In backlighting (i.e. sun directly behind the subject) the face is in the same light as it is in open shade — skylight.
Skylight has two components, a dominant downward direction which models the face and a omni-directional one which creates a wrap around fill effect. In terms of cause and effect skylight and a LightSphere or StoFen in a small indoor room are similar: a downward modeling vector and fill coming from many other directions.
What the photographer in the video is doing by adding flash from near the axis is canceling out the natural directional modeling of the skylight with flat flash. The net effect is the same as a direct flash shot indoor in front from near the camera — flat — but it will look better overall than an indoor flash shot because of the rim lighting effect of the sun.
Had the photographer put the flash on a camera-flip bracket over the camera the cause and effect would change. Whenever a flash is moved off axis it's role changes from even fill which lifts everything, to key light which creates a modeling highlight pattern. When raised 16-18" above the lens the downward angle of the flash will create the same "mask" pattern of highlights in a full face view as the key light in a centered "butterfly" pattern where the nose is pointed at the overhead key light.
With a backlit subject the nose will also be pointing at the brightest part of the skylight also creating a faint "butterfly" pattern on the face. But during most daylight hours the skylight angle is so high that the brow will shade the eyes. The solution to that problem is to lift the face up into the skylight to get the skylight's "key" component into the eyes — before adding any flash. In other words before adding flash outdoors pose the face into the skylight as if you were shooting with window light using the directional component as key light.
The easiest way to get the face up into the light is to bring along a short 3 step ladder, get the camera and flash several feet over the head of the subject then say, "Hey look up at me..." The camera/face angle winds up similar to what you'd see in the viewfinder on the ground, but now the eye sockets will be well lit rather than shaded. The same thing happens at the beach in a situation like the video. The model down by the water line is several feet below the photographer higher up on the beach so the vector of the flash, even though it is near the axis winds up downward relative to her eye line.
With the face up into the skylight, modeled with a skylit butterfly pattern the next step is to add flash in a way that will not cancel that soft modeling from God's own softbox. The logical way to so that is to add the flash from the same 45° downward angle the "key" component of the skylight is hitting the raised face. That's where the flash bracket, and the ladder come into play.
If the lens of the camera is about 8ft off the ground (to get the subject looking up and skylight in the eyes) and the flash is 16-18" higher than the lens the flash and the skylight angles will be a similar 45° relative to the eye line. The net effect is very similar to using an umbrella as key light centered over the camera - brighter and more parallel (collimated) light rays in the center from the small flash source, with more diffuse omni-directional light from the edges.
This defies conventional wisdom and terminology of any flash used in sun as being "fill" flash, but the role of light sources change depending on their direction. In back light the sun becomes the "hair" light not the "key" in lighting parlance. Before adding flash the skylight is both "key" creating the highlight pattern on the face, and "fill" from its 360° wrapping of the subject. Flash added from ground level which winds up level with the chin will act as flat even "fill" raising everything the camera sees. But when the flash is move off axis it's role changes to one of modeling the face by creating highlights — the role of the "key" light.
Where does the fill come from in that backlit / raised "butterfly" flash on bracket scenario? The same place it did before the flash was added — the wrap around fill component of the skylight, not the flash which is now raised and creating the overlapping flashed highlight pattern. When the flash is raised the shadows the flash does not hit remain as dark as they were before the flash was added.
The lack of fill when flash is moved off axis and becomes the frontal "key" light becomes obvious when the flash is moved to the side instead of up which results in a noticeable and unflattering dark shadows on the nose and other parts of the face. Those distracting harsh dark shadows are not seen in the butterfly lit / full-face view because both skylight and flash are "key" lighting the entire front of the face. The nose shadow falls out of sight under the nose, and the sky-only filled darker sides of the face work in a good way to frame and slim the appearance of the front planes of the face. Butterfly / full-face is a strategy which works quite well outdoors if: 1) the sun is kept directly behind the subject and off the face and front of clothing, and 2) the flash is raised so it hits the face at a 45° downward angle relative to the eyes with face tilted up to get the skylight into the orbits.
What you don't have with any single flash scenario, indoors or outdoors, is independent control over the lighting ratio. Indoors with a StoFen the distance to ceiling and reflectance of the room is what creates the spill fill that lightens the shadows. Outdoors the "spill fill" comes from the sunlight bouncing off the sky. It is more predictable (3-stops below sunlight in incident strength) but given the short range of digital sensors not bright enough to make shadows look "normal" if the sunlit parts are exposed below clipping at "Sunny 16".
So if you want control over the lighting ratio, indoors or out, you need independent "key" and "fill" sources. If in the backlit situation above you were on a step ladder with subject looking up and flash 16" above the lens you'd get a flattering pattern on the face but rather dark shadows on the sides where only the skylight reaches. But if you were to place a second slave flash at chin level just in front of the ladder to create flat even fill you could dial in whatever tone you desired for the mood of the lighting with the lighting ETTL lighting ratio. The fill at chin level and centered will be "neutral" in the sense that it will not change the modeling the skylight and raised "key" flash is creating. The only difference you'll see after exposure is adjusted for the addition of the fill in the highlights is lighter, softer looking shadows.
What modifier size difference will does on a cause and effect level outdoors where there is no ceiling bounce isn't seen in the shadows it is seen in the highlights. Direct flash because it is a small source creates very small distinct "catchlights" and specular glare. If you set up a lighting pattern and change only the variable of modifier size outdoors you will observe as the modifier gets bigger the catchlights in the eyes reflecting it also get bigger and the reflections off the cheekbone become larger and less specular also, unless the skin is oily. So size matters, but mostly for the rendering of the highlights — diffuse highlights look "softer" than specular "hard" edges ones perceptually. The clues a shot is flash lit don't come from the fact the shadows are dark as much as the fact the face is covered with specular hot spots on the cheeks which are lower than those seen in natural lighting from above.
Indoors when modifier size is increased more and more of the light from the larger footprint bounces off the ceiling and walls. Just consider how a LightSphere, while very small, can light up a room like an overcast day with flat even light. A bare bulb flash will do exactly the same thing. Diffusion and soft shadows are a function of how many directions the light comes from and the relative strength of the vectors. If the light from all directions is more or less equal you get flat lighting (no modeling). As the the light becomes more dominant from one direction on the front side it becomes the "key" modeling component over the base "fill" being bounces around the room or off the sky. Indoors even when direct flash is used there is some "spill fill" effect. The more the diffuser increases the footprint of the flash the more light is spilled vs. hitting the subject directly and it creates more diffuse lighting on the objects and faces it hits.
Once you grasp the how the roles of light sources change based on angle relative to subject its easier to sort out the cause and effect of what each actually contributes and how best to combine them to achieve the desired look.
