New Heights News

We rely heavily on photogrammetry for the production of our realistic climbing game New Heights. In this series of articles we will attempt to explain why, how and how we got to this point.

For our climbing system to work it needs something to hold on to, because climbing in the air or on a flat wall would be quite difficult. So we need some (collider) meshes, but how detailed do we need them? About 100M - 500M verts, those kinds of models. They are rocks (think more like whole cliffs), which have to be super detailed, because you have to actually climb on the details. More detailed = more better for us. Luckily, we have a constraint: these meganormous rocks will NOT move.

Since we are mostly developers, and a little crazy, we thought it would be a good idea to look into other ways of creating these 3D models than to just model them, because it seemed like it would be a lot of work to model realistic looking mountains that would also be fun to climb on. When you model a cliff for the purpose of it being climbable it’s also easy to focus too much on “creating routes”. This would not be optimal for the realistic experience that we are trying to create as it would simplify the puzzle of climbing too much.

Scanning seemed to be the obvious answer as an alternative. 3D scanning is already used extensively in games and even movies so how hard can it be? After doing some research into methods of 3D scanning it came down to photogrammetry (based on photos) or LiDAR (based on infrared lasers).

A quick search later, we found that LiDAR could be adversely affected by sunlight. This would obviously be a bad thing for us as we would be going outside to scan the cliffs and boulders. Also looking at the prices we would much rather work with photogrammetry, since hooking a good outdoor LiDAR to a phone or drone would be quite pricey.

A major advantage of photogrammetry is that the model will already have a texture and not just a texture, but actually how it looks in real life. At least how it looked on the day that we did the scanning. Changes in weather conditions, like a cloud moving in front of the sun, will show up in the textures of the final models.

So it is decided photogrammetry is going to be our way forward for creating these meganormous rocks. In the next part we will go over running some photogrammetry tests and fundamental decisions we made for how we will be doing the photo scanning.

All around me, everyone (including myself) uses some form of stylized art for their game because you can accentuate what you want to stand out the most. You can also simplify, making art easier to model, set-dress and combine. If you do it well, your game would also stand out from all the others because of your style.

Knowing this, we started to define the right style for our current game, a sports / simulation / puzzle climbing game called New Heights. For the demo we found a way to stylize its look, using asset packs and retextured high detail models of rocks. Very highly detailed rocks actually, because we need all the features of the rock model that a player can climb.

This was one of the reasons for us to make our own 3D scans of rocks. Most rock models online are too low detailed and sculpting very detailed rocks by hand is very time consuming. But we now ran into the stylizing problem.



For the demo we used Substance Painter presets for generating nice textures, mostly based on curvature. But still, as you might see in the demo, it proved difficult for us to create the right amount of detail for the rocks. Whatever we did: the original textures always turned out much prettier and more interesting than whatever we came up with in Substance Painter. We also tried ‘stylizing’ the textures in photoshop with some filters, but it kind of came down to getting a lesser result than the original.



Then we realized something. We are creating a game that is based on one primary visual principle: show beautiful rock. Sure, also beautiful nature and nice looking climbing, but 90% of the screen will be filled with rocks / cliffs all the time! And somehow, whatever nature itself came up with, it is consistently nicer than our attempts at stylization. Of course! Being in nature is a huge motivation for many climbers like myself. So why would we take that away by stylizing it ourselves?

Now we are taking the raw, 3D scanned materials as the base for the entire style. Everything else has to fit the looks of the rocks. For now that means a fully, utterly realistic art style, even with our small indie team. And wow, this is going so much smoother! There are enough great asset packs out there that can help us (specifically, a big thanks to NatureManufacture). Suddenly we can look around on the internet, and check what photos we want to make a reality! It seems to be the perfect timing for us to make this realistic looking game.



Now we are facing the challenges as well. Animations need to be smooth and plentiful. Our climbing system will have to look even better. But at the moment we are mostly just relieved. Because coming up with - and creating the scenes, is quick, easy and looks great.