This submission is for the Atari 2600 game Steeplechase, a port of the arcade version. Simplistic though it may seem, we manipulate a jump route to beat the hardest difficulty and end input as soon as possible. We are the yellow horse.
Game objectives
- Emulator used: Bizhawk 2.8
- Aims for fastest time
- Uses hardest difficulty
So I decided on a whim to try a TAS for this and my first two completions actually lost to the AI despite me not missing a jump, so that made me dive into it further to see what was happening.
We use game 6 as it has the hardest horses and "random" hurdles with "random" distances between them. To jump a hurdle, you're supposed to set the jump height with the paddle and push the button to jump. There are 3 hurdle width but 4 jump heights, and this is a big deal.
You know they say all horses are created equal, but you look at me and you look at the rest of the field and you can see that statement is not true! Y'see normally if you go one on one with a 2 width hurdle, you're supposed to jump it at height 2. But I'm a genetic TASing freak, and I'm not normal, so I get to save frames and jump it at height 1. Then you add a 3-wide hurdle to the mix? Your chances of winnin' drastic go down.
Basically what I'm getting at is that you can save time over the CPU by using the minimum possible jump height, which they don't do. The longer you're in the air the more you stop moving to the right, so you want to be in the air as little as possible. Also, every jump you make changes the next spawn (this is the "randomness" of game 6)
For the one width hurdles we use the lowest height, so does the CPU, nothing special here other than to say that there is a window of frames where you can make the jump, and each of those frames change up the next spawn. In general I think jumping as late as possible is the best option.
For the two width hurdles we also use the lowest height as we can clear it with a frame perfect jump. This is one of the ways we stay ahead of the CPU; they always seem to use height 2 and we gain a handful of frames by not being in the air as long.
For the final 3 width hurdle, the CPU uses height 4. We use height 3 to gain a few frames on them but also to manipulate nothing else to spawn. By routing this specifically this way we can end input way sooner. I had tried quite a few jump routes and the second closest one was like 40 frames slower because it kept spawning one more hurdle whereas this submission's route doesn't and I'm not sure if there's room to squeeze out any additional frames with a different route.
As un-exciting as the game may be, I was surprised by the depth. If you were to play this RTA I think it would be very challenging; the AI on game 6 doesn't miss any jumps so if you miss one you're probably done, not to mention the frame perfect precision required to nail the lower height jumps.
