Customization for Navigation capabilities

[kgschoueri]

It seems that by default there is a kind of buffer around the view rotation axis when one looks around in the scene. I guess this relates to the original purpose of the smithsonian viewer, which is to have you look at an object at the centre of the scene, but it looks a bit weird when you're looking around in a room like in my case. Perhaps there is a way to customise this, but I couldn't find it.

I greatly miss a "collider" option which allows one to set a physical bounding shape around some of the objects in the scene so that you don't find yourself trespassing walls or inside furniture pieces 😅(It happens frequently, especially because of the issue of the buffered rotation axis mentioned above).

[sdumetz]

This is how we configure scenes with interior spaces:

You can improve navigation in this type of scenes by disabling "autozoom" in scene properties. Additionally, you can set the camera Z-offset min value to negative and max to 0. Finally, disable backface culling on the model. It makes it transparent when viewed from the back side. You just have to ensure the "back side" of your model is on the exterior.

It works quite well in some cases but it's clearly an ugly workaround and won't handle collisions.

Can't speak for the Smithsonian but it's definitely on the roadmap for us to improve navigation inside rooms (beginning with #266). Colliders in particular though, I think they might not be the best "value for money" feature right now.

[gjcope]

@kgschoueri This is mainly because we currently only have orbit navigation. Orbiting has a focal point and then an offset distance from that point and angles describing rotating around that point. I believe what you are seeing as a 'buffer' is actually the offset distance.

What we need is "first person" navigation that operates in a wholly different way. It is definitely on our list, and has been for some time, it's just a relatively complicated task with the way things are currently set up.

As @sdumetz mentioned, you can somewhat simulate standing in one place and looking around by reducing that offset distance (camera z-offset) from the focal point to 0, but it doesn't provide all of the flexibility of a first person navigation mode. You can see an example where we did this here: https://3d.si.edu/object/3d/command-module-apollo-11:d8c63e8a-4ebc-11ea-b77f-2e728ce88125 or here: https://3d.si.edu/object/3d/main-floor:bf46d303-9620-466b-8ea7-5f86f1541a1f (see the tour)

[kgschoueri]

@sdumetz and @gjcope thank you both for the valuable information! I will pass this along to the person who was having the questions.