Download API and metadata

download
vr

(Rupert Rawnsley) #1

If I download a GLTF model through the Download API, is it also possible to get access to metadata like annotations, VR scale and viewer position, background sounds, etc…?


(Paul Sketch) #2

no, there’s no metadata exported in gtlf, only 3D data fitting in gltf spec


(Rupert Rawnsley) #3

Thanks Paul. Any plans to make this available or is it something that Sketchfab are always going to limit access to?

Background: We do VR in classrooms where teachers can pick 3D models and send them to the students headsets. I want to add access to the Sketchfab library, but the scale and position metadata is essential for scene-type models. Asking the teachers to scale and position the models is a friction point we are keen to avoid.


(Waleguene) #4

Hi @rawnsley

As Paul said, we don’t currently distribute these metadata through the download API since the main point of the API is to access 3D content.
I will just mention the scale, that is a topic that has been discussed (internally) a few times now, because it’s the one that would make the more sense for the purpose.

We are aware that there is a need for this kind of data, especially for VR/AR application. In a perfect world, you want to download an object and have all the info to be able to display it as it’s real world scale, and centered.

The main problem here is that we don’t really have this info, except for the few models where users explicitely setup the VR Human in the Sketchfab Settings editor.
We do most of the job with the bounding box, and the way we compute a default scale for VR (when it’s not set by the user) is more to have something to see, than to guess the real scale of the model.
It would mean that for most of the models, you will not have the data, or you will have it with default values, which will not be a great help.

We don’t have the bandwith to fix that now, but we are aware of it and plan to improve this in the future


(Rupert Rawnsley) #5

@waleguene thanks for the detailed answer. We have come to a similar conclusion about the scale issue and ended up with some bounding-box based heuristics as well, but it’s not ideal.

I really like your VR scale-and-center tool, but I guess there’s no reason for the creators to use it until VR becomes a more popular viewing option.

I think we we will probably end up curating the models we make available to the teachers and adding supplementary data for scale and position. We already add things like teaching notes and curriculum suggestions, so it’s not too bad.