How MeKiwi ported Cave Digger to 7 different VR platforms

Hello Jaakko, introduce yourself to my readers!

Hello, I’m Jaakko, the team leader of a Finnish indie VR company called MeKiwi. I’m the director and designer of Cave Digger.

Tell me something more about Cave Digger…

Cave Digger is a weird west mining adventure in VR. It’s my first VR game and I love it, warts and all. It all started in November 2017 with the pre-production for a demo of Cave Digger. We released the demo on Steam in May 2018. The demo got more than 100k downloads on Steam so we decided to push on and now it’s available on Steam, Oculus, VivePort, WMR, PSVR, Pico Neo2, Nolo X1.

You have made your game compatible with 7 platforms! Wow, it is quite a record… how have you managed to do that?

It is a lot of work porting and optimizing the performance and the controls for each different platform. Some are easy like VivePort, which is basically a Steam build with the Steam stuff removed, but then on the other hand you have Sony’s eldritch submission process and things like Quest performance optimizations. We are using Unity3D with VRTK, which made things easier.

You mentioned about VRTK. Online there are many people both liking and disliking this library. Can you tell us your opinion? Why is it good and why is it bad?

Using VRTK gave us a kickstart at the beginning to get things like teleport and grabbing easily working. Also, it made things easier getting the other platforms working. On the other hand, our version of VRTK wasn’t maintained properly and some of the platform SDKs for it were fairly broken. So that caused some extra work.

Porting between PC headsets is easy, but then you also worked on Quest, and that is pretty hard. How have you managed to reduce and optimize your experience? What has been your workflow?

You’re right, the Quest porting was the real challenge. With the game optimized for PCVR, we had a real problem on our hands – how to get it to run and also look good on the Quest.

We had to reduce draw call counts from 1000+ to 150, vertex count from a million to 200k and optimize CPU load to have the game running at constant 72fps. So lots of profiling of single-game sections, coming up with creative and not so creative solutions, and staying up late. OVRMetrics was really useful in realtime A/B testing on the device.

Also, we had lots of problems with RAM usage… we were using around 1 GB and found out that Android’s Low memory killer was randomly killing our foreground app, especially during scene reloads. That was lots of fun. After weeks of optimizing and profiling memory usage, we ended using an empty loading scene between the reloads so that some memory would be freed and the OS wouldn’t kill us.

Going deeply technical, how you managed your workflow for handling so many different Unity projects? I mean Git, the build process for every bug fix, etc…?

Git is of course a god-send. We tried to keep things in the same branch and same scene as much as possible just for the reason of back-porting. We did all the PCVR releases from a single branch, using the same scene. PSVR and Quest had a branch of their own because the optimizations needed for both platforms were similar. Then the bug fixes need to be committed to both branches, which sometimes can be nasty because of the delta.

Also, the flat version has a branch of its own. There’s been so many fixes and improvements on it that it would probably be easier to implement VR on the flat branch than back-port all that stuff to the old branches.

You have decided to go back from VR to flatscreen. Usually, companies take the opposite route. Why did you take this decision?

The fans asked for it basically. Also as we’re into publishing as well, so we thought it could have been useful to release the experience on all possible platforms.

From a UX standpoint, what have been the difficulties of porting an interface made for VR to pancake mode?

Well, creating an interface in the first place! Cave Digger VR was very minimalistic in that regard, having only a few in-world options. So this time we actually have a menu with the usual quality and customization settings.

But yes, the controls, the UI, the 1st person view, the animations… all had to be (re)designed. The biggest problems were around tool usage: now that we had the tools “stuck” to the screen as opposed to being in the VR player’s hands, how to make sure they were satisfying and practical to use? That required lots of testing and tuning.

We avoided changes to the content and concentrated on the actual 1st person experience. Of course, we needed to add some basic FPS mechanics like running, jumping, and crouching. Since VRTK was so deeply ingrained in the game, we ended up cloning the VRTK Simulator mode and creating a separate First Person SDK based on it.

Let’s talk about stats! What are the platforms where you have sold the most? And how is the feedback of the game on PC?

PSVR and Quest by far, both having a rating of 4 out of 5 stars. Quest for example has sold more than 5 times than PCVR put together. The next milestones are 100,000 copies and 1 million $ in revenue. Both VR and flat games have very positive ratings on PC.

What was your favorite dev platform?

Personally PSVR because I did it myself. Not that is was easy or smooth, but gratifying in the end, as it took 7 months going through all the hoops in Sony submission and testing. It was a valuable experience and the next release will go much more smoothly.

Tell me one big lesson that you’ve learned in all this process and that you want to share with my readers!

Game dev isn’t a bed of roses. Persist, trust in your idea and even us indies can make it. Also, finish your projects.

What are your future plans?

Making more games! There’s a big announcement coming from us soon, perhaps related to a sequel.

Something else to add to this interview?

Thanks for having me. Hit me up on linkedin if you have a game idea or an idea for a Varjo XR experience!