Horde3D

I suppose this makes sense in a way. But it stumped me for a while.

I was making an animation for a test character. I added a bunch of mocap data from the CMU mocap database onto the character and exported a walk cycle from frame 80-100 ish. What actually exported from max was fairly strange, in that it exported those frames, but when I reloaded the file in max, the frames were back at 80-100 (so basically it seemed to export the frames by the original offset). So when I tried to play the animation loop in Horde, the frames were offset and didnt play correctly because I believe horde simply takes all frames in the DAE as its animation.

I'm thinking that I might modify colladaconv to allow output of start-end keys for anim files. That way i can model a whole sequence of anims in one DAE and then give the collada conv parameters to output specific clips rather than the whole motion. That would better match the animation clip structures I setup in the game too.

anyway, just thought I'd share.

I'm not so familiar with max, but if I understood it correctly you propose a command line argument for colladaconv to limit the frames exported from the DAE file to the .anim Horde file, right?
Would definitely make sense.

Yes, thats basically it. So you could do a batch export from one single .DAE file to multiple .anim files. Just supplying different command line arguments.

Hmm, that behavior is kind of strange but both, OpenCollada and ColladaMax seem to have it. In that case we would really need some clip handling in ColladaConv. Command line arguments are ok but don't work well with the batch processing that we have in Beta5 (or rather the svn trunk right now).

I would suggest to introduce optional meta files that can contain asset specific processing information. For example, ColladaConv would check for a file "asset.dae.cc" and if available, parse it when processing "asset.dae". For animations it could contain clip data while for textures it might in the future have information about the desired compression.

The 'Star Wars: The Old Republic' team have an interesting data pipe they call 'OMP', and they call these meta-data files 'side-cars' (because they attach onto the main data file).

Where possible, I'd prefer this data to be contained within the data file (extra data in COLLADA XML nodes, XML inside PSDs, etc), but a corresponding meta-data file is a good idea where that's not possible.

My preference would be to not have another file to worry about too. But I can see marciano's point. I'm not sure that collada provides a route for animation clip creation anyway.

Animation clips would be wonderful if supported by ColladaConv!
Currently our animator has to re-export the same file for every animation clip it has.
This produces a lot of DAEs where only the animation data is different.
Which then need to run through the ColladaConv converter, producing a lot of unnecessary files.

I would definitely prefer to extract as much information as possible from the raw source files. But if it is required to modify exporters for that (e.g. for writing the COLLADA extra nodes), I think it is better to have our custom meta data files that would work with all source files. Collada has the concept of animation clips but I'm not sure how well it is used by exporters.


Thanks for the link DarkAngel, a good read. The knight model that we use is coming from Rob, the guy who gave that presentation

I'll maybe take a look at adding in clip support then. Might take a few weeks because I've got other things on my mind right now.