After seeing the tedious ( but very awesome ) workaround that had to be created in order to get the FaceShift Plugin working in VVVV i was just wondering if there are any plans for fbx support.
The reason im opening this discussion is that ive created my first gaming character with working Bones - even for eyes. I have created 45 custom scuplted facial expressions and teeth in Zbrush and it would be WILD to have an easy import option.
All of these can be exported in FBX for Unity and every parameter / transformation works perfect even with faceshift + body bone animations.
If there is an easy way to this using collada or Assimp let me know :)
In my experience loading time for Collada is really awfull and for both Object and Collada i always have to switch between many versions of 3dsmax / autodesk FBX Converter / LithUnwrap / MeshLab etc to get anything correctly imported in VVVV. In assimp 3ds files do not work for me.
Maybe im wrong and someone has a great idea how to import blendshapes other than using lots of x.files and vertexbuffers.
I know that each 3D Format is tricky and requires different settings for different programs, even .FBX can act completly different in a different software.
Blendshapes/Morpher/vertexbuffer Animation are quite a powerfull animationtype, not just for facial but also abstract animations.
The current situation makes it very complicated to export import such animations, therefor you cant really preview things & change them easily in a 3ds software to reexport them.
It really makes the workflow complicated and prevents experimentation and for me thats what vvvv is all about :)
someone should ask me if i would like some cheese with my whine.
We are working on a similar project, and FBX would be the weapon of choice for us too. Some research tells us, that it is a very complex format with many versions and inherited approaches … So it won’t be a quick thing to do.
Yep, FBX is THE standard right now, I think. Exporting collada or .x from Maya is a huge PITA. I tend to use polytrans for this, but that’s not a nice workflow. Really.
the fbx sdk is c++ as i know was the only problem, but honestly thouse formats are pretty much same as collada. u still not gonna do morph on cpu and pointcash is not that hard to implment…
@antokhio
You are right! The Collada format supports export and import of “Deformations” ( blendshapes, Morphershapes) its really not a very compatible format anymore.
After trying several import and export approaches it will skrew up my Charakter in one way or another. Right now it will destroy my UV mapping and IDS.
While trying to load Collada’s in VVVV nodes will blink red for a second, then wont load any kind of mesh untill i kill the node Collada or Assimp node and reload them / Restart vvvv.
The file Size is much higher and again, the loading times are scary.
I still carry around 3dsmax 2012 - 2015 in 32bit and 64bit, simply because its that much a hit and miss situation for me. :/ Using converter tools i find in 10 year old Planetquake forum posts etc is just really frustrating.
Found this post describing the same issues i have with the format. The format itself is fine but new software packages dont seem to support it properly:
yea that part is hell in 4v, the max biped also in max axis plane so it’s ok for baked animation, and not if u want to drive it. also i used 2008 maya, u can see build name in astroboy example…
the most part of characters stuff works in dx11, but there is no examples yet, and vux still have it in the state of implementing
Didn’t test latest assimp build yet, but seems it supports fbx so that would make things easier than using fbx sdk (even tho from what I’ve seen it’s more complete, but if assimp can cover standard features it would be a good start).
Since I suck at 3d modelling, if someone can post the following:
One 3d object (with proper normals/uv and a texture)
Maybe it would be good if people provided these examples to Vux from several 3d packages? Seems part of the problem with 3d assets > vvvv is that there are slight differences between exporters
Maybe a bit offtopic but i got this .dae file working with collada loader in dx9 but doesn’t work with assimp node in dx11…
File was exported with the last blender, any idea to make it work in dx11?
i can provide a Character but i can not share it public, its an ongoing game project so it would be great if its just for your testing. If your Ok with not sharing it and only using it for testing i can give you a link via skype / email :)
It consists of 5 Meshes ( Eyes, Teeth, Hair, Body)
The Body and Head are 1Mesh but have seperate Textures and UV.
Character has been rigged with http://www.mixamo.com which might be interesting for a lot of people.
Added 2 Eye Bones to the Mixamo Rig to work with Faceshift.
Has 44 facial expressions in Morpher/Blendshapes that work with faceshift which works well in unity.
I have one Version exported for Faceshift and one for Unity. Both want their own export Version. Unity always wants the uptodate version. 3DSmax 2014-15
ok, i did some research and talks about the fbx / assimp stuff. please correct me if i am wrong but as far as i know now assimp do not use the autodesk sdk at all because of some licensing issues. so they wrote an own fbx importer from scratch.
in my understanding this is not really what we want because we dont know which feature level is supported right now and how. And we also dont know if it ever will make use of upcoming features. … so you have to deal with two sides - one is the fairly complex fbx format itself - the other is the custom implementation through the assimp library. At the end this means to me that we would always hang on that custom development someone started and maybe hopefully will taking care of it in the future.
i would say the only way to go for future save support of the industry standard would be to start wrting a managed wrapper for the autodesk sdk. therefore we have to define the first basic features (such as mesh loader, bones, blendshapes, camera …) as well as a standard export software because every company such as autodesk or maxxon uses different approaches in building up a model and a scene. And of course we could also talk about some funding for this project.
edit: antoher thing to research is the licensing behind the sdk - is it free to use in all cases?
