Spatial 3D stereoscopic video is easy to shoot on modern iPhones, but you’ll run into its limits pretty quickly. Third-party apps can improve things, but they have limitations too, and if you’re used to shooting with a couple of dedicated cameras already, why not use them?

Well, if you’re tempted to try making some stereoscopic 3D video with a couple of cameras sitting on the shelf, you should know what you’re in for — it’s harder than you think and there are plenty of mistakes to be made. On your first few stereo shoots you’re bound to get something wrong, but if you practice now, you’ll hopefully avoid catastrophic failure down the road when you get hold of a real camera.

Experiments like this are a great way to push your skills, but don’t get your hopes too high. Even if you don’t use something this janky to shoot real client projects, making it work with what you’ve got will give you real-world experience shooting in 3D that you can use if you step up to a pro setup down the road. So, with some regular 2D filmmaking gear to hand, how can you hack together a quick stereo 3D rig for viewing in a headset or on a 3D display?

As an extra reminder that this is an experiment, I’m not worrying about proper 3D Spatial Audio at all. Video’s the focus here.

Matching up two pro cameras

As very few cameras allow you to shoot at a high resolution with a dual lens, most 3D professionals shoot with two separate cameras. I’ll be using my Lumix GH6 and GH7, each one shooting at 5728×3024 at 50fps, with widescreen to fill a regular frame, high resolution for image quality, and higher frame rate for sync and to get closer to reality. (Note that open gate on these cameras tops out at 30fps, and it’s harder to make 4:3 video fill the frame in an Apple Vision Pro. This 16:9 resolution allows for higher frame rates.)

This setup brings flexibility, but also several worlds of pain, because you have to now consider:

• Matching the cameras and the lenses
• Alignment of the two cameras
• Placing the two cameras close enough together
• Matching white balance between the two cameras
• Syncing recording between the two cameras

That’s a lot to get right. Hopefully you’ll have two identical cameras; if you have two similar cameras with the same size body and sensor then you can make do, but try to minimize any differences. Action cameras are an option here — use whatever you’ve got.

If you’re trying mirrorless, two identical lenses is probably a harder ask, but you might have a couple of kit zooms sitting at the back of a drawer, or a favorite workhorse lens you’ve bought twice. If you’re using zooms, be sure to zoom to the same end of the range on both lenses, to ensure consistency. I found that a relatively wide frame, 14mm on micro-four-thirds, worked best. Set each camera to the same aperture, and consider stopping down to keep the whole scene in focus — if light permits.

White balance is something you can control, though you should perform a manual white balance with a grey card — each lens is different. To help with media management, set the same time of day, and if you can control file naming, give each camera a unique prefix, including “A” and “B” or “L” and “R”.

With the cameras and lenses matched, time to line them up, because if the two cameras aren’t pointing in the same direction on the same plane, sometimes the footage is unfixable. Yes, it’s possible to move or rotate one angle a small amount, but if the two cameras are facing different directions, that’s a harder issue.

A GH6 and GH7, with matching kit lenses and synced settings… but a huge gap between them

While the professional solution is to buy a specialized mount, if you want to experiment then you can just grab a few Arca-Swiss plates and use them to mount your two cameras side-by-side onto a double-width plate. In an ideal world, these mounting plates would all include locking pins to keep cameras facing the same way, but hey, this is not an ideal world.

Next, you’ll probably run right into the next issue: the distance between the lenses.

Considering inter-ocular distance and depth

Most cameras are too large to maintain a human-like inter-ocular distance (around 63mm) when positioned side-by-side, and worse, if you want to be able to film anything close up, you’ll need to minimise that distance (hypostereo) all the way down to just a few mm, impossible with any interchangeable lens. In stereo photography, the 1/30 Rule says that the closest object to the camera should be at least 30x the inter-ocular distance. Hmm.

The professional solution is to use a beam-splitter rig, placing the two cameras at 90° from one another. This does make mounting easy, but it’s probably not something you’ll be buying as part of an ad-hoc experiment. What to do?

It is possible to place two cameras closer together by rotating one 90° counterclockwise and the other 90° clockwise, with the bottom edges of each camera facing each other. This brings its own issues, though. You will sacrifice horizontal resolution, complicate mounting, and now that the two sensors are offset 180° from one another, you may experience rolling shutter differences. A fast-moving subject in one camera will now be scanned in opposing directions in each camera, and this may not be something you can fix.

So can you just… go with a huge gap between the cameras? My GH6 and GH7, horizontally adjacent, have an inter-ocular distance of about 150mm. You can shoot like this (hyperstereo) but you’ll have to be quite careful with shot composition, because you’ll be exaggerating the apparent depth of everything in the shot, and if there are objects both near and far, your eyes won’t be able to make all the objects converge.

Put another way, the depth of the image has too much dynamic range.

To avoid issues and follow the 1/30 Rule, keep everything far away from the camera, and try not to have any visible objects in the first 4.5 meters or so of the shot. If you have objects near and far, it’ll be uncomfortable to watch. Also beware of anyone walking from near to far — perspective becomes a little wonky.

Perfect sync is not simple

With two near-identical cameras facing the same way, close together, capturing a very similar image from two slightly different positions, all that’s left is to make sure they start and stop recording at exactly the same time. Sadly, this is hard. If you’re lucky enough to have two Blackmagic Micro Studio 4K G2 cameras, you can use Genlock to keep the cameras exactly in sync, but very few other solutions are good enough.

To cut a long, painful story short, wireless shutter release systems are different for every brand of camera and require custom cables that aren’t always included with purchase of a wireless receiver. Now that I have a complete set of Hahnel Captur (one transmitter, two receivers) I can sadly confirm that at least when used with a GH6 and a GH7, this doesn’t result in a frame-accurate synced recording. It is possible to sync timecode between two Lumix cameras via the flash sync port (and the adapter cables that shipped with each camera plus a separate BNC cable) but this does not allow for synced recording.

Sadly, sync between multiple cameras isn’t common, and it’s not always something you can add.

What about two iPhones?

You can shoot with two iPhones (in Log!) using Final Cut Camera and Final Cut Pro on a separate iPad to control recording, or using Blackmagic Camera to sync recording across both phones without an iPad, but, short of borrowing from a family member or colleague, I suspect most pro shooters don’t have two matched iPhones. If you do, this could be a decent solution, but care must be taken when mounting them.

If you flip one phone over to get the two lenses closer together, you could see rolling shutter issues. This can be overcome by mounting the two phones just offset, caseless, one behind the other, but that introduces a slight positional difference between the two lenses. None of these options are perfect.

Set all the settings, twice

You set white balance on both cameras earlier, but double-check all the other settings to make sure both cameras agree on every setting. Check that your cameras are still aligned. Autofocus should be off, because the two cameras won’t adjust at exactly the same time, resolution should be at least 4K, and frame rate is up to you. If you want a cinematic 24-25fps, that’s fine, but shooting at a higher frame rate (50 or 60fps) will minimize any time gap between the two angles caused by imperfect sync, and you can always conform to a lower frame rate in post.

Anyway, if you’ve gotten this far in the somewhat janky process, time to record! Double-check all your settings, make sure nothing has shifted, and start recording on both cameras. Grab several different shots with different kinds of subjects, as you might be surprised what works well in 3D and what doesn’t. If your subjects are moving slowly (or relatively static) then life is easier, but grab a few takes of each shot just in case.

How do you combine these clips in post?

After you’ve copied all the files, drop them into the same place, then load them into DaVinci Resolve (Studio) or Final Cut Pro; either way, it’ll get a little messy if your clips aren’t perfectly synced.

In FCP, you’ll want to create a stereoscopic project with the same width as the clips. In the Info Inspector, set Stereoscopic Conform to Monoscopic, and in the Video Inspector, make sure to uncheck Stereoscopic. Making sure they’re in sync, position them one above the other at the same point in time, and change Position X to place the clips on the left and right of the stereo frame. Add an Adjustment Clip above both clips so you can correct them together, and turn Stereoscopic ON for that Adjustment Clip.

In Resolve, you can select a matching pair of clips, right-click, then choose Stereoscopic 3D > New Stereoscopic 3D Clip and then choose if the clip you clicked on should be the left or right eye.

The Stereoscopic 3D Sync command here offers a method by which properly created pairs of clips can be synced in bulk, but if there’s any jank in your workflow, this can fail. If you’re experimenting, manual sync is fine. Create a new timeline from one of your stereo clips.

Slipping half of a stereo clip in time is easy in FCP, but in Resolve, you’ll need to use ⌘⌥< or ⌘⌥> to slip by a frame. (That’s Control-Alt-< or > on a PC.) Thank goodness for shortcuts.

And the rest of the edit?

In FCP, correct Convergence in the Inspector to control the depth of the shot, but there’s no built-in tool to correct stereo alignment. With a dual-camera setup like this, you can move each of the clips around manually, or you can use a plug-in called Spatial Kit which makes alignment easier. Full disclosure — I made this plug-in, and yes, I do benefit from purchases. You may need to correct color on each side of the clip too, and you’ll find viewing options (including Anaglyph for red/cyan glasses and Both Eyes) in the Viewer’s drop-down menu.

In Resolve, head to the Color page and click on the Stereo 3D icon to the right of the central panel. Press the link icon next to Convergence and you can scrub on the numbers to control the depth of the shot. The Vision tab lets you control how the stereo is shown to you, and Anaglyph and Side by Side are options in the Out menu here. To correct stereo alignment, you can use two different automatic options to the right, under Stereo Alignment, or fix them manually using the Eye controls to the left. Color Matching options to the lower right help make your two shots look the same.

Exporting in FCP is easy if you export to the Apple Vision Pro preset; it makes an MV-HEVC clip, though this tops out at 4096px per eye. Take a guess at the correct metadata, but you won’t be able to provide a baseline (inter-ocular distance) of greater than 100mm — because that’s crazy, right? You’ll have to send to Compressor if you want to export the full 5728×3024 per eye original resolution as MV-HEVC, but Vimeo won’t recognise this as a Spatial video file. Yes, it’s a bummer that I can’t share my 6K original. Export File is fine if you want a side-by-side version.

Exporting from Resolve is also straightforward: in the Deliver page, make sure you export to HEVC, then head to the Stereoscopic section and choose Both Eyes As > MV-HEVC. The same metadata options are here too.

After all that…

A janky setup with cameras that are too far apart can still capture a few shots which look pretty good, but that huge gap between lenses causes some real weirdness, and it’s just not going to work for pro shoots. The image quality is certainly better than an iPhone in “spatial” mode, but when you move your head, the image will warp around in unpleasant ways. Some shots just didn’t work at all, but enough did. On the plus side, discovering which shots really make your head hurt a little is all part of the learning journey. Take them out of the edit, figure out why they didn’t work, and remember the lessons for your first great stereo camera. 

Also, congratulations: you’ve done something few others have. Stereo photography has been around for over 100 years, but really high quality stereo video? It’s not common, and that’s because it’s much harder to pull off. 

To enable everyday stereo shoots, we need more reliable cameras which support stereo video, and which offer a regular, narrow field of view. Immersive 180° stereo is great, but you don’t always need immersion, it’s hard to repurpose immersive in 2D, and to do immersive well requires an even more specialised, more expensive camera.

In an ideal world, a known manufacturer would sell a high quality small camera with two lenses, or a pair of cameras with built-in, frame-perfect sync support. The visionOS 26 support for Wide FOV action cameras, which I have tested, means that two of those synced together somehow could be an affordable not quite 180° 3D solution.

In the meantime, rig up some gear for fun, and even if you don’t end up using this for great 3D video, it’ll work very nicely for 3D stills. But that’s for another article. Enjoy!