Collada to GTLF online converter (https://cesiumjs.org/convertmodel.html) is doing the conversion however the orientation is changed 180 degrees. Attached reference model.
The original collada model is opening with Google Earth in proper orientation.
Earlier I used the same converter and it was working fine (with no change in orientation). Please suggest how to resolve this.
I have a similar issue. I set up Cesium1.2. I have been using the webtool to convert .dae files (which in GE are oriented due north) to .gltf. When I import them through CZML with [x,y,z,w] set to [0,0,0,1] my model points nearly due east and is healed over to the the right 42 degrees (like a sailboat).
I hope Anshika's and my observations have the same solution. Best, erik
Scratch my previous post. The behavior observed has the sailboat headed nearly due east and healed to the left at [0,0,0,1] while headed nearly due east and healed to the right at [0,0.7071,0,0.7071].
I understand that I am ruining my axis somewhere between the .dae file and being rendered in Cesium, but I can't figure where. Best, erik
There was a bug in the previous version of Cesium, which we fixed in 1.2.
Models are rotated into a Z-Up frame. In the previous version some models were upside down, so you may need to change the axes you are using. For example, the Sandcastle 3d Models example was fixed to use eastNorthUp instead of northEastDown.
Also, the conversion page was incorrect, so if you converted the model before version 1.2 was released, then you will need to do it again.
Tom - So if I have code that does not necessarily know which version of Cesium (1.0 or later) it is deployed in, can I just check Cesium.VERSION, and if it starts with 1.0 or 1.1, use northEastDown, otherwise use the more correct eastNorthUp?
Partially answering my own question, I did try using the northEastDown transform for Cesium 1.0, and at least the models are drawing right-side up, but they are going in the wrong direction now, compared to the same 3D Model examples in Cesium 1.2 Sandcastle demo. I was trying to keep Cesium-GWT compatible with 1.0, so I'll just include this caveat in my example code.
You’re approach is correct, but as you discovered, there are other bugs regarding orientation in 1.0 that cause the model to be in the wrong direction. I wouldn’t worry about trying to put special code in to support it, as it will be wrong for certain classes of models no matter what you do. A doc note sounds like the right approach (plus 1.2 is completely API compatible with 1.0, so there’s no reason not to just upgrade versions).