Today, The Climb launches for Rift. It’s our debut full-scale-production VR game, and partnering with Oculus has been an amazing experience. We’re now into the very first generation of mass market VR, and new forms of gameplay, storytelling, and experiences are emerging. We’re hugely excited by the work teams around the world have accomplished already, and about what’s coming in the months and years ahead.
At Crytek we like to think we have a track record of pushing visual boundaries and adopting the latest techniques and technology. VR was instantly exciting for us and a natural fit – cool tech, new challenges, and the opportunity to join that vanguard of developers breaking new ground. We’re looking forward to hearing feedback from the VR community now that The Climb is out in the wild, but in this blog, we’ll look at the lessons we’ve learned and some of the techniques we’ve employed to bring our artistic direction to life.
Aiming for immersion in our imagined perfection
When you’re in VR, the audiovisual effect fools your brain into believing that you’re there. You don’t need to approach photo-realism to deliver presence, as long as you’re providing a consistent world. However, for The Climb we knew we’d need to really push the visuals. Once we’d established a fun core game mechanic, we started thinking about the setting. It made immediate sense to use the immersion of VR to set our gameplay in environments that were inspired by areas of outstanding beauty around the world, so people could experience what it’s like to be high up on a rock face in the Alps or in Halong Bay. Why not take players to the most spectacular places you can, right?
The aesthetic we aimed for was hyperrealistic, an imagined perfection. To meet the expectations necessary for players to feel like they’re in a real environment – and also to satisfy the visual expectations of PC and console gamers – was a challenge given the technical requirements of VR. But for the experience we had in mind for The Climb, VR demanded that we achieve a certain level of quality. We learned a lot of lessons and had to rethink a lot of our assumptions about game development as we progressed.
Doing detail right
Detail, in the near field, for objects that imitate real world objects, became far more of a fixation than we ever imagined. In VR, everything that is up-close is open to inspection and subject to our natural curiosity. And when something doesn’t look as consistent as the brain expects, that sense of presence can be quickly broken. When players climb in our game, there are naturally a lot of assets in the near field. You’re close up to rock surfaces and come across a lot of different elements: checkpoints, vegetation, insect life, and so on. Our art assets needed to be of such a quality that they’d hold up to scrutiny in VR.
This was the first time we worked with assets in the sub-millimeter area in the CRYENGINE editor. Stitches on the gloves of the player’s avatar needed a huge amount of attention. We use bells for checkpoints, and to make them feel right we had to add precisely detailed, intricate engravings. The carabiners we use for save points had to have every little rivet authentically designed and placed. Most of those small details go unnoticed when you’re playing – it’s when they’re missing that you notice and it becomes a problem.
However, it’s not just about creating those details as close to a realistic standard as possible in a VR environment. You have to consider how to use art assets for storytelling, too. For instance, we have some cables fixed by screws to the cliff face that you can haul yourself across from one rock face to another. When we first created them, they were detailed, and they were accurate… they would have met the standard required for a current, 3D game. But in VR, because of that near-field view and our instincts to perceive and process the world that surrounds us, they just felt too perfect, uncanny even. In VR, that sense is really exaggerated, more so than in a flat-screen experience.
So we had to work through the story of the cable. How would it have got into the rock? What screws would real rock climbers use? How long had it been there? Was the screw weathered? So we did the research, found an authentic screw that would be used for this purpose, attached the cable in the way it would be in real life, and applied decals around the screw so you can see cracks in the rock showing how it was inserted. We now have more accumulated knowledge on screws than we ever anticipated! It’s a lot of commitment for a screw, but with the aesthetic we wanted, we found we had to go to this length of detail again and again. When your brain accepts the detail of everything near you, it’s tricked into assuming that everything else in the world around you is detailed too – which works really well with the huge vistas that we provide in the game. The effect of the work on the smaller details amplifies the overall effect of the world.
Always be testing
Testing was more crucial than ever for our VR development. Player comfort, what things look like in VR compared to the screen, how it feels, and figuring out where the eye is naturally drawn to in a scene are all elements that demand testing in the VR environment. In addition to the demands of detail we just discussed, scale became a really interesting challenge. What you think might look impressive and provide a bit of a “wow moment” in the CRYENGINE editor might have a totally different effect on screen. Or it could be more impressive than you’d imagine. We built up experience as we went along and over time we could more reliably predict (and develop processes around), for example, the sort of rock shapes that would work. Nonetheless, constant testing in the VR environment was absolutely invaluable to ensure we were on track.
More geometry required please
Typically we’d go straight to normal map baking to make an object look great in a game. But we found that in stereoscopic vision many assets felt flat, like they were made out of paper. An asset with those kind of properties would be a problem in any game, but once more, the requirement of immersion to deliver a great VR experience heightens the problem. It didn’t mean that we jettisoned the use of normal map baking entirely, but we had to be smart about where we could use it so that the process would still hold up. For instance, the billboard background techniques we’d normally use to efficiently get that sense of a big landscape felt just too fake. We had to work at how we composed our scenes, where we could use billboards, or how we’d layer them, and used far more real geometry in the distance instead of only using billboard assets. It’s more work, but it pays off.
Optimization. It’s a big deal.
Running all this extra detail and using more geometry while hitting 90 fps was obviously a big challenge. On the art side, to ease up the demands of draw calls, which has a big impact on the frame rate, we focused on reducing the amount of individual objects in the editor by grouping them together and then shaping, sculpting, and layering them. But optimization was something that we realized every team had to think about and commit to. And while CRYENGINE has plenty of grunt, we made some improvements to the engine (which VR developers using CRYENGINE will benefit from), particularly when it came to lighting and performance, to achieve the effect we wanted.
We’re proud of what we’ve achieved. Pre-release, people have been excited not just about the game, but about the graphics too. We aimed high with creating realistic looking, authentic assets not just because we wanted to see how good we could make the game look as a collective (ok, maybe that’s a part of it), but in VR, more than in a PC or console game, it dramatically and directly impacts the experience. For a game like The Climb, player expectations are set by previous games and the world around us, or how we imagine certain places to be. Meeting that expectation is what makes the immersion hold up. Ultimately, in VR, you really are creating worlds, rather than levels. To create ours we had to rethink a lot of how we approached development and the techniques we used to create our assets. The details might not always be noticed, but we hope the overall effect will be.