I’ve played through the first half of the mobile VR port, which is due for release next week. As we teased earlier this month, developer Vertical Robot has gone to great lengths to keep this version of its excellent adventure game looking incredibly sharp. I’m sure there are differences between the Quest and high-end versions but, truthfully speaking, I’d be hard-pressed to find them.
Check out our exclusive footage below; Red Matter doesn’t shy away from draw distances and the lighting work is easily the best I’ve seen on the platform. Textures are super crisp and the game’s physics all seem to be in place.
It’s difficult to comprehend how Vertical Robot pulled this off. So we thought we might as well ask them. Below is a recent Q&A we did with the team’s cofounder and technical artist, Norman Schaar.
UploadVR: Approximately how long has the Quest port taken?
Norman Schaar: We started working on the port right around the time we shipped Red Matter for PSVR back in December. So it must have been around 8 months for a two-person team. It’s a bit hard to tell as we have other company-related matters to attend to, which have certainly robbed us of some time. That being said, also quite a fair bit of overtime went into the port so we could ensure Red Matter would release on Quest sooner rather than later.
UploadVR: Why was it important for you to set this standard for Red Matter’s visuals on Quest?
Schaar: We found ourselves in a very similar situation two years ago when we started the development of Daedalus for GearVR. We were looking at the landscape of GearVR games and realized that most games at that time were extremely simple flat shaded games. We can only speculate as to why that was, but it certainly isn’t a hardware limitation for the most part. Although thermal limits on GearVR was something to keep in mind.
Having had experience with both triple-A console games and mobile games, we simply knew we could do better. There are certain effects that we pulled off years ago on an iPad 2, so it had to be possible on GearVR and Quest for that matter. Better graphics on mobile VR are absolutely possible with the proper time and know-how. A lot of old tricks can be used, but we also developed new ones along the way.
Visuals are definitely one of our strong suits. It’s one of the aspects we have a lot of experience in, so it made a lot of sense to make use of that skill set here. Especially in mobile VR were great graphics seem somewhat lacking.
We also like to argue that better graphics enhance immersion. A lot of our improvements focus on the notion of providing additional depth cues to the player. Parallax corrected reflections, the raytraced laser reflection, multiple point light reflections. All these add layers of depth to the image and help you better understand the depth in the world. This is especially true nowadays, where we don’t have varifocal displays with proper image defocusing. Any depth cue that you can add will help, even if the player only realizes on a subconscious level.
UploadVR: Red Matter on Quest has a one-time install. What’s happening during that?
Schaar: During the first launch of Red Matter, all the shaders that the game will ever need are compiled. If we didn’t compile them during the first startup of the game, the player would notice framerate hitches as new shaders showed up on screen, which would make for a terrible experience.
We did try to experiment with shipping Red Matter with precompiled shaders so that you could avoid the compilation on Quest altogether. But ultimately we ran into issues that were beyond our scope to fix, so we reverted back to this solution which was a better-proven path.
Vulkan should make this whole process easier, so I would expect upcoming games running on Vulkan not needing this compilation process at all.
UploadVR: What were some of the hardest aspects to either maintain or improve when porting to Quest?
Schaar: Most of the tasks were mostly simple but just very time-consuming. A lot of time went into manually reducing polycount of each mesh in the game. I don’t believe there was a single mesh left unturned during this process. We could have used simple optimizations algorithms, but those still don’t give you as good results as someone who is manually optimizing could.
But in terms of what was the hardest, I think keeping track of performance, benchmarking, etc was oftentimes very hard. This was especially true at the beginning of our development when perhaps Quest firmware was not as stable as it currently is. So bugs unrelated to the game were sending us into wild-goose chases, only to realize that we weren’t at fault. We ended up developing a workflow in order to make A/B comparisons within the same game execution through a series of debug menus and a fixed camera. This helped us make the whole experience less frustrating.
Another task that was quite hard was adding all the changes to Unreal Engine’s built-in Merge Tool to fit Red Matter’s needs. With these changes, we were able to more efficiently merge draw calls together and create better texture atlases. This paid off immensely.
UploadVR: Will the optimizations here be utilized in other versions of the game?
Schaar: For Red Matter on Rift our goal was to meet the minimum specs with great visuals, and I believe we succeeded there as well. Not much time was spent on optimizations on Rift as we were meeting the min spec just fine, instead, those efforts went elsewhere: Environment art, lighting, user experience, etc.
That being said, there absolutely is more room for optimization on both PC and PSVR.
The optimizations we have done on mobile are not easily transferable to PC in practice, as the mobile rendering engine is somewhat of a different beast from the console/desktop renderers.
UploadVR: What would you say to other developers looking to start out on their Quest ports today?
Schaar: Optimize from the start, keep your framerate in check at all times. Don’t let it slip throughout production or you are going to make your life more difficult in the long run. There is one thing that is better than optimizing: not having to!
UploadVR: What changes would you like to see made to Quest hardware/development pipeline to make porting easier?
Schaar: In terms of hardware, we believe it’s a very impressive first-generation 6DOF VR HMD. There is still quite a bit of room ahead of us in terms of optimization. This is not the best Quest is capable of, things are only going to get better from here. If hardware needs to be revisited, we would prefer it to be focused on improving comfort.
This story originally appeared on Uploadvr.com. Copyright 2019