of primary control/articulation on a character. In the little look I had at Mixamo they seemed to be oriented at least a little toward that approach.

 

You may find yourself to be the pivotal part of your own solution in that if you can express the elements of rigging that keep you from reaching your goal (i.e. the painful vs ideal workflow) you may inspire a solution that improves beyond your original goal. What I'm suggesting here is that what you want isn't really a FBX converter/plugin but a more effective way to rig and animate your characters. At least that is what I perceive so feel free to correct me where I'm off. File formats are like hardware in that the closer someone programs to them the less compatible they are with where we really need to go.

[Kombowz]

 

 

*if* you could rig like Mixamo in A:M I assume you wouldn't have much need for Mixamo

 

That would be nice and more than likely easier for the Dev of A:M to accomplish

 

for example

 

I have a OBJ that imports fine, but has no bones...

Auto assign points works well if the bones are positioned well, but takes FULL control over each cp so no CP is shared by any two bones.

IF there were some sort of parameter that you can set the strength of CP sharing amongst bones for CP inbetween them, that would be perfect!

Like a sphere that's an inch around. the center of that sphere would be more shared then the outer part of the sphere. and you can make the sphere as big or small as you like per joint. maybe before selected auto assign?

I believe this is exactly how the mixamo auto-rig works based on looking at the rig setup and parameters in Cinema 4D

 

Can this be implemented into A:M perhaps?

[Rodney]

IF there were some sort of parameter that you can set the strength of CP sharing amongst bones for CP inbetween them, that would be perfect!

 

 

A:M has that sort of parameter adjustment in at least two varieties: Bone Falloff and CP Weighting.

 

Do you have the most current FBX Converter from Autodesk installed (from 2013)?

Short of paying for some other solution, that's what you'll want to have in order to get a proper quad based mesh/OBJ.

[Kombowz]

Well, I'm using Mixamo Fuse to create models, then it exports as an OBJ.

 

I don't mind putting bones in the character all that much, it's just getting it to move nicely in-between those bones.

 

The OBJ created seems to look just fine. I don't think I will need an FBX converter as it won't help me with the FBX that is produced with the Mixamo Auto-Rigger

 

I will try experimenting with the bone fall-off and CP weighting properties to see if this will help me any.

 

It's really a shame that something so innovative such as spline based modeling never got the recognition that it deserves...

 

Have you ever tried modeling with polygons? What a nightmare...

[Rodney]

Have you ever tried modeling with polygons? What a nightmare...

 

I'm not a purist so much as a realist and don't have issues with polygons where understood properly.

And if those models you create with polygons won't move or be distorted in any way... they'll work really well imported as Props.

There are still a few issues with texturing those (some transparency settings don't appear to operate properly) but for the most part those are largely resolved.

 

Polygons are easier to manipulate at the lower end of the scale but inversely more difficult to manipulate at the higher end (esp. when animated).

I suppose the reverse could be said concerning splines as splines are less beneficial in simple static or non-continuous constructs (there is an inherent tension with spline based patches that doesn't exist with flat planar polygons). I equate this to the difference between bitmaps and vectors; bitmaps work quite well except where they do not. As do vectors.

There are classic tricks used to suggest the movement of static images such as the replacement of one static image for another in rapid succession but the real promise of automation (and computer animation) is to move beyond that into actual movement of objects. It's taken 100 years to approach that type of animated movement outside of puppetry and stop motion.

 

I don't think I will need an FBX converter as it won't help me with the FBX that is produced with the Mixamo Auto-Rigger

 

I've mainly used the FBX converter as a conduit to transform/subdivide triagonal meshes into quadrilateral meshes (the one being useless to me while the other is not).

I'm sure there are strictly OBJ converters that can do that type of conversion but I've yet to find one that produces meshes in the same way that an interim pass through another file format such as FBX does. But note that Autodesk's free FBX converter supports both tris and quads so one must tell it to override the default which is set to spit out triangles. (the good news... in the interface it's a simple checkbox)

[Kombowz]

I will try the FBX converter anyway to see if it makes the model better/worse

 

What will the FBX converter do? convert the FBX to OBJ to be imported to A:M?

 

I will have to test it out when I have time, but a brief use for it would be helpful.

 

I appreciate your help with all this, thank you.

[Rodney]

What will the FBX converter do? convert the FBX to OBJ to be imported to A:M?

 

I've used it for several conversions.

Of course the first sort is FBX to OBJ.

 

A less obvious usage would be the OBJ to OBJ conversion that converts tris to quads.

The problem with approach this is that the FBX converter doesn't directly convert OBJ to OBJ (that's not an option) so an interim/extra conversion is required.

You have to convert OBJ to FBX and then FBX back to OBJ (but with the 'triangulate' checkbox unchecked/off).

 

There are a couple of programs I tested out (MakeHuman and... ah... I forget the other one but it is a lot like MakeHuman).

Here's a topic where I posted a few models and others plussed up the results:

 

http://www.hash.com/forums/index.php?showtopic=45450