Disclaimer: Just “thinking out loud” here so take it for what it’s worth.
If I were taking this shot, my technical objectives (challenges?) would be:
- Clean background free of distractions (no big deal—just rig something up)
- Separation of bottle from background (i.e., eye-catching/dramatic contrast—should be no problem with white background, light positioning, and curves tool in post)
- Slightly directional/off-axis lighting to prevent an ugly flat look, but no so much that any part of the label is buried in shadow (just experiment and watch out for too much fall off)
- Keeping specular highlights at bay
The last point is, IMO, what will separate a “decent” result from a “great” one, as it’s a bit of a challenge with shiny rounded surfaces.
I’d approach the problem by trying to use a light source that is bigger than the bottle itself in all dimensions. A window is an obvious choice, but personally, I don’t like being at the mercy of mother nature’s whims, so I’d probably go for a strobe-based solution where I have full control (again, this is just me, YMMV). I don’t think I’d even bother trying the umbrella because it’d just show up in a fugly ribbed reflection, and I don’t have a large softbox (yet), so what I’d do is rig up 2 or 3 off-camera strobes in a vertical line and “double diffuse” them via 2 layers (separated by a few inches) of something like tissue paper or parchment paper. The dual layers help smooth out the “hot spots” from the strobes, thereby creating a large and even light source that you can move around to taste. I’d probably end up rigging some ad hoc gobos to keep light from going places it doesn’t belong. It’s hard to pre-imagine every possible problem on shots like this, so some on-the-fly ingenuity is always going to be required.
You might also want to try placing the bottle on a glass surface and lighting it from underneath, in addition to the key light, for some interesting effects.
A long exposure “light painting” approach using a light wand might also yield a striking result, after some trial and error.