Back in October I was given the opportunity to shot some 360° Performing Arts content for a Google Cultural Institute project. I had yet to shoot any 360° content prior to this so I used the time before starting the project to familiarise myself with the workflow and to get as much experience as possible.

Although it’s all fairly simple to do thanks to advancements in stitching software such as Kolor AutoPano Video, I thought I’d stress test the 360° rigs to the max to discover as many potential problems as possible… and naturally the first thing that came to mind was strapping a rig up to Ashley Holland and getting him to climb some buildings without any safety equipment.

To give the best TL;DR I can on how it all works, filming 360° content can be done on a variety of rigs and cameras but we chose one using GoPro’s as it was the lightest option with the best quality output.

The Freedom360 rig we used held 6 GoPro Hero 4 Black Editions, with each one recording either 1440p50 or 2.7K 4:3 25p depending on the frame rate we wanted and light conditions. The GoPro’s have to be shooting in 4:3 mode as to capture as much of the sensor as possible for stitching.

As of this moment in time there is no easy way to sync the GoPro’s properly unless you get the Odyssey rig (which costs quite a bit), so the more frames you capture per second, the better you can sync frames. Syncing is done in post and can be achieved by analysing motion (twisting the camera around for a few seconds), sound (giving it a few claps) or in some cases, by flashing the cameras with a light. We did all 3 methods after hitting record just to be safe. The GoPro’s were either set to record manually or by using the Smart Remote.

What started as a little test ended up being an adventure in DIY body rigs and a human flag 40-storeys up a lift shaft.

Body Rig version 1 - Aluminium

Above you see Ash modelling version 1 of our body rig. Made from the finest wooden and aluminium materials from B&Q, it was our first attempt at getting usable, shake-free footage from the 360 rig. The back support had zero cushioning, making the climb that bit more uncomfortable. Despite the rig digging into his back with every movement, it was quite light and simple to make.

Although Ash made it to the top (with Bobby by his side), the resulting footage made it look like we connected the 360 rig to a vibrator instead of an aluminium pole.

That faliure aside, we still learnt a lot from our first test, and obviously still got some usable material like the above ‘little planet’ image. In that image the same body rig is still being used, and another visible positive is that you can’t actually see much of the rig due to the size of the aluminium pole.

Below is an unedited stitch straight from AutoPano Video without any other manual adjustments. Although you can see the wood, the pole isn’t visible as it was right in the centre of 3 GoPro shots and therefore ‘stitched out’ by AutoPano.

Rig placement/direction/angle on the pole is another thing we experimented with…

Body Rig version 2 - Plastic and duct tape

This rig happened to be a last-minute build as we were blessed with some rare sunshine in London. As you can tell it was put together in just under 2 hours and was really, really light.

The plastic was a lot less uncomfortable on Ash’s back than the previous wooden build, and had 2 pipes coming off it to help lessen the vibrations we experienced on the previous test.

The size of the pipes in the GoPro shots obviously meant they’d be present in the final stitch but it wasn’t all that bad. The rig did however look like crap, and that was before we ruined the aesthetics with more duct tape for structural support.

Aside from its looks, the rig worked really well on what was probably the most challenging climb Ash did. We did end up reshooting this climb on our final rig but, again, we still got some cool content from it.

The image above is at the top of the climb with the rig Photoshopped out.

Below is another climb we tested the rig on and, although it’s projected in Little Planet mode, you can see how much of the rig appeared in the stitch from AutoPano Video.

Body Rig version 3 - Wood, sponges, plastic & a selfie stick

The third and final rig we used was mostly Ash’s creation. After his first hand experiences of getting impaled by a wooden back support and then looking like a fearless plumber, he came up with this cushioned lovechild of the previous two builds.

We shoved a selfie stick – wrapped in kitchen sponges at the base for absorbing (get it?) some of the vibrations – into the PVC pipe and wrapped it with tape to secure it even more.

The above photo shows the rig with extra tape around each extension point of the selfie stick – this was to prevent it turning and therefore possibly unlocking/shortening during the climb.

The setup you see here, with the stick in the middle of 3 GoPro’s, was by far the best for producing an image with no stick visible at all after stitching. It was also the most portable and lightweight rig, despite looking fairly suspect when walking around with it.

Post Production

Although the whole concept of creating 360 videos may seem tedious, the post production process can actually be quite simple depending on what you’re doing.

I’ve been using AutoPano Video from Kolor (now owned by GoPro) to stitch my videos as I had previously used AutoPano Giga for creating panoramas and was impressed by how well it worked – easily surpassing Photoshop for speed and stitch quality when using a load of images.

Kolor have uploaded a video describing the entire 360 workflow, so watch that if you’re interested in the details.

As Ash’s climb was around 40 minutes of GoPro footage, I only fixed up the horizon and stitch on 4 different segments totalling approximately 4 minutes. Fixing the horizon/stitch involves a process similar to keyframing in After Effects, and was only necessary as Ash’s body, and therefore the rig, was moving from side to side during the climb.

Contrary to what a lot of people think, the stitching isn’t a one size fit’s all kind of thing. If objects are moving closer or further away from the rig you will get parallax issues due to the distance between the GoPro lenses. Think of it like holding an object in front of your eyes and alternating between your left and right eye – each one will see something different.

A way to avoid this is by having an ‘exclusion’ zone around the rig, usually around 3-4ft or so, to minimise parallax issues as much as possible. Obviously in our case this wasn’t a possibility as the rig was going within inches (or sometimes hitting) things during the climb.

To fix this AutoPano Video gives you the option to keyframe the stitch on the timeline. Usually simply hitting the Stitch button gives a good result, but if you need to fine-tune your stitch you can do so in AutoPano Giga quite easily. Luckily the auto stitch worked quite well for us 90% of the time despite the various obstacles and distances during the climb.

Unfortunately the rig did hit something on the way up which knocked it out of position slightly – putting the selfie stick right in the middle of one of the GoPro shots. The stitch in the screenshot above shows the result of our ideal position, stitching out the stick completely. Ash wasn’t quite feeling up to the task of going back up to do another flag in freezing temperatures so we decided to go with what he had.

The video is best experienced with a headset (even Google Cardboard), but still works well in the YouTube app on mobile phones, so give it a play on that if you can.

If you’ve got any other questions, leave them in the comments below!

One Comment

Leave a Reply