Summer 2017 Challenges Theme: Landscapes

My only attempt at landscapes in the Unreal Engine was back in October 2015 when I attempted to create a Martian environment from a small reference photo and artistic interpretation/imagination. This is never a good idea because if you’re trying to recreate something that exists, then you should surround yourself with a room full reference material and take your own (if possible). The results weren’t too bad:

However, there were several things wrong… well several things that could be improved thanks to developments in the Landscape toolset and some new knowledge including new tools and workflow suggestions.

  1. The terrain was hand-painted, and whilst it actually does resemble some low-lying dunes, up close the terrain form was pretty boring. This could quite easily be improved by examining an external terrain creation tool (World Machine).
  2. The material applied to the floor was poor. Basically I created a couple of tiling rocky textures from reference imagery. The main issue here was the resolution… they were only 512 x 512 which looked fine from a distance but awful underfoot – so not suitable for a game world at all. Additionally, the normal and roughness maps were created in Photoshop using the 3D filters – so wildly inaccurate! I need to find/create better textures for the landscape materials… preferably 2k+.
  3. The rocks were fine in the dim light but didn’t hold up well when in near-sight of the camera. Also, they were not great for performance because they didn’t have LOD applied to them. The original Martian landscape was created using some Unreal’s experimental features (at the time), the Landscape tools now include much better support for creating and editing ‘foliage’ on terrain.
  4. The rocks didn’t blend into the terrain. I tried to cover this up with low light, but if this scene is to work as a game environment, it needs to hold up better under harsher/different lighting conditions. Maybe even a day/night cycle.
  5. I used static “baked” lighting to create the best static scene I could. However, as above, this wouldn’t work as a game environment because one would expect the weather/time of day to change and for this, all the lighting that influences the terrain should be movable or “dynamically” lit.

Identifying and setting Challenge(s)

It’s good to have targets with your experiments to keep you focused. I’m going to set myself 2 challenges as part of this ‘Landscapes’ experiment:

  1. Recreate an example that combines multiple techniques from tutorials
  2. Create another Martian landscape that is an improvement over the 2015 attempt

With each challenge I’m going to look at the 5 areas for improvement listed above and use these to evaluate my performance. In some ways it’s like being a student again but this time I’ll be in control because I’ll be the one establishing the framework and assessment. Just like a self-initiated degree project!

Here are some good guidelines to creating your own challenge(s):

  • Approaching experiments in this way is far better than just “mucking about” or simply experimenting with the tools. By all means do a little of this first, but when it’s time to formalise and evaluate how much you have understood, introduce some discipline and establish a challenge with scope, aims and evaluation criteria.
  • It might sound boring but it’s not a great deal of extra work and will encourage creativity (and potentially innovation) because you are working to restrictions.
  • Keep your challenge simple and quick (limit the scope). You don’t want to get bored and you want to be able to evaluate your work as quick as possible. Break a larger task into smaller challenges each with it’s own unique set of goals.
  • Stick with it. Try to stay focused. There’s usually a lot to take in (unless you have a very focused task – in which case it should be quick to complete).
  • Don’t muddy the waters.┬áIf you fancy trying something beyond the scope of the challenge you are currently working on, write it down and set it for a separate or the next challenge.
  • Feedback. Whether you’re happy or not with the end result, share it. It’s easier to hear feedback on something you didn’t think was your best work, but be prepared for constructive criticism even on something you consider “finished” or – and I hope none of think this too often – “perfect”. Something can always be improved and all comments will help you evaluate the success of the challenge and identify areas in your work to improve next time you approach a similar task (just like I did with the above evaluation of my Martian landscape).

But before I tackle any of the above, it’s time to explore two external tools and resources that will make the creating landscapes so much more interesting: World Machine & Megascans Library.