I don't know how familiar you are with the format, but the file is really correct on that part. I do admit immediately S3D exports DAT-models inefficiently (storing the same UV-coordinates for each map channel), but the UV-indexes however for each map channel are different. I could compress the file by removing the unused UV-coordinates in each channel, but this does not make the file 'more correct' from the file specification point-of-view. I will do this at some point, but it had no priority to me and made things only more complex...
I just think the problem is with the materials (specifically: nested multi materials), not with the way I store the different UV-channels and their indexes, but I may be wrong. Possibly it HAS been the problem all along and I should have not been so lazy leaving the compression out:rotfl:
You've given me some extra motive to figure this 'bug' out though. I'll keep investigating and working on it to see if I can fix this.
I can also analyze the Max-plugin source code to see if the problem lies there, or if there are workarounds. I may even submit a bugzilla report to FeelingSoftware if I think the problem is with the plugin.
To be continued
[edit] Test (nested) multi materials in Max on a sample model (a box), preferably with a diffuse, AO and spec map. Export it to DAE and reimport. Each time I tried, Max lost the map channel assignment in the material, but did import each UV-set (same as when importing a S3D exported model). I had to manually correct the material to have the model render correctly again. If you find otherwise, please send me a sample (Max/DAE file) so I can analyze it.