Lighting is a key component of photography, on an equal footing
with other aspects such as composition and content. In many cases,
photographers actively illuminate their subject with a variety of
lights to obtain a desired look. Lighting a scene is a challenging task
that is the topic of many courses and books, e.g. [Hunter et al. 2011].
Not only the notion of “good” lighting is elusive and heavily relies
on one’s subjectivity, but the traditional way to set up the lights itself
is complex. Positioning and setting the power of each flash is a
nontrivial and tedious task; further, most lights are accompanied by
modifiers that also need to be adjusted, e.g., a snoot to restrict the lit
area, or a diffuser to soften the shadows.
While post-processing the result in photo editing software is common,
this step has almost no effect on the lighting which essentially
remains the same as what was captured at exposure time. Recently,
a few photographers have introduced a new workflow to control
lighting that relies a lot more on the editing stage. Instead of a
single photo with many light sources, they take many photos with
a single light located at different locations each time. Then, they
load all the images as layers in photo editing software and carefully
composite the images to produce the final image. There are several
advantages to this workflow accounting for its increasing popularity.
First, capture sessions are shorter, easier to set up, and require less
equipment. Second, this workflow permits considerable control by
enabling arbitrary layering and post-exposure adjustment over all
the lights, allowing room for experimentation. For instance, one can
easily control the region affected by a light source with a mask and