I am currently updating the collada branch from master. i do not know how long it will take. I think that updating master from collada is currently not a good plan since i first want to be sure that the new collada module is clean, documented and stable.

Merge the changesd from collada onto master? Will that work?

You are free to take the .dae file i added as a reference

you probably need to merge from master branch. I will take a look at this.

In T55790#521464, @Gaia Clary (gaiaclary) wrote:

i am working on a major rewrite of the exporter and importer. This rewrite is available in the collada branch (based on master) and it very likely has fixed the reported issue. However i have not updated the collada branch since a few months because i had only very little time.

You could build blender now from the current collada branch to see if this solves the issue. I myself plan to finish the collada-rework and migrate this to Blender 2.8 in the time from september until december-2018. Should your issue not already be fixed in the current collada branch then i will take care of this as well.

i am working on a major rewrite of the exporter and importer. This rewrite is available in the collada branch (based on master) and it very likely has fixed the reported issue. However i have not updated the collada branch since a few months because i had only very little time.

I’m sorry, but we cannot investigate an issue we cannot reproduce, so will archive this report until someone can give enough elements to consistently reproduce it. Thanks for the report anyway.

No answer since ages, more than time to close!

i believe that the order of appearance of the data should not be important for the importer. In fact the importer processes the imported file twice so that it first collects the data and then assigns the data to blender objects. Please can you take a look at the collada branch? maybe things are different/better/resolved there...

Hm, looking at this part of the code, I can observe:

Well, if it is the only one full log, then we have already seen it on screenshot. I have no more information about this

As far as i know:

One problem, seems no one can explain where it is

Normally blender becomes very verbose when it breaks something. If you could get a crash log we could get a better info about where blender fails exactly.

I tried to export in root directory of both disks, in any case of export it creates empty file and crash blender. Not sure that it is collada issue, but all other formats to export works fine. Maaaybe collada using some dll which have been changed by windows

Then it indeed seems like something has gone wrong on your windows update.

It was mistake with wrong screenshot, but anyway. I turned off all addons and export nothing again

I tested it not right now. But i resave user settings to factory settings, open blender in debug mode and export nothing

1. Your screenshot indicates that at least the Avastar Addon is running.
2. And what is "measurit: Imported multifiles" ?

I already tried it, as i said in first message. Disable everything, restore user settings. I also reinstall blender deleting all user settings from AppData

Please try:

I cant test it on another windows 10 computer, for some reason people dont updating their windows for lastest version. I saw something like this problem in topics of about 2010 year, but not sure is it the same. I dont really know where is my crash log. I can provide screenshot of powershell (or console) where i try to export empty scene without selected objects

I personally can not yet test with Windows-10 but another user has confirmed that the Collada exporter works with Blender 2.79 on Windows-10. So currently your issue can not be reproduced.

Sorry for long delay. I said that collada export crashing blender with anything, even i try to export default cube or empty scene, in my case. Console crashing with "Unrecognized Exception" as i said too

And now it is also in Blender 2.8

The reworked animation exporter/importer is now in master

I have started a rework of the animation exporter. The fix in D3070 already solves the issue, but it needs some refinement and further testing.

if you refer to this file:

The demo files have been removed, so i can not check if this issue has been resolved or not. Please can you resubmit the missing files, so that i can test again?

This issue is fixed in master. You may want to double check that the problem is fully solved.

Hi Gaia,

note to self:
It looks like the export can be done in a straight forward way, see rna_Mesh_calc_tangents().
The FBX exporter does it similar to this:

Thanks for reporting. At the end i found the problem was in the way how the importer handles errors. Your files should now import with warnings about missing images.

I opened T53966 to collect ideas for how to support group instances in a good way.

According to Collada the Node library is used for organizing scenes with many objects into manageable subgroups (see citations at bottom of page). If i understand correct, then your request introduces one library Node per instance_geometry. This adds one level of hierarchy to the exported data: