Sometimes you have an idea you can't keep out of your head. I've long been interested in the rare fighting games that break the conventions set by Street Fighter II, and wanted to take my own stab at a unique combat system. Kumite is a martial arts game that returns to a more deliberate, methodical style of play, inspired by games like Nidhogg, Virtua Fighter and Karate Tournament.
First I decided what to cut from the fighting game template. Combos, jumping, crouching, blocking inputs, anti-airs and overheads are gone. Special moves and unique characters are gone. It not only brings the amount of art and code needed down to more reasonable levels, but removes many fighting game concepts like the air and wakeup games in favour of footsies, where matches have the controlled pace of real martial arts fights. Damage is clearly readable and players know how many mistakes left they can afford to make in a match. 3D movement is intuitive and Ring outs mean players have to respect movement and positing much more than in traditional fighters. There are only three buttons used and no complex inputs.
The stance system is used to great effect in games like Bushido Blade, and keeps blocking as a tactical challenge. Guessing whether the opponent will attack high, medium or low is a raw 33% guess, but players telegraph their stance and are locked into longer attack animations than the average fighting game, so faking out a stance and countering opponent attacks is critical in Kumite. Evading into the foreground or background bypasses the stance guessing for a block and leaves the opponent open for a counterattack, but Evading without an attack to dodge puts the player at a large frame disadvantage and open to a Circular Attack.
I like to think all of this keeps much of what makes the genre great in a way that scratches a certain itch in seasoned fighting game veterans, and is more approachable to newcomers.
The problem is, fighting games are very difficult to make. Neither my art or code skills are up to the task. To put my energy somewhere, I created this design document. The process was satisfying, and a valuable exercise in systems design.
I'd like to make Kumite someday, once someone gives me a team and a budget. In the meantime, the Kumite document can be viewed on Github.