Is mirror displaying the original image. Is there a constraint for the reflective property of mirror. Does mirror reflect with 100% perfection?
[Physics] Reflectivity of mirror
reflection
Related Solutions
What you are observing is total internal reflection. Snell's law tells you that, for a ray transmitting through a surface $n_{1}\sin\theta_{1} = n_{2}\sin\theta_{2}$, where $\theta$ represents the angle of reflection from the surface, $n$ represents the index of refraction of the substance in question, and the labels 1 and 2 represent the source medium and the destination medium.
If $n_{1}>n_{2}$, as would be the case for light leaving water ($n\approx 1.33$)and entering the air ($n\approx 1$)inside of your watch, simple algebra will tell you that there is a range of $\theta_{1}$ at which you will find that Snell's law predicts $\sin \theta_{2}>1$. For this range of angles, since you can't solve for $\theta_{2}$, light cannot be transmitted, and must be reflected. So the watch looks like a mirror. In fact, if you flip over, and look at the surface of the water, you will find a portion of the surface of the water looks like a mirror, too!
So taking your questions one at a time 1) The concave mirror will form a virtual image if the object is placed closer to the mirror than the focal point of the mirror. The formula for the position of the final image is $s'=\frac{sf}{s-f}$ where s is the object-mirror distance and f is the focal length of the mirror. You can see that if $s<f$ this will be negative which implies a virtual image. The virtual image means the rays do not actually meet but appear to come from a point. In the diagram I have drawn real rays as solid and virtual rays dotted. So in the google glass case the LCoS screen must be closer to the mirror than the focal point (indeed from your photo of the prism you can see the curved edge is very slight implying quite a long focal length - on my diagram I rather exaggerated the curvature). The presence of the 45 degree mirror turns the real rays toward the eye while maintaining their relative angles and so the virtual rays now appear to come from a point straight ahead of the viewer.
2) You're right that at 2.5m the rays from the virtual image will be nearly parallel but it's still not infinity. Parallel rays would only result if the LCoS screen was actually at the focal point. At 2.5 m the normal eye can comfortably focus on the image when needed while allowing it to blur out when focusing at more distant or closer objects in the real world.
3) To the mirror: It is hard to tell but this is probably just formed by having two slightly different refractive index materials. Ideally lower on the right than on the left. This means the reflection from concave mirror to the eye will be by total internal reflection and so will be quite bright. Where as the spurious first reflection (shown by narrow lines in the diagram) will be a dim partial reflection. It is the image from this reflection you see when looking from the wrong side. You don't see the other image because the rays from it are directed away to the eye. It is possible there is a further partial reflection from the eye-side flat surface but you are not likely to see this because its rays will be mostly reflected back toward the concave surface by TIR. It's very likely this surface is in any case anti-reflection coated to stop just such spurious images. If they had used a partially silvered mirror instead of a prism then you might see some light from the main image on the wrong side due to dust scattering on the mirror surface but it would be a very blurred image. Using a bonded prism pair, probably assembled in a clean-room, this shouldn't ever happen here.
It's a pretty clever little arrangement! There is a neat little java app here where your can play with the imaging properties of concave mirrors. It's a little old so I had to add it to my exception list as modern Java complained it was a security risk.
Best Answer
An electrically conductive first surface mirror cannot be as efficient as a multi-layer first surface dielectric mirror. The reflection mechanisms are different - absorption and re-emission vs. optical resonance.
What is a "correct" image? If you want high fidelity return, use a fused silica corner cube operating via total internal reflection. Do you want to reverse intermediary optical aberrations? Use a phase conjugate mirror. Viewing your face in a phase conjugate mirror will be a spiritual experience.
Don a pair of circularly-polarized 3-D movie glasses, then look in mirror. Close one eye, Whoa! What will be the result of closing both eyes?