Horde3D

I've been able to export collada files from Poser Pro and import them using ColladaConverter. However, there are a few problems:

1. In a direct export/import, there are some "holes" in the mesh
2. No morphs are imported (even though they are visible in the collada file)

Taking the collada file through Maya solves the mesh hole problem, but not the morph problem

Could someone show me a collada file with morphs that correctly import?

Morph targets should work. Could you provide a collada sample file that we can debug?

Since the file itself isn't technically redistributable, I tried to clean out all of the point data while leaving the structure of the file intact:

https://umdrive.memphis.edu/aolney/publ ... ped_m1.dae

Thanks

bump

To make my question a little clearer, here's a comparison b/w the jeanette model that was posted previously (based on that 3ds file, exported from 3ds 2010 using OpenCOLLADA for 3ds Max; Version: 1.2.5; I haven't tested this in the engine for non interesting technical reasons) and the Poser pro model.

Files at https://umdrive.memphis.edu/aolney/publ ... ometry.zip have 1 section of geometry each. As is clear, the Poser pro has a much smaller geometry section. Most notably absent is the triangle data.

My question more specifically is

1) What parts of the <geometry> are necessary for a morph import into Horde3d
2) If I can't export these morphs using Poser pro -> export, then I will have to bring them in as meshes individually. Does that mean I should load the collada figure into a Maya scene, then each individual head morph as a separate mesh into the same scene? Can those head meshes be anywhere, or do they need to be in correct relative position to the body (all overlapping and situated on the neck)? Any pointers would be useful b/c I've always worked with skeletal animation in the past. In this project I want to use skeletal animation for the body and morphs (hopefully) for the face.

Hmm, don't you have a minimal test model that can be debugged? If you can't make your model publicly available (which I understand), you can also send it to me or Volker via mail and we will treat it as confidential. But an invalid collada file does not help too much since we can't run ColladaConv on it.

Ok, I sent a link in an email to the address given on the Horde3D website. Thanks

Thanks for the test model. The morpher problem is fixed now in the SF svn (ColladaConv always expected an accessors for a source, although it is optional in the standard).

The reason for the wholes was that the model has a very small scale for which the standard epsilon for the degenerated triangle check was too large. I fixed this as well.