Written by Josh “Grubby Wombat” Richards,
Images by Will “Churrasco Wombat” Passos

---

It all started with a simple question: “What the hell is that empty area?”. Will had only just finished several dozen dives in Pines Cave to collect photogrammetry data, and he was now starting the painstaking, months-long process of using Agisoft Metashape to turn the 10’s of thousands of still images he’d collected into a high resolution 3D model. Eventually, that 3D model would lead to a game-changing map to support new divers experiencing one of the most renowned and frequently dived cave sites in the Mount Gambier region.

But I couldn’t see that yet. It was August 2024 – Will was hard at work and proudly sending me screenshots of the model’s point cloud… and all I could see was an absolutely massive gap in the data on the northern side of the Crazy Czech Room, the end of the known cave.

The Crazy Czech Room (aka the “CCR”) is unusual for being a 20m wide room created by rocks collapsing directly under a road which large forestry trucks drive over, and since it was discovered by two visiting Czech cave divers in 2002 it’s been the end of the known tunnel in Pines Cave. Getting to the CCR requires weaving through a rock maze down to 34m, jumping across to the “Surprise Tunnel” then swimming at depth for ~70m, before weaving back up a rockpile to 14m into the Crazy Czech Room itself. Will and I had been diving together to the CCR for years, and I knew that he’d scanned it with his photogrammetry rig pretty comprehensively, so there was no way the data should be missing from nearly half of it. I also knew this photogrammetry business could be tricky, but given Will’s experience surely there wouldn’t be issues with modelling a chamber this big and open?

For years Will had been testing and refining his photogrammetry setup, and after a lot of testing and tweaking in a wide variety of dive sites, it felt like he’d worked out the quirks. Collecting the data itself continued to be a lot of hard work though – swimming slowly into a cave while repeatedly twisting this heavy rig (carrying 4 GoPros and 3 BigBlue lights) back and forth in order to scan all the walls, then needing to do it all over again on the swim back out. But the level of detail Will could achieve with this setup was absolutely extraordinary. When the Soggy Wombats discovered the Iddlebiddy Extension, one section was found to be covered with pristine and exquisitely-coloured clay blocks. After I swam Will’s photogrammetry rig through this section, the 3D model he then produced was detailed enough to display the individual, millimetre-wide bands of colour in every single clay block.

The only time the modelling would have issues was when the software couldn’t get “alignment” as it tried to match the details on the walls between the sequential images from the cameras. This was generally because the camera’s field-of-view had changed too rapidly from the diver pushing the photogrammetry rig too quickly through somewhere tight, or because silt had obscured the details in the rocks and walls. It just didn’t make sense that there’d be a a massive gap in the model caused by an alignment issue given Will had been swimming slowly around a big, open room like the CCR with a ton of light and four GoPro’s snapping high resolution photos every second.

We also had absolutely perfect modelling of the South-Western side of the Crazy Czech Room – an area that was a lot smaller and siltier, and where I had long suspected we might find more tunnel if we just looked hard enough. But the model clearly showed that the South-Western side was a dead-end, with all the leads I’d planned to go back to and push turning out to be just loose boulders half buried in clay and leaning against impenetrable walls of dolomite.

The utter completeness of the model on the South-Western side of the Crazy Czech Room only deepened the mystery around the incomplete modelling of the room’s northern side. It was as if all these GoPro’s and lights somehow couldn’t see a solid limestone wall. And the only other reason explaination for why the GoPro’s weren’t seeing a solid wall on the north side of the CCR was if it wasn’t actually a solid wall.

There’s an old saying in cave exploration – you haven’t really checked a cave until you’ve touched the back of every wall. Gaps in the cave roof or floor can easily unnoticed until you’re right on top of them, so before you can truly say “There’s no way on” you have to physically place yourself at the furthest point inside every “dead end” and really look around. Even the “Surprise Tunnel” in Pines was only found because someone floated up to the roof of a chamber that everyone else had said was a “dead end”, and from the roof they spotted an angled restriction that was invisible to anyone following the line below. But Will’s photogrammetry data was starting to show us something we couldn’t have predicted – we might be able to find leads just by looking for the spots in the model that didn’t have any wall data.

We’d experienced something vaguely similar a year earlier in Iddlebiddy cave, where Will was collecting photogrammetry data as I poked my head into random holes hoping they’d lead somewhere. There were clear gaps in the model when Will scanned the jumbled rockpile West of the cave’s entrance, but we assumed it had been caused by the cameras moving too quickly, along with a lack of light and experience. It was only when we returned to scan that section more slowly and with more light that I spotted a hole in the roof – a hole which led to a previously undiscovered surface lake, and eventually onto the 100m of new tunnel we’d name the “Iddlebiddy Extension – A New Hope“.

Will couldn’t get to Mount Gambier for a few months, so he couldn’t do another photogrammetry scan of the Crazy Czech Room to try and fill the gaps in his model. Instead, I decided to investigate the northern side of the CCR myself to see if the gaps in the model were were real or not. It took just one dive to discover those data gaps translated into very real gaps in the cave – gaps we likely never would have found if the photogrammetry data hadn’t suggested we were missing something big.

Swimming to the northern side of the rockpile in the Crazy Czech Room, there was a small vertical drop into a low and very silty flattner. As I squeezed down into it, two things struck me immediately;
1) There was a series of loose rocks on the left side of the flattener, and I could easily shine my light between them and back out into the CCR – it was instantly clear those rocks had obscured the view of the flattener for anyone trying to spot it from the CCR,
and;
2) The flattener was packed full of clay and yellow silt, but the moment I stopped moving the visibility would start to clear surprisingly quickly… much quicker than this type of silt should be able to settle.

Looking ahead, I saw a stark white wall that contrasted with the 40cm high flattener I had just squeezed into. Not being able to see any clay against the white wall (and knowing a cave hasn’t been properly checked until you’ve reached the roof and floor against every wall) I wriggled forward a couple of meters to find that the floor suddenly dropped away, and that another flattener opened up at 90 degrees to my left. This second flattener was on an aggressive angle and looked incredibly tight – the sort of restriction that’d require a wetsuit and 7L cylinders to squeeze through, rather than the drysuit and 12L cylinders I was currently wearing.

Luckily I’d brought the cave explorer’s secret weapon – a GoPro and fill light on the end of a selfie stick.

I pulled the selfie stick out of my pocket, turned the light and camera on, set it to record, then jammed it through the restriction and hoped. With no way to watch the GoPro’s display, all you can do in these situations is slowly twist the selfie stick around and hope that you’re not filming close-ups of wet rocks or that a smear of clay is covering the camera lens. I spun the selfie stick around a few times, tried waving it up and down to see if I’d bump anything on the other side (it didn’t), then started to pull the selfie stick back – my buddy and I still had a 200m swim back to the surface through a nasty U-shaped depth profile, and neither of us wanted to accrue endless deco because I’d spent too long poking around in some silly little hole.

As I pulled the GoPro back though, it clipped the roof and created a shower of silt. I was about to push the selfie stick back and try to pull it through the gap properly, but I froze when I saw the silt from the roof was moving towards me. That made no sense – pushing the GoPro back away from me should have pulled any silt back with it, but here was silt clearly moving towards me? That’s when the lightbulb came on – the silt was moving towards me for the same reason that the first flattener was clearing so quickly. there was flow in here! Noticable water flow in Mount Gambier caves is incredibly rare, but when you see it, it’s generally because there’s something big ahead that’s pushing water into a small space. No one had ever reported any kind of water movement in Pines before, yet here I was looking at clear evidence of flow… and this was flow coming from a flattener which leads out of the Crazy Czech Room!.

Back near the surface as we were clearing our decompression, I tried to explain to my buddy what I’d just seen… but in my excitement, all I could think to scrawl on my slate was “BIG LEAD FOUND FLOW”

Back at the accommodation we immediately plugged in the GoPro and downloaded the footage from the selfie stick. Sure enough there was a decent size room on the other side of the 2nd flattener! It was a jumbled room inside a rockpile, with some scary looking boulders suspended from the roof… but this was a room which was plenty big enough to turn around inside and to hide a way onward too!

It’d take another month before Will and I could get back to Pines Cave, but now we knew we were onto something really big. Using a series of stage cylinders I managed to get to the second flattener with streamlined 7L cylinders and a wetsuit, so I squeezed through the 2nd flattener and into the jumbled room. It felt initially like there were leads everywhere, but it soon became clear that most were deadends – boulders leaning against solid walls just like we’d seen in the South-Western corner of the Crazy Czech Room.

Finding a way on from the jumble room would take nearly a dozen dives – each time wriggling through that tight and aggressively angled 2nd flattener (soon christened “The Gadi Gap”, with each time it getting a little bit easier as the limestone slowly wore down. I eventually managed to find a way out of the jumble room and picked my way under, over, and between the boulders to a point about 5m below the centre of the Crazy Czech Room. But once again, it felt like everything was a deadend – all I could see was boulders leaning against solid walls, or deep hollows full of trapped clay. Getting Will’s photogrammetry rig in wasn’t an option at this stage either – the whole rockpile was just too small and silty to avoid the model devolving into a mis-aligned mess.

I was also ready to give up. Sure we’d made some amazing progress, the flow was very exciting, and I had to admit I hadn’t checked every hole or wall. But wriggling and poking between these boulders so far back in a complex cave was definitely scary, and I was starting to suspect the flow was being fed by an impassable, 10cm wide vertical fissure I’d seen in the roof the first time I’d squeezed into the jumble room.

Now being able to wriggle through the Gadi Gap on 12L cylinders and a drysuit though, I had a little more time and gas to find a potential way out of the rockpile and into new tunnel. I’d swum back to the deepest point I’d found in the rockpile (~22m), and couldn’t see anything but solid walls and boulders. Frustrated at another big dive without finding the way onward, I saw I was close to turn pressure and decided to lock the reel and turn back for the surface. Just as I turned though, my eyes caught the flicker of an unexpected shadow – wasn’t that supposed to be another solid wall?

I swam back along the line to where I’d just seen the shadow, and moved my light around. Sure enough there was a basketball sized rock sitting on a small mound of silt, with a solid dark wall half a meter behind the rock. I moved the basket-ball sized rock out of the way and looked to see the dark wall behind was phreatic! Everything else I’d seen in the rockpile was either fallen rock from the roof and walls, or the exposed walls that were left behind by the same collapse. Phreatic passage is formed by water moving through underground tunnels which deposit minerals on the walls, and has a distinctive dark orange-brown colour which sets it apart from the bone-white of a limestone rock collapse. There was also clear flow coming through this hole straight towards me – a sure sign I’d finally found what I had been looking for!

Finding a potential way out of the rockpile was exciting, but I was presented with the same challenge I’d had when I’d found the Gadi Gap going into the rockpile – how do I know there’s enough room on the other side to turn around and get safely back out? Luckily I had my secret weapon again – I dragged the GoPro/Fill-Light/Selfie-Stick combo out of my drysuit pocket, turned it all on, and jammed it through to hopeonce again.

I didn’t bother writing a note to Will on my wrist slate this time – we’d talk about what I’d found once we were back the surface, but there was no point getting excited just yet. I’d done plenty of fruitless dives into this rockpile already, and while a phreatic wall was exciting, there was still no telling if it had any real potential without carefully checking the GoPro footage. So we drove back to the accommodation, copied the video to my laptop, and hit play.

Looking at the footage, I knew it’d be tight… but there was definitely space to turn around on the other side! Better yet, there were several big dark areas down low which the selfie stick’s fill-light hadn’t been able to light up. And the nearby walls that it did light up? They were ALL phreatic. This was it – we’d just found solid evidence of an unexplored, phreatic passage with measurable flow in one of the region’s most well-known caves!

The very next day we were back for the big push – I picked up the reel I’d locked the day before, wound it back in until I was in-front of the hole I’d spotted the phreatic wall through, and sized up the restriction. I’m not a big guy, but there was no way the hole was big enough to wriggle straight through – I’d have to unclip one of my 12L cylinders, swing it in front of me, and then try to push through. The restriction was a lot tighter than the Gadi Gap which had let me into the rockpile, but it was also a lot shorter – instead of dragging myself through several meters of squeeze, it was an uncomfortably tight pinch only half a meter long. And as soon as I was through, the only way was down.

There wasn’t enough room to clip my cylinder back behind me, so I just kept it out front as I dropped straight down through a small phreatic passage and into a large, inky black room 10m below. In a vertical squeeze with my exploration reel in one hand and my cylinder in the other, there was absolutely no way to inflate my wing or drysuit to slow my descent – I crashed directly into the pitch black silt at the bottom and threw up a mushroom cloud that killed the visibility instantly. Locking the reel, I pumped gas into my drysuit to ease the squeeze, got myself off the bottom, and finally clipped my cylinder back into place before trying to look around.

My crash landing had clouded the room in a heartbeat, but it was still obvious there was flow in this room too. Near the bottom at 34m the silt hung unmoving, but near the roof the water was crystal clear, with a distinct “silt-cline” which separated the clear water running along the roof from the murky mess I’d made below. Looking through the clear water near the roof I saw more an open space to the North, so I headed that way and found the roof opening up slightly before spotting another clear opening in the wall to the South.

Folks often ask what it’s like to find new tunnels underwater. Most of the time it’s a lot of hard and boring work knowing you’re probably close to finding something, but that something may be just another small room full of deadends. As exciting as it was finding this murky black room (which we later named “The Void”), deep down I was certain this would turn out to be yet another deadend – a small and short new piece of cave to tell folks about, but not something most divers would make the effort to bother seeing.

Very rarely though, you’ll poke your head into a new hole and be rendered entirely dumbstruck. I first experienced it when I wriggled down a silty tunnel in Englebrecht’s East and popped out into a previously undiscovered room 30m wide – a room which led to a >400m extension of a cave which people had previously believed was only 50m long. The next time I felt it, I was looking up through the roof hole in Iddlebiddy and seeing my exhaled bubbles break on the surface of a lake directly above. And I felt it once again as I swam out of the Void, through the opening in the wall to the South, and into an epic underwater tunnel that our team would quickly christen the “Crazy Wombat Tunnel”.

Running the line would be easy, because this tunnel just kept going on and up. After swimming up to a rock at the top of the visible tunnel, I looked over and saw the tunnel just kept going on the other side. After dipping through a low point, I found myself faced with another huge room which was utterly dominated by yet another rockpile. Except this wasn’t the rockpile that a visitor to the Crazy Czech Room might see – this was a rockpile 100m deeper into the cave, one that no living being had ever seen before.

I tied off the line at the top of the rockpile, looked briefly for a way on, and then turned for home. There was no way anyone would believe that I’d just laid 100m of new line in a cave that we’d been told was “fully explored” two decades earlier… at least not without footage.

We were back again the very next day, but this time I came armed with my 2nd secret weapon – an Insta360 X4 camera. I’d been trialling the X4 for a few months in the ocean and around large cavern areas to figure out how to setup the lights and capture as much as possible with as little impact on my diving as possible. I’d even brought it along on an earlier dive to the Crazy Czech Room with some of the my fellow Soggy Wombats, but I’d dropped it off in the Crazy Czech Room before another attempt to find a way through the rockpile. This time I knew I could get through the rockpile, there was an enormous tunnel on the other side, and the best way to capture the journey was on a 360 camera.

And the best bit about filming on a 360-degree camera, besides being able to prove you’ve found a new tunnel? By slicing out frames from the footage, we were then able to process the spherical images in Metashape to start creating a 3D model of what we’d discovered. The images from the X4 are about ~10 times lower resolution than the collective resolution of Will’s photogrammetry rig, but what it lacks in resolution it more than makes up for in ease of use and flexibility in challenging cave conditions. So while the model lacks the detail Will could get with his 4x GoPro setup, the shape of the cave the X4 produces is still accurate and it’s also a whole lot easier to through a restriction than it is pushing Will’s full-sized photogrammetry rig!

Naturally collecting photogrammetry data in a new section of cave also means producing new maps and models, as well potential new leads to check whenever there’s a gap in the data. In our case, the new model of the Crazy Wombat Tunnel puts the current end of Pines cave just 15m away from another of the area’s well-known caves – Stinging Nettle! The dive-able part of Stinging Nettle runs to the north of the doline, while the end of the Crazy Wombat Tunnel is approaching the Southern side. But there’s still good evidence for a connection between the two caves – the southern-side of Stinging Nettle has a small, dry drainage hole whichappears to be pointing directly towards our newly discovered tunnel.

I don’t know if our Soggy Wombats team will find the connection to Stinging Nettle, or if we’ll find a way around the new rockpile that block the end of the Crazy Wombat Tunnel. But one thing is for certain – we would never have found any of it without collecting and processing photogrammetry data, or by asking “What’s missing here?”. By using a variety of cameras we’ve been able to produce 3D models which help us map the caves we already know, to help spot potential new leads, and then be able to check those leads safely with a selfie stick before someone tries to squeeze in. Rather being toys for filming each other doing cool things underwater, the Soggy Wombats are using these cameras to collect scientific data which furthers our understanding of these caves, and directly enables the type of exploration our team uses to make discoveries in places that others have dismissed for decades as “already explored”.