rethink scene navigation

[roambotics_scott]

There are a few other ideas that touch on this but it's a huge / fundamental problem that needs to be addressed soon and is almost to the level of showstopper.

Another user described scene navigation as it stands as trying to perfectly position a pinata with a baseball bat while wearing boxing gloves.

The goal should be to make it something that's so easy and natural that you're not even aware of it.

Minor caveat: my prefered input device is the trackpad on my MacBook Pro. There are some chauvinists who think that's silly and think a mouse is better, but a multitouch interface is much more natural and intuitive. We evolved to point to and manipulate things with our fingers, and however used to a mouse you may be, it's an extra layer of abstraction, extra steps, an extra transfer function, and another unnecessary piece of hardware. Also, the issues I have aren't trackpad specific because to the app, any pointing device should be indistinguishable, so preemptively, please don't say «just use a mouse»

Obviously this is on the radar because things have changed in the last couple of updates, but it's not improving and the last one was a huge regression that behaves inconsistently and has actually rendered F360 nearly unusable for me. That makes me worry that the current approach is more shotgun debugging (where you try random things and see if they work) than rationally rethinking and starting over.

In order to design things effectively, it's essential that the user be able to efficiently move around them, get close to features, build, and refine them. The goal there should be to make it as intuitive and easy as holding a part in your hand, translating and rotating it in your hand, turning your head, bringing it closer to and further from your eyes, and adding things that are impossible with a real part like the ability to look around from a detail from an ant's perspective.

The following are essential:

1) reasonable translation

Regardless of the current camera viewpoint, moving two fingers from one extreme of the touchpad to another should move the view proportionally by as much of the screen (by default and with user configurable scaling).

Currently, it takes a huge (at least large single digit and sometimes more) number of full sweeps left to right on the trackpad to move by one screenwidth. That means that when you drill down to a small feature, it can take hundreds of swipes to get from one edge of a part to another even when the magnification would only require a few screens. That forces the user to zoom in, work, zoom out, pan, zoom back in, and repeat which is much more time consuming.

2) reasonable rotation

Shift+two fingers should work consistently with click+drag moving the view cube [in the latest update this is somehow sometimes inverted and sometimes not; I can see an argument for inverting it to correspond to looking around rather than rotating an object in your hands, but it needs to be consistent and some way to easily do both even without a VR headset would be ideal].

When you're «zoomed» into a part, looking and moving around should correspond to moving your head from that perspective.

It seems that the camera magnification is being used rather than translating and rotating the camera. That would explain a few things like the pan scaling depending on how close the view is to a detail, and it being easy to lose sight of something in the view and rotate through solid parts.

The ideal for me would be something like ant-man view. I can hold a part in my hand, look around it, and then I can shrink myself to the size of a detail, quickly walk / fly around, and look around while I'm there without things behaving erratically.

This has been done much better elsewhere.. maybe it'd be a good idea to talk with a game developer who has done something similar to figure out how they do it ?

Please do something about this, @carl_bass. To give you an idea of how big of a deal it is, I'd take it over a free Carbon3D printer with service and and supplies for a couple years, or even a few hundred k$ of cash.