[Bf-committers] Importing Assets: FBX VS Collada

[Stefano Corazza]

I'm compiling the key responses on the topic, thanks for the great 
feedback/comments. Here also some reality checks.
Sorry this is long but i'm trying to wrap the topic up.

>>Re: baking animation into meshes instead of using a rig:
>>This might work for uses related*solely*  to still picture & animated film development, but it does pretty much strand the game development folks trying to use Blender as an option in >>their pipelines. Animated armatures with skeletally attached meshes being imported and exported into/out of Blender is kind of essential for this area of Blender use.
>>Benjamin Tolputt


Benjamin well said. Game development CANNOT use baked mesh animation. 
Also for film it is extremely inefficient and a memory sink. The future 
is a cross compatible rigging system. And it is already there for FBX 
and Collada (every Autodesk software exports proper Collada!!)


>>cool to hear that FBX maps to Blender well. If it's a sane format perhaps it is ok for it to become standard .. by Blender folks reverse engineering and documenting it :p . Even if it's >>not publicly specified by Autodesk I suppose they can't break it as then many apps with support for it would break .. unless they'd require always using updated versions of the SDK.

Autodesk will NEVER release FBX SDK openly. I tried for years to 
convince them. They are a public company, so they cant do things that 
weaken their position in the industry. I talked to top management many 
times around this so make peace with it. Many apps DO NEED to upgrade 
FBX SDK every year or every couple of years as they are breaking the 
workflows at every release. Unity3d, Unreal, Maxxon, Mixamo, etc. we ALL 
UPGRADE FBX SDK every year of couple of years. The effort of reverse 
engineer FBX SDK in Blender every year is a waste of resources in my 
opinion when Autodesk already provides a FBX <-> Collada conversion.


 >>  a lot of game makers use special engines with own formats and are 
quite happy to directly use Blender python for export.
 >> For game makers we should keep fbx to work too - but even better 
would be if they start directly supporting .blend files. I'd love to see 
efforts for that.

Every decent game engine supports FBX import/export. The time of 
import/export plugins is GONE. Even for Maya and Max it is getting 
harder and harder to find them.
With EVERY game engine importing FBX and with a FREE and maintained FBX 
<-> Collada converter out there it seems obvious to me that Collada is 
the way to go.
Especially when you have a import/exporter that is already 90% there.


 >>Ton is asking why Mixamo does not create the new importer:
 >>Then why not do it for 2.7x? That import API is there since 2009. You 
are really welcome to pick up this development job and become a module 
owner for the Collada importer. It's the >>stakeholder principle. If you 
depend on a feature to work, you can also help it to be realized. If you 
- on the other hand - just seek for volunteering help, that's OK too. 
But I know Collada is >>incredbly impopular to work on. If you look for 
someone to be contracted for it... you could ask Gaia Clary. -Ton-

A negligible fraction of our paying customers is using Blender. Why? 
Because you cannot reliably import assets from any other 3D 
package/service!! So we are not a "stakeholder" in the strict sense but 
would love to help the students/hobbyists/indie community that is using 
Blender. Since you changed Python version with 2.6x and later we would 
have to rewrite from scratch the plugin. We only have  a small team 
engineers so we can't do it. The best option is to finish the job on the 
current one that is almost there.

What I can offer is test cases and QA service for the last 10% of the 
Collada importer/exporter fix (support of multiple meshes/bind poses) 
and Mixamo stocks as payment.

If anyone is interested please contact me directly stefano at mixamo.com

I hope you appreciate the passion for solving the problem and the amount 
of real industry information on the topic. Establishing a standard is 
not an easy job. But we are almost there!!

Long life to Blender and the Blender community!!

Cheers

Stefano