23 Samurai battle it out against 23 Ninjas. Although it's still in a raw version it's shaping up pretty good even with some animations missing. Wait till I add blood effects, different weapons, dismemberment and enemies with horses. It's gonna get so awesome. Every character is a unique entity with his own stats/skills and style familiarities. For this video everyone starts with the same stats as to be balanced so the outcome is always random. Everything you see is AI I don't do anything.
Although it is a stealth game there would be times where the above would be necessary, of course on a level you would have to take care and think where this battles would take place. Kinda like in Total War games but instead of wide open areas you will have small areas and in cities/villages and other landmarks. So let me give you an example, you have your ninja been spotted by 5 samurai, as you already know without having first observe your enemies the battles would be very difficult so what will you do then? Run inside a house where you have strategically placed some of your own characters and let the slaughter begin!!!
I had a friend come over and we placed bets everytime on who is going to win, it was so freaking awesome, kinda like multiplayer without actually playing . I'm going to make this a mode in the final game and you would unlock characters by playing the main story.
Spend the whole week restructuring most of my scripts and optimizing them. As you can see I'm pretty happy with the results, given that prior to optimization with only 5-6 enemies on screen fps dropped to 1-2 and now it can stand 80+ agents. Most of the drop it seems it happens when all of the enemies have in their sight the player because it's based upon the physics engine so it's not NavMesh's fault. Anyway i think it's pretty good results. Of course in the final build they might not be that many on screen because of draw calls and other reasons but it's nice to know that the pathfinding can handle much without killing performance.
I've also spend the week experimenting with MoCap to streamline the process and sped up things for the game. I had some good results with it but there are some tests left with it and it will be ready to go. More on that later...