Controlled light 3D portraits from a stereo rig based on Sony RX100 III cameras

Karsten Bruun Qvist

1 Introduction

A recent article on this blog described a stereo rig assembled from two Sony RX100 III cameras, using generic Arca Swiss compatible parts for mounting, and commercially available cables, connectors and radio transceivers for triggering. Several options were explored for synchronization, including direct wiring of the USB/multiports of the two cameras, and two versions of triggering with Yongnuo RF605 transceivers. With all three options synchronization was accurate to at least 1/1000 of a second (the limit of the test method).

The article also shared work-flow considerations for setting up the cameras, configuring the rig in various ways, and processing of the photographs captured. In the present article we will apply the same stereo rig to making stereo portraits, using controlled flash lighting.

In the infancy of stereo photography, which coincided pretty much with the infancy of photography in general, stereo portraits were in high demand, but the interest faded, and today classic style portraits are not a common stereo photography subject, although exceptions do exist.

Studying stereo portraits a few years back, I made an observation that intrigued me: When looking at a stereo portrait on my computer screen, the eyes and face looking back at me turned when I moved my own head and upper body from side to side. I experienced a sense of interacting with the portrait in a way that is not possible with a normal 2D portrait. It seemed more engaging to me.

This experience initiated an interest in exploring what stereo photography can offer classic style portrait photography. Fabulous portraits can obviously be made in 2D, so it seemed clear that the basic portrait quality of a stereo portrait would have to be at least decent, in order to allow evaluation of whether 3D can make a meaningful difference in the photographic description of a person.

Since controlled lighting is a very important factor in classic portrait photography, I decided to make some stereo portraits using off-camera flash as light source. This article describes the challenges encountered, solutions and workarounds found, and also presents a couple of the portraits produced.

2 Off-camera flash synchronization

The Sony RX100 III has no hot shoe, or other provision for triggering an external flash, or for attaching a controller for off-camera flash. This is definitely a disadvantage for controlled light portraiture, but a workaround suggests itself.

Most speedlights and studio flashes can be triggered by a sudden change in light intensity caused by another flash, when set to what has traditionally been called optical slave mode. It is therefore possible to use the pop-up flash of the RX100 III as a master flash to trigger one or more off camera speedlights, or studio flashes, set to optical slave mode.

Since the RX 100 III uses a TTL flash system that sends out a pre-flash, the reflection of which allows the camera to determine how much light the built in flash should send out for the actual exposure, we want the off-camera slave flash to go off with the second burst of light, not with the pre-flash. For this to happen, the slave flash need to be set to the so-called S2 mode, rather than S1, which fires on the pre-flash. Flashes triggered in slave mode must be operated in manual mode, meaning that you set the output level on the unit itself. This is not necessarily a disadvantage if you work in a relatively static, studio-like situation.

As one might expect for a camera with a leaf shutter, it is possible to use shutter speeds at least as fast as 1/1000 sec with the Sony RX100 III without suffering loss of flashlight - as long as we are using only one camera.

Having seen excellent synchronization between two cameras, in the previous article, and great flash synchronization with one camera, you would be forgiven for thinking, as I did, that synchronization of an off-camera flash in S2 slave mode to both cameras simultaneously would be excellent. But you would be wrong! At a shutter speed of 1/250 sec, the result was one well exposed frame, and one black. Same thing at 1/100 sec, and at 1/50 sec. Only when shutter speed was reduced to 1/5 sec were both frames equally exposed, in a reliable way. So what was happening here?

To find out, the camera synchronization experiment described in the previous article was repeated but with one of the pop-up flashes activated, and the two cameras directly wired through their multi-ports. The results revealed that when the pop-up flash of one camera was activated, the exposure of that camera happened about 145 ms after the camera who’s flash was not activated. An effect of this nature should perhaps have been anticipated, as different events are set in motion in the two cameras when the trigger is pushed. The camera where flash is not activated proceeds to making the exposure immediately, while the other first sends out a pre-flash, then computes the amount of light to output, and only then proceeds to making the exposure.

If activating the flash of one camera delays the exposure of that camera, one might be hopeful that activating both flashes would delay both cameras to a similar extent, and thus improve the synchronization of both cameras to an external slave flash. It did, to some extent: when both pop-up flashes were activated the difference in timing between the cameras was reduced to around 15 ms.

Although this is an improvement, in practice it was still necessary to use an exposure time of 1/5 sec to ensure a reliable synchronization of both cameras to an external slave flash in S2 mode. However, an exposure time of 1/5 sec is not necessarily a game stopper in a studio like scenario, where one can effectively eliminate the influence of ambient light on the exposure by turning up the flashlight. In that situation the effective exposure time is dictated by the duration of the flash light, typically 1/250 sec, or shorter.

.. an observation that intrigued me: When looking at a stereo portrait the eyes and face looking back at me turned when I moved from side to side

3 Making portraits

Setup

As I wanted to be able to control the lighting situation completely with flash, the contribution to total exposure of existing ambient light in the room needed to be neglectable. With an exposure time in the order of 1/5 sec. needed for both cameras to 'see' the external flash light (section 2), ISO had to be set to its lowest possible value, 80, and aperture to around f/9. The built-in 3-stop neutral density filter was also used to limit the influence of ambient light.

The external flash, a GODOX AD200 Pro mounted in a 90 cm reflective silver umbrella with a layer of diffusion on the front was run manually and triggered in S2 optical slave mode by the built-in pop-up flash of one of the cameras. To avoid that the pop-up flash contributed to the exposure and created additional catch lights in the eyes of the subject, its flash exposure compensation was set to - 3 EV, and the flash was flipped up to avoid any light from it traveling directly to the subject.

Examples

The portraits below are shown in several formats. First as an anaglyph that requires red-cyan glasses for viewing. Since anaglyphs do not treat skin tones kindly, they are presented in black and white, rather than color. As always with anaglyphs, they may show up very well on some displays (such as the one they were created on), while showing massive ghosting on others. So, if the anaglyph does not work for you, please use one of the free-view versions below. Check the tutorial on how to do that here, if needed.

When you look closely at the portraits, especially the anaglyphs, and preferably on a desktop size screen, or larger, I encourage you to notice what happens when you move your head and whole body from side to side. Spend a bit of time looking at them from different angles, while moving. Ask yourself: does 3D add anything of value to the portrait, to the characterization of the person?

Anaglyph version for viewing with red-cyan glasses

Free-view versions: Top row left-right pair; bottom rom right-left pair

The most interesting question perhaps is whether 3D has the potential to offer anything of real value to portraiture, the characterization of a human being

4 Reflections

While the basic synchronization of two Sony RX100 III cameras is excellent, at 1/1000 sec, or better, it turned out to be an entirely different story when you use the built-in flash to trigger an off-camera flash in optical slave mode. While it is possible to work around an exposure time of 1/5 sec in a studio-like scenario, given that the effective exposure time will be determined by the off-camera flash, it does put you in a situation with limited options. For instance, it would be a challenge to make a portrait with a small depth of field, at e.g. f/2.8. If an exposure time of 1/5 sec is too restrictive for you, one way forward could be to add a delay circuit to help the cameras synchronize better.

Another possibility is to consider the use of constant light instead of flash. So-called COB LED lights are becoming increasingly relevant to photography as more powerful options become available, with designs that are compatible with traditional light modifiers such as umbrellas and softboxes. A LED light option, like Godox SL200 II does, however also come with compromises. To have the same level of control over the lighting situation, you would be restricted to working a lower levels of ambient light, due to their much more limited light output. Working with larger apertures, and smaller depth of field would not just be an option, it would be a necessity, as would be the use of higher ISO values.

At the end of day, the most interesting question is not which light to use, but rather whether 3D has the potential to add anything of real value to portraiture, the photographic characterization of a human being. Or, is it just one more demonstration of a 3D effect, as so much of the stereo photography we see online? If you have an opinion on that, please add a comment below, or send me a message.