Seeing something from only one angle means you have only seen (what?)% of its surface area at most? Is there a logical/mathematical way to derive what the very maximum percentage of surface area you can see from one angle of any physical object?
For instance, if I look at the broad side of a piece of paper, I know I have only seen 50% of its surface area (minus the surface area of the very thin sides). Is 50% always the maximum amount of surface area you can see of any object from one angle?
Assumptions: This is assuming we aren't considering transparent/semi-transparent objects or the extra parts we can see with the help of mirrors. Just looking at the very surface of an object from one particular angle.
 A: Consider a thin piece of paper, then as you said you can see only half of it's surface. Now, take an arbitrary point in the middle of the paper, and extrude the sheet from this point, ie. create a pyramid. You can increase the apparent surface as much as you want (just by increasing the height of the pyramid), so you can see from 50% (included) to 100% (not included) of the surface area of this object. Looking at the other side of the pyramid, can see 0% to 50% of the object.
A: There is no such upper bound. 
As a simple counter-example, consider a thin right-angled solid cone of base radius $r$ and height $h$, observed on-axis from some large(ish) distance $z$ away from the cone tip. You then observe the tilted sides, of area $\pi r\sqrt{r^2+h^2}$, and you don't observe the area of the base, $\pi r^2$, so you observe a fraction 
\begin{align}
q
&=\frac{\pi r\sqrt{r^2+h^2}}{\pi r^2+ \pi r\sqrt{r^2+h^2}}
\\ &= \frac{\sqrt{1+r^2/h^2}}{r/h+\sqrt{1+r^2/h^2}}
\\ &\approx 1- \frac rh
\end{align}
of the surface, in the limit where $r/h\ll 1$, and this can be arbitrarily close to $1$ so long as the cone is thin enough and long enough.
A: As a completely tangential type of answer. Consider a neutron star; due to the General Relativistic bending of light in curved space we are not bounded by the dull constraints of Euclidean geometry!
If the radius falls below 1.76 times the Schwarzschild radius for its mass$^1$, then then all of the surface is visible, when viewed from any direction (e.g. Pechenik et al. 1983; Kraus 1998). Light rays emitted tangentially to the surface from a point on the neutron star that is opposite a (distant) observer would be bent through 90 degrees whilst travelling to the observer. According to that observer, the antipodal point would actually form the circular boundary of the observable disk and hence they could "see" 100% of the surface.
For neutron stars with radii even less than this (as a fraction of the Schwarzschild radius), then there are multiple paths for light to reach the observer for some parts of the surface and they could be seen more than once on the visible disk.
e.g. a neutron star just below the critical $R/R_s$ value where the entire surface can be seen. (From http://www.spacetimetravel.org/ssm/ssm.html )

$^1$ @SeleneRoutley asks whether such objects are theoretically possible or stable? A full answer to this is contained within my response to this Physics SE question, but to summarise:
GR does impose a limit on the mass/radius ratio of neutron stars, but the exact limit depends on the equation of state governing the structure. The diagram below shows the theoretical mass-radius plane, adapted from Demorest et al. (2010). The shaded part top-left and marked "Causality" is outlawed by GR and any equation of state where the sound speed is less than the speed of light (as it must be).
I have added a thick red line which marks where the radius equals 1.76 times the Schwarzschild radius. Any neutron star above this line would  have a completely visible surface from any viewing direction.
Now there are possible equations of state (marked as labelled loci in this diagram) that allow high-mass neutron stars to exist above the red line. However it is currently not possible to measure the mass and radius of any neutron star with enough accuracy to place them unequivocally above this line, but the measurements for some are consistent with this.
The phenomena discussed above are "routinely" taken account of when inferring spatial structures associated with the pulse profiles of rotating neutron stars at X-ray wavelengths (Sotani & Miyamoto 2018).

A: There are not bounds in either direction.
Invisible majority:
Take a rod (a solid cylinder with circular cross section, and disks for end caps) of radius $r$, axis of revolution running along the $z$-axis, lower end cap at $z=1$ and upper endcap at $z = u$.  Place the observer at the origin.  The lower endcap is the only portion of the cylinder that is visible, so the visible fraction is \begin{align*}
    &(\pi r^2) / (2 \pi r^2 + (u-1) \cdot 2 \pi r)  \\
    &= 1 / (2 + 2(u-1)/r)  \text{.}
\end{align*}  By increasing $u$ or decreasing $r$, we can make this last expression as small as we like.
The basic idea here is that we can arrange for the object to have a small "shield" close to the observer, which occults a huge surface.
Invisible minority:
Take the spherical shell of inner radius $1$ and outer radius $2$.  (This is the set of points with $1 \leq \text{radius} \leq 2$.)  Drill $n$ cylindrical holes of radius $r$ along rays from the origin.  For the same reason that the curved wall of the cylinder in the above case is invisible, the curved wall of these drillings is visible.  Suppose we choose $n$ so that half of the inner sphere's surface has been drilled through.  (We could do better.  We can easily exceed 50% by close packing the spherical caps corresponding to the drillings on the surface of the inner sphere.)  Then the ratio of visible to all area is \begin{align*}
    &( (1/2) \pi + n \cdot 2 \pi r ) / ( (1/2) \pi + n \cdot 2 \pi r + (7/2) \pi)  \\
    &= 1 / (1 + 7/(1 + 4 n r))  \text{.}
\end{align*}  Notice that if we reduce $r$ by a factor of $2$, we increase $n$ by a factor of $4$, so by decreasing $r$, we may increase $1 + 4 n r$ as much as we like.  That is, we can make this ratio as close to $1$ as we like.
The basic idea here is to make most of the surface area barely converge toward the rays of vision so that only a small amount of the surface is occulted.  Then fill in the volume "behind" these convolutions so that the resulting back face has a small area.
