We can do this on timescales on the order of 1 minute at the moment, if you'll allow me to stretch the boundaries of the question a bit. There are essentially two ways to do this:
- Slow light, where the refractive index of a medium is drastically increased, in the process slowing down the pulse velocity of a beam of light.
- Stored light, where the quantum states of photons in a laser beam are mapped onto atomic states in a crystal with the help of a second laser, and then converted back into photon states.
The second example is potentially more fruitful, and it is this method which enabled physicists (Heinze et al. 2013; pdf here) to "store" light for one minute inside a 3-millimeter Pr3+:Y2SiO5 crystal:
Figure 1(a), Heinze et al. 2013. The setup for light storage is much more complex than you presumably want, and involves many different optical elements.
Part of the reason the team was able to achieve such long storage times was their use of evolutionary algorithms to find an optimal pulse shape; these algorithms could potentially be extended in the future to reach longer timescales. Unfortunately, any storage mechanism will be subject to optical losses, and will be limited by the simple fact that it isn't an ideal environment.
There are, of course, some caveats. For example, the light being stored comes from a laser, so you would need a large bank of lasers to slow down complex images. You also have the issue that only one pulse is being stored in each crystal at once, and that pulse doesn't represent a long snapshot of time. Plus, the storage efficiency is not great - 0.4% by Heinze et al., with the possibility of doubling it to 1% if certain technical problems can be accounted for.
The upshot, though, is that yes, we can indeed see store light for noticeable timescales. The efficiency is poor, and the light has to come in the form of a laser, but it can be done, and it can presumably be improved upon.