When we started creating Rainway we set out to build a service that was user-friendly and did away with manual configuration. Part of this process takes place in our advance game analysis engine, Mist(we have a running joke that all internal projects are named after water-related things). Mist is a powerful engine capable of scanning a users computer in just a few milliseconds going through millions of different game titles. This isn’t just limited to PC games, but also ROMs dating all the way to the 70s, making Mist as versatile as the PC platform itself.
Mist’s streamlined heuristics engine can locate known game installations quickly and piggyback on those results to find even more games. Once a game has been found, we extract metadata on it to help us identify it in our systems, which allows us to do everything from metadata populating to game optimizations. We can do this because our analysis engine is capable of identifying which DRM is applied to the game, what game engine suite was used to create the title and even apply patches to the game to help skip intros and force things to start faster.
Mist scanning a new computer instantly
To elaborate further on what makes Mist possible, we have to talk about metadata. Our metadata project goes hand in hand with Mist. We gather data from two sources, which in turn adds up to about 1.4 million games with about 2 million alternate names. With such a high volume of data, it would take around an hour for all of it to import properly. Included in all of this was the banners and screenshots we have to secure from five different sources. Getting all of this information to communicate with each other was quite a challenge for the team.
We ended up having to strip away at the names until it was the bare minimum. Yet even with this, we ended up encountering some issues with standalone expansions such as FarCry Blood Dragon and the StarCraft II expansions. Since we ended up having to simplify names a lot to prevent duplicates from occurring, the system we had in place didn’t recognize them as the standalone titles. This, in turn, was causing more issues further down the line with banners being assigned to the wrong game, and the search function not finding a game at all. To combat this, we had to end up adding several variants of simplified names. This turned out to be more dive for searching. At this current stage, we grade names based on similarity, if the associated game is a standalone game, the platforms the game was released on, and if it has valid genres associated with the title.
Mist and our metadata project are just a few examples of the powerful pieces of tech we’ve created to bring Rainway to a wider audience. As we roll into out beta, which is set to launch on January 20th, we are excited and can’t wait to share even more tech deep dives with you. Getting to share with our future user base is what drives us to keep creating more innovative technology.
Till next time!