Devilish is a breakout/ pinball type arcade game. You control two paddles and try to continually advance by breaking blocks and killing monsters.
This run utilizes hardmode, which saves several seconds. Hardmode (mostly) only reduces the amount of seconds you have per level, and thus the number of frames counting your bonus points takes at the end of each level. Bosses also have more HP, but the time saved from bonus ticking down makes up for it.
The red X blocks temporarily disable your red panel, so we must avoid them.
The main technique we can use is to flip the paddle while ontop of the ball to send it backwards. Other than that, you don't have much control over the ball. It does not bounce if you push the paddle against it. It only bounces if it is coming *towards* the paddle. This makes it difficult to move right and left, as the game biases towards downward momentum.
I manage to skip over one gate and destroy it from the opposite side. Unclear how this glitch works. The paddle is exactly 1 pixel off, but likely it also requires the bounce angle to be perfectly aligned to go as straight as possible in order to clip to the other side. Controlling the bounce angle however is not trivial. It seems to be based on a rotating sin-wave that only advances after each bounce. The boss can't die while attacking, so his HP actually underflows.
The clock is just bonus points. Doesn't seem to really matter what you do, but might as well play around a little.
There are a few paths that branch and lead back. The top path seems to be faster mostly because changing the scrolling direction is awkward to trigger. Turning the scrolling once, instead of twice, is therefore seemingly better. The tree boss is easy to do in one cycle.
There are a few paths here. The cannons are faster than using the dragon powerup because the game normally only can get up to 30 FPS... But in the cannon, you hit 60 FPS until you land. This essentially means you move twice as fast while shooting out of the cannon.
The optimization in the boss section (break all the blocks) is ok, but could probably be improved. Still, this was one of the better attempts.
The cannon in this stage isn't running at full 60 FPS, because of all the overlays. Even so, it's still slightly faster to take the cannon than the break through the blocks. Note: In game clock only ticks on non-lag frames, so you can get a lower in-game time by breaking the blocks instead.
The boss is straight forward. You can only hurt the left twin.
I take the shorter worm, because it places the ball closer to the right side of the screen.
The bounce ball bonus game ends when all the balls are dropped. You can use the paddle to flip them backwards, ending the minigame faster.
There are a few forking paths here. There is a promising looking path, but you'd need to be able to evade the ball grabbers while going directly over them. I don't think it's possible. So this path was chosen.
The boss fight is a little tricky at first, but once you get in the sweet spot, you just flip the paddle back and forth to juggle the ball. If you *don't* do this, the game will make the ball untouchable and kick it out.
Early Input End
Immediately after the boss dies, I end input. The game plays itself, and reaches the final goal, and the ending scene. The savings is something like 30 seconds because we don't need to count the dying animation.
If you win on hardmode, you get to see a very special bonus scene where a knight hugs a princess. I have no idea why this hug is locked behind hard mode.
Final notes - Possible improvements to look into for the next person.
Despite the game having some lag issues, there is not much you can do about it on the micro-level. The game mostly doesn't really support any FPS between 30 and 60. If the CPU is overloaded, it drops to 30 FPS immediately and mostly never recovers. So I don't think there is any point in trying to optimize lag frames. I think the lag decreases when the ball is coming towards the bottom of the screen. This probably indicates something about the collision detection algorithm, but I don't know how you might take advantage of this.
I don't see a lot of room for improvement left in this game. I think the biggest gain would be learning how to skip the skull gates consistently. The bosses are pretty good, but I wouldn't be surprised if it's possible to save more frames on them. These were my best attempts, but different bounce RNG going into the boss could open up a faster option.
Thanks to Moonberry2 for providing me an example run of this game. It helped set my expectations of what times I needed to beat.
Memory: Replacing file with one with unknown rerecord count since count was bugged.
Memory: Optimization appears good. I asked about 2 player and the author responded that it was unlikely to save time and honestly, from looking at it, I agree.
While at first glance the game appears similar to breakout/arkanoid, this game has a flavor of its own and presents some interesting strategies and unique mechanics. I personally found the run rather entertaining and the audience did as well.
Accepting to Moons.