atom feed95 messages in [Bf-committers] Collada importer/...
FromSent OnAttachments
10 earlier messages
Domino MaramaJan 5, 2012 2:55 pm 
Ton RoosendaalJan 6, 2012 2:58 am 
Gaia ClaryJan 6, 2012 3:42 am 
Sergey SharybinJan 6, 2012 3:48 am 
Thomas DingesJan 6, 2012 3:51 am 
Campbell BartonJan 6, 2012 4:08 am 
KnappJan 6, 2012 10:51 am 
Morten MikkelsenJan 6, 2012 12:20 pm 
Francesco ZoffoliJan 6, 2012 4:04 pm 
skotiJan 6, 2012 5:11 pm 
Morten MikkelsenJan 6, 2012 5:34 pm 
Przemyslaw GolabJan 7, 2012 12:41 am 
Ton RoosendaalJan 7, 2012 3:30 am 
skotiJan 7, 2012 5:03 am 
Morten MikkelsenJan 7, 2012 6:49 am 
Juan LinietskyJan 7, 2012 7:06 am 
Morten MikkelsenJan 7, 2012 7:38 am 
Juan LinietskyJan 7, 2012 7:49 am 
Juan LinietskyJan 7, 2012 7:52 am 
Juan LinietskyJan 7, 2012 8:22 am 
Morten MikkelsenJan 7, 2012 8:29 am 
Domino MaramaJan 7, 2012 9:04 am 
Erwin CoumansJan 7, 2012 9:07 am 
Juan LinietskyJan 7, 2012 9:10 am 
Erwin CoumansJan 7, 2012 9:23 am 
Morten MikkelsenJan 7, 2012 9:47 am 
skotiJan 7, 2012 10:16 am 
Gaia ClaryJan 7, 2012 10:34 am 
KnappJan 7, 2012 10:34 am 
Erwin CoumansJan 7, 2012 11:15 am 
Alexandr KuznetsovJan 7, 2012 11:33 am 
Juan LinietskyJan 7, 2012 12:04 pm 
Juan LinietskyJan 7, 2012 12:06 pm 
Juan LinietskyJan 7, 2012 12:08 pm 
Erwin CoumansJan 7, 2012 12:13 pm 
Sergey KurdakovJan 7, 2012 12:32 pm 
SebastianJan 7, 2012 1:22 pm 
Juan LinietskyJan 7, 2012 2:34 pm 
Morten MikkelsenJan 7, 2012 3:09 pm 
Juan LinietskyJan 7, 2012 3:24 pm 
Morten MikkelsenJan 7, 2012 3:36 pm 
Campbell BartonJan 7, 2012 6:45 pm 
skotiJan 8, 2012 1:57 am 
François T.Jan 8, 2012 3:24 pm 
Juan LinietskyJan 8, 2012 3:31 pm 
angj...@gulftel.comJan 8, 2012 4:34 pm 
Juan LinietskyJan 8, 2012 7:16 pm 
Campbell BartonJan 8, 2012 8:39 pm 
Juan LinietskyJan 8, 2012 9:37 pm 
Kalle-Samuli RiihikoskiJan 8, 2012 9:55 pm 
35 later messages
Subject:Re: [Bf-committers] Collada importer/exporter kickout
From:Erwin Coumans (
Date:Jan 7, 2012 9:23:35 am

You are right that in some cases an exporter is better, but in many cases a C/C++ .blend importer is a better to go.

I just wanted to remind anyone reading this thread that there is an easy way to extract any data from Blender in C++, including animation curves, skinning info, textures etc, without the issues of COLLADA and FBX.

I'm not familiar with baking, but you might be able to store the baked data in the .blend file.

Thanks, Erwin

On 7 January 2012 09:10, Juan Linietsky <> wrote:

Erwin, That looks awesome and really useful, however the main advantage of an importer is the ability to bake everything (IK and other constraints) just like it's displayed in Blender.


Juan Linietsky

On Sat, Jan 7, 2012 at 2:07 PM, Erwin Coumans <


Instead of going through the COLLADA or other intermediate format you can directly extract any data from a .blend using this C++ .blend parser:

AnimKit with the .blend loader includes a small sample that loads a .blend file, extracts textures, meshes, animations, skinning and physics info. It shows a skinned skeletal animated character using AnimKit and GLUT (keeping dependencies to a minimum)

Cheers, Erwin

On 7 January 2012 07:38, Morten Mikkelsen <> wrote:

In my case I do not need morphs. I do need animation and skinning though. And obviously also geometries and materials. And it sounds like you have this covered? I have no sense of loyalty toward OpenCollada so if this is a viable solution I am for it. Can you make it available somewhere with instructions on how to install it so people can test it?

Cheers and thanks,


On Sat, Jan 7, 2012 at 7:06 AM, Juan Linietsky <> wrote:

Hi guys,

I made my own Collada exporter in Python and that's what I've been using. It's less than 1k lines of code and does not depend upon any library or anything, but it exports everything except morphs. I don't have much time to work on it at the moment, but it's so simple and complete



anyone else want's to work on it, it should be really easy. I'm also



expert at Python or Blender API so someone more experience can probably shape it up better. It's also much more stable than the official one (due to it being so small).

Feel free to do anything with it or integrate it into Blender, just credit me on the file. I would love to work more on it, or even make a proper importer since I have a high level of expertise in Collada, but I



little time and must work to earn my meals.


Juan Linietsky

On Sat, Jan 7, 2012 at 11:49 AM, Morten Mikkelsen <


Yes that's a very relevant point. A collada solution with just positions, UVs and normals is not a solution. In that case you might as well use obj format. I went through the hard work of writing a collada importer



get skinning and animation into my tech frame-work.

On Sat, Jan 7, 2012 at 5:03 AM, skoti <> wrote:

I know Collada importer / exporter is problematic (I wrote an importer for my engine and I know that everything in the Collada can be



N different ways).

- If you want to use the model in Second Life / Google Earth, you have to use Collada, if you want to use models in engines WebGL/Flash3D practically have to use Collada (is there any web engine with FBX importer?), Most game engines use Collada for importing data (support for FBX a few). FBX exporter also has bugs. Well, that FBX is in a blender, but it is not usually an option for people using Collada.

- If someone uses Collada it not for the base mesh + uv (then



obj) and for skeletal animation, lights, cameras and their animation (motion, color, light and their intensity), multiUV (uv for color, normalmap ... + uv for lightmap). And all this in current Collada exporter works well.

- No other exporter does not work here as well as Collada - ofc has bugs, but it has nothing what could replace the Collada in this



the future, Alembic can replace Collada, but not for several years.

IMO, better to leave Collada, until you will be able to replace



success to other formats like Alembic (FBX is not popular in the software and you can not replace him Collada in most tasks).

On 7 Jan, 2012, at 12:30, Ton Roosendaal wrote:


I realize the proposal was harsh, but it was meant as a public


as well (to khronos, opencollada team, etc). I don't want to




the hard working devs here. We do have collada IO work at some



that has been proven useful in several cases. The job is just incredible hard to achieve.

To move forward:

- userts who successfully applied .dae could also check whether


exchange format would have done the job as good. Tried FBX?

- note that for basic mesh (+uv) export, a quite simple script



the job already. It is probably a few days job for a py scripter.

- we are currently including about 100 MB of opencollada libs


make a

feature work that's meant to be able to exchange (I+O) full shots



environments, with character animation and so on. That's what



designed for, and that's a target we can't support.



Ton Roosendaal Blender Foundation Blender Institute Entrepotdok 57A 1018AD Amsterdam The Netherlands

On 6 Jan, 2012, at 21:21, Morten Mikkelsen wrote:

1) Official announcement that Blender drops Collada support 2) Move Collada support into a branch, out of trunk 3) Create a tracker "orphanage" or "branches" or so, where we



reports that are not in support (anymore).

I just want to say though I am not up for the challenge of




this sub project I would be sad to see Collada support taken




I use it often also with skinning export. I know many users do and eventhough I understand everyone's frustration I prefer the implementation that is there now (which I use often) to no built-in version



I am not disagreeing with any of the frustrating aspects of



you and others have mentioned. Just saying personally I do use it and would very much like to




With bugs and all :)

I have seen many other commercial tools with awful crappy .3ds importers and exporters but bad as they were they didn't take them out because people were still relying on them and using them.




in them.