One of the most significant challenges that The Game Band faced with the Apple Arcade launch was ensuring that it was compatible and played beautifully across iPhone, iPad, Apple TV and Mac. “It also gave us some peace of mind that if we needed to trim down the memory usage even further, we could do it in a piecemeal approach (e.g., target animations or meshes next) without worrying about needing to overhaul too many systems at once.” In practice, The Game Band found that it was “pretty painless” and “a marked improvement over the previous Asset Bundle workflow,” according to Sorg. The Unity team helped The Game Band achieve a staggering 60–70% reduction in memory usage – and that’s after the new asset management system let them add an additional 500–700 MB of deck textures. Integrating the Addressable Asset System was a key factor in the studio’s ability to attain an incredible reduction in memory usage. “We ended up marking just the lowest-level assets as addressable, and it turned out that, to get our game running on iPhone, we only needed to manage the card texture memory.” However, it proved to be a bit unwieldy,” explains Sorg. “We started out by attempting to mark all our highest-level assets (the ones that reference all our animations, meshes, textures, etc.) as addressable, which did allow for loading all our scenes asynchronously. The Game Band team experimented to find the right way to deploy this powerful feature set in their game, knowing that they could implement the system just where it was most helpful. Without their support, this fix would have been very difficult to come by.” “One key area they were able to help with was identifying an issue with how we were setting up animations and playable graphs. Our biggest performance optimizations to cross the finish line were based on fixing bottlenecks they helped us identify,” says Sorg. “Unity helped walk us through profiling and provided context for many of the readings we were seeing. CPU times were also more than halved – on low-end devices, performance improved from around 10 fps to 24 fps, eventually hitting their target launch framerate of 30 fps. The team pinpointed issues causing slow startup times, which helped The Game Band to dramatically accelerate from around 60 to less than 10 seconds on lowest-end devices. To boost GPU performance, they used the Shader profiler to recommend changes that would reduce the cost of VertexColor-type shaders. The team recommended integrating keywords to activate and deactivate room transition calculations as needed, as well as culling off-screen character and card elements using CullingGroupAPI. The Memory Profiler identified vital areas where memory usage could be improved, leading to recommendations for better systems to compress and downsize textures, changes to the settings on imported meshes, and performing regular resource audits – together, these suggestions helped the studio reduce memory usage by 60–70%. Also, for testing and debugging, we relied on Unity Remote heavily for Apple TV,” says Brandon Sorg, who acted as both the gameplay designer and technical director for Where Cards Fall. “The Profiler and Memory Profiler were both huge benefits to identify performance bottlenecks. Testing on a 2GB device, they found that Where Cards Fall sometimes crashed at startup as the game loaded, and they also identified key CPU performance issues that would impact gameplay on higher-spec devices. When Unity engineers looked under the hood, they saw that it was essential to enhance both CPU and GPU use to improve the game’s performance on both high-spec and lower-spec devices. The first front to tackle was core functioning.
0 Comments
Leave a Reply. |