Issues & Errors
Issue 01
Graphics Cards & FPS
Throughout the course of the project a number of pertinent issues materialised causing significant unexpected delays. It is entirely expected to combat various modelling, texturing and assembly issues throughout the process of creating a game environment. This project, however, encouraged a whole host of unexpected and unprecedented issues.
​
The first major issue to appear revolved around a significant frame rate dip when using certain machines. Throughout the course of the project, work took place on two separate PCs of drastically differing specification, and surprisingly the environment ran significantly better on the lower end machine. This could be for numerous reasons. Firstly, the lower end machine contained a gaming graphics card, while the higher-end machine contained a workstation card. On paper the workstation card is more powerful, but when running the game environment it performed significantly weaker (sometimes with as much as a 30FPS difference). After researching online it appears the gaming graphics card drivers are optimised to deliver maximum performance for games, while workstation cards are optimised for rendering. Thus, it is expected that games and other real-time applications will run better on gaming graphics cards, but the difference was staggering. Unfortunately this issue was unsolvable, so I decided to optimise the card to work on both cards efficiently.
A number of approaches helped when optimising for both cards. Firstly, I dramatically decreased the amount of real-time and stationary lights in the environment. All atmospheric lighting (skylight and directional lights) are entirely baked so as to improve performance. But with all of the effort allocated to dynamic elements within the environment, is was not ideal to bake every light as these dynamic elements could then not cast shadows. To combat this I simply added smaller movable lights around dynamic elements to cast a false dynamic shadow.
​
The most recent version of Unreal Engine contains significant changes to level of detail systems. In hindsight, these features would have been incredibly beneficial when optimising the game for weaker cards, but time did not allow for exploration of these new features.
In hindsight it would have made sense to target just one graphics card and optimise for that, as opposed to trying to please both cards.
Issue 02
Over ambitious
After commencing this project it quickly became clear that I had embarked on an impossible task. In some ways this was intentional - The aim was to commence the environment during the final masters project with a clear intention of continuing afterwards (When discussing the art bible, Fig 4.12 depicts the intentions for the overall environment, while Fig 4.30 narrows down to the specific assets within this project).
​
However, the approach taken was very linear. I proposed commencing at the entrance of the amusement park, and reaching deeper into the park as time progressed (Fig 26.1). In retrospect it would be more beneficial to initiate the project from a more important asset, and work backwards (Fig 26.2). Such a method would result in more interesting assets much earlier on in the process.
