Collada Import & Export
Tue, Jan 29
This issue has already been fixed in the master branch (for Blender 2.80). I also believe that this particular issue has been there since ever. Maybe back porting the fix to 2.79 might not be too important?
Mon, Jan 28
I also have this same issue.
Sun, Jan 27
(Messed up these comments forms default to Change Status: Resolved!)
Sat, Jan 26
Jan 21 2019
YesI show you example of a collada file with multiply materials/textures that actually displays correctly in Mac, Yes.
Blender 2.79b is fine no error.
@Kent Davis (mac4kent): thx for checking
@Philipp Kant (philipp) Allright I see you say "Only selected UV Map" I click make uncheck then export. Still same problem.
Just one more thing: Have you tried unchecking "Only selected UV Map" under Export Collada Texture Options?
We talk about Materials:
One can be show (Original).
Two and Three doesn't show
Have you checked importing that .dae back to blender?
(Since you seem to be judging this by inspecting in the MAC filebrowser)
@Kent Davis (mac4kent): Blender uses the same codebase on the Mac as other platforms. There may be sligt differences caused by differences in GPU's and other things, but Blender is basically the same on Mac, Windows and Linux. For this reason, the platform (Mac) is very unlikely to be the cause of most bugs.
I show you youtube. https://youtu.be/wntTiNXQCgM
Please do it use Macintosh not use Windows.
Did you use Macintosh?
Jan 13 2019
@Gaia Clary (gaiaclary) Sure, but the thrust of this (COLLADA) ticket IS losing primitives. The materials are lost because the primitives are discarded @Trevor Glauz (TSGlauz) says this defect is something core to Blender.
Jan 12 2019
Blender should not loose vertices, edges or faces on import, at least not with Collada.
Removing duplicate edges/faces is just how blender works since it is considered invalid data. You can have duplicate vertices, though. And with duplicate vertices, you can overlap faces that use entirely different vertices that sit in the exact same positions. You can also duplicate triangles into separate meshes and overlap them that way.
Excuse me, can I ask about the "primitive deletion" thing? Something like this happens in many importers, and it is making working with Blender very difficult, since I work with game data, that is always importing/exporting. I provided an example of an OBJ file (3DS does it too but less so) in a bug report a little while back and a real **** (awful developer) here named Sebastian tanked it immediately saying deleted polygons is a beloved feature of Blender. I'm just curious if there could be a connection. (In games overlapping elements is required and fundamental, and also in Blender I've never had a problem with this. Normally importers that lose data or destroy things are not published, and heads roll. I don't understand Blender culture. It's like a den full of nasty punks. The opposite of what you want for software work.)
Jan 10 2019
Thanks again for the patch.
It is now commit to master
You're right, that's an unnecessary extra normals calculation.
Sounds good! Much appreciated!
Thanks for this report!
I fixed the set enumeration to start with "0" (like in blender 2.7) instead of "1"
Thanks for the patch! It looks ok to me.
- fix unreported: collada mesh importer: Remove duplicate calculation of Normals.
Jan 2 2019
Dec 14 2018
There are many reasons for getting parsing errors with Collada. But really, this report system is for reporting errors related to Blender and not so much about importing issues into target engines. Unless a target engine complains about wrong Collada format, then we are back at work here. But then we also need to know what the complaints are exactly. And a demo blend file of course, as always :)
Uh Gaia, I have another object with a dae parsing issue that has a normal scale ( the transform scale all 1.1.1? Is this what you mean?) Do you know what issues are likely to be causing this? It's hard to find recources on making models for Secondlifeand opensim :/
Dec 13 2018
Ok close this task.
More than a week without reply or activity. Due to the policy of the tracker archiving for until required info/data are provided.
Dec 9 2018
I am sorry, i overlooked that the textures are present in the zip file. I give it another look later. It may also well be the case that the issue has been resolved as i am steadily (and admittedly only very slowly) working on fixing the Collada exporter since about 4 years now. Also with Blender 2.8 everything changes :) The new Collada exporter is already in the 2.8 Beta but it has not yet been tested to the very details.
@Gaia Clary (gaiaclary) I'm not sure what you mean by "updated blendfile with all the textures included". The .zip file I submitted contained all the textures in an external directory, because that's how the bug presented in this case, and given the intermittent nature of the bug, there's no guarantee that changing the structure of the example will preserve the issue.
Dec 5 2018
@Gaia Clary (gaiaclary) Seeing that the collada releases from 1.6.64 to 1.6.68 were all tagged in less than two months I would just make 1.6.68 the required version. If anyone is updating from 1.6.63 there is really no reason to build any version before 1.6.68.
Dec 4 2018
Thank you for your input! Actually I found a solution that seems to be working! Somebody happened to tell me that it's a bug and I should report it :P I probably should have tried a bit harder to research first!
While adding the fix, i found that the related definition files are available only since Collada 1.6.68 So this fix will only work for Collada 1.6.68 or newer.
Dec 3 2018
This particular "DAE parsing issue" is caused by negative object scaling which the importers to SL and OpenSim seem to reject (see error logs in the Viewer). However negative Scales are actually not forbidden by the Collada specifications, so this is not an error in the Collada Exporter, but an error in the Model.
@Gaia Clary (gaiaclary) Guess that one is for you :)
In the worst case we can check for the existence of a COLLADAFWAnimationClip.h file in CMake, but would be nice if OpenCollada had version defines.
I could not find an easy way to get the library version number. So i asked in the opencollada project: