I’m guessing it’s because Rhino can’t detect a “pinch zoom” situation…similar to the same issue related to the scroll wheel. When you click and hold the mouse (or a keyboard key), Rhino knows what is happening at that point, and can therefore act accordingly… And in the case of Rendered mode, “acting accordingly” means “degrade the viewport quality as much as needed in order to maintain certain frame rates”. Once the mouse (or keyboard) is released, Rhino then knows it can render the viewport in full quality. So in order to “dynamically downgrade” the rendering, there must be physical hardware event that can be detected… It is most likely that a pinch-zoom just gets mapped to a standard system event, which will not put Rhino into any kind of dynamic state…thus, there will be no downgrading.
But, just to be sure this is what you’re seeing, you can confirm it by:
Turning off the Skylight (or shadows in general) and see if things update any faster. If so, then it’s because the Skylight shadows are what’s bogging down the display, and given what I said above, Rhino cannot detect the event in order to degrade the quality.
See if you have the same lag in the Arctic display mode…it also uses ambient occlusion style shadowing, so I would expect that it would also suffer the same problem.
Also, when rotating the view with your mouse, take note of the quality of the shadows…you will see that they drop in quality as you spin the view, but as soon as you release the mouse, the view updates with a better quality version…It’s this dynamic-downgrade mechanism that provides for smoother frame rates.
Whatever the case may be, it’s possible that this is something that can be detected and special cased… I won’t know until I look…but I would also expect things like two-finger dragging to pan would also suffer the same lagging for the same reason… The “screen events” just get mapped to single system events, which are then processed by Rhino one at a time…and they do not act the same as when a mouse button is held down continuously… But I could be wrong, and I’d need to do more testing here…