VR Sketch 11.0

Hello everyone!

We have released VR Sketch 11.0. Download it here.

This release contains support for SketchUp on OS X together with Oculus Quest. This means that Mac users can now start to make full use of VR! This is code that is still not thoroughly tested, but it should work. Follow the instructions to install VR Sketch 11.0 on your Quest, and also install VR Sketch 11.0 on your Mac—you need both.

Important! Mac users need SketchUp 2020 to properly edit models: in older versions, the “open group” command from VR does not work. Note also that the Quest is limited and won’t display models that are too complex (or sometimes it will, but at a horrible frame rate; it is not recommended to work with bad frame rates!).

This release also contains many other improvements that are useful with PC too. In comparison with 10.1:

  • Support for viewing text labels (a.k.a. “leader text”). You can hide them again in the settings in VR.

  • Point cloud support.

  • Some important fixes to snapping rules, which should work better now (more stable, less surprising snappings; snapping to face-face or face-line intersections, like in SketchUp; and various more fixes in this part of the code).

  • “Cut” and “Paste in place” commands let you move the selection between groups.

  • Animated components can be manually dragged with the teleport tool, not just clicked on. Also, clicking will now animate between more than two positions if the component was configured in this way in SketchUp.

  • Shooting videos: the camera position was frozen in some cases, instead of following the controller or the headset that it was supposed to follow.

  • Another attempt at fixing Vive Cosmos controllers.

  • Compress the textures transmitted to the cloud servers (for cloud models or Quest-connection-via-cloud). Useful if your model has got many large textures and your internet upload speed is limited.

  • The Quest is now able to reduce the texture resolution more than before, and will do so to keep the GPU memory usage within bounds. This should help with out-of-memory situations (e.g. black screens).

  • And many smaller fixes and changes.

(update: see VR Sketch 11.1 for hand tracking)

Have fun!

Armin Rigo

1 Like

Do these forums have the ability to pin posts? It might be helpful to keep the most recent release post pinned to the top.

Good point, thanks. Yes.

Now that you support the Quest, have you looked into finger tracking?

Yes, I’ve been looking at it recently. I still have to decide how much of the UX to support with hand tracking. In my prototype, right now you can click inside the initial dialog box, and then teleport around when a model is loaded. I guess we’d need at least a way to invoke the initial dialog box again. It would also be cool if the hands are shown in SketchUp if you’re connected to it. But supporting anything more is unclear. In theory we could map all the buttons of the controllers to specific hand gestures, but I doubt it would be very usable: the hand tracking logic is still quite approximative, so we would quickly run into limits where just picking up the real controllers is so much better.

Still, it is a technology that will only get better.
Maybe if you hold up a hand like you are looking at a watch, that could cause a menu to appear on the back of the hand, and you use the other hand to press buttons on the menu?
When you say approximate, how approximate are we talking? A way to quantify would be to create numpads at different scales. Have the numpads display several random numbers to be entered. Then you can record exactly how many entry errors you have at each scale.
What really interests me is if there would be a difference in accuracy between a numpad freely floating in space & one positioned like a watch so that you might have a better propreoceptive idea of where the buttons are in space.

At the moment, hand tracking works relatively well as long as all the fingers are clearly visible and the two hands are not too close or overlapping, from the point of view of the headset. It works poorly for situations where they aren’t. For example, pointing an index finger in a direction straight away from your eyes doesn’t work (the hand disappears). Also, tapping on a watch-like dialog box means your hands almost overlap; in this case they tend to both disappear too. I also thought about bringing a dialog box when you show your palm, like you’re holding it, but now I realize that it has got the very same problem.

If that’s the case then I can’t thing of any good UI. This does seem like the sort of problem that will get better over time though, so its probably still worth brainstorming.

It doesn’t mean no reasonable UI exists! For now, we can design around the current constraints. That’s also how it works when designing UIs for VR with controllers: we are constrained by hardware. For example, we need to minimise typing on VR keyboards—and, as far as I’m concerned, having a lot of dialog boxes in VR is also a mismatch, but we can’t avoid to have some.

In this case, maybe we can do some gesture with one hand, like a “look at watch” or “show the palm” gesture, and have the dialog box “grow” out of the watch or palm, but into a space that is no longer overlapping. For example, look at your left wrist/palm and the dialog box grows from there to the right, and ends up clearly to the right of the hand it started from.

You also have to consider the ease of certain actions. Touching the thumb & pinkie finger should be reserved for very rare actions. Could we maybe have a “toolbelt” that is always visible & stays in a constant position relative to the body around waist level?

Please try out the just-released VR Sketch 11.1. It’s not quite working well enough in my opinion. It’s OK if you want to do a quick trip around a model and ran out of controller battery, but that’s about it. Also, there is the idea that it’s easier for someone who doesn’t know VR at all, but I’m of the opposite opinion—just give him one controller, which just works, instead of this (cool but) unreliable hand tracking. So I don’t think I’ll invest more time in it unless we get some ideas that could be used to work around the unreliability.