However, physical latches don't detect contact, nor do they present an image or move the image. So, those first two steps aren't taught by physical latches.
Um, yes they are. Physics instructs my physical latch that it has been touched, and physics causes the image I receive of it to move while I continue to maintain contact with it and move my hand. Physics may also cause it to move back to its original position if I remove my hand, depending on the design of the physical latch.
Now, the real thing is certainly relevant prior art - you couldn't get a patent claim to Mie scattering, since that's inherent in why the sky is blue; and you couldn't get a claim to having virtual smoke rise from a virtual fire.
If your 'simulation' is throwing so much computing power at it that you can use actual physics to design the fire, smoke, and atmosphere, and just let them interact with believable results, I don't think it deserves a patent. Your processor might, if it isn't merely a progression of currently-patented ideas. However, if your simulation is a bunch of special algorithms that effectively reproduce the effect of real life without having to calculate what all the pieces are doing, that's may be worthy of a patent, and will doubtless require something more than and 80-year-old physics reference.
The fact that computing power has improved to the point that we can track physical contact and move high-res images with a responsiveness that is indistinguishable from reality by the human mind doesn't make using physical analogs in that environment innovative - that just makes sense. If they want to patent the painful, almost-intuitive design of the quicktime interface, circa v5, feel free. There's nothing obvious there, from the setting panel that can't be used on a low-res screen on up.