Sometimes I manage to make the ghost move on mobile but I don't understand how to do it in a consistent way. Maybe there should be a tap area of 90 degrees above the ghost (from -45 to +45) to move it upwards, 90 to the right to move it rightwards, etc.
On mobile, you just swipe in the direction you want to move (e.g. swipe up to go up). You can also queue the move ahead of time. So even if you are nowhere near the intersection yet, you can swipe up and trust that the ghost will automatically turn once it reaches the intersection.
Seems the solution is to immediately leave and follow Pac-Man to the bottom-right (by alternating right-down) and chase him across the long corridor on the bottom. Keep following him and you'll just catch him, since he never goes for the power pellets.
Interesting! I found them mostly ok, but I was playing with a keyboard. They definitely remind me of the controls of Pacman games I grew up with where (as sibling comment notes), you have to decide where you're going before you get there.
I wonder if tweaking the input buffering or adding some frames after the turn where you 'snap-back' would help, similar to ghost jumps in Mario.
I grew up playing Pac-Man and Ms. Pac-Man in dimly lit arcade parlors and bowling alleys.
The controls in this version are, for lack of a better word, sluggish compared to the tight responsiveness of the originals on a four-way joystick or using a keyboard with MAME. Even when you press an arrow key to "move" the ghost, there’s a noticeable delay, almost like it’s polling for the key-up event instead of the key-down.
The code was written by Claude, unfortunately, and hence no controls were probably even considered, or no tokens were left.
I jumped in with the love in mind, too, but when I checked the source repository, and saw the actual source... and then the contributors... it was... I am sorry... it hit hard...
What does it mean for the ghosts to be "unable" to reverse direction? Isn't that just another way of saying that the original ghost AI was programmed to never choose to do so? There are, in fact, a lot of things they never choose to do. The whole appeal of this idea is to see if you can make better decisions than the original ghost AI.
pretty nice. I had a multiplayer demo a decade back where one player would be pacman and rest would be ghosts. and it would swap as you cycle. Whoever collects most coins won.
I think at one point Namco had a multiplayer Pacman Party Game which had a similar premise - whoever manages to eat Pacman gets to be him in the next round.
Why are people still upvoting obvious AI slop garbage?
1. Claude couldn’t do a proper fence algorithm for the walls?
2. Controls feel horrible.
3. It’s literally impossible to catch Pac-Man? You do not move fast enough and the Pac-Man AI is programmed for perfection so it does not make deliberate mistakes for the human player to take advantage of.
4. The tile based movement is not smooth, very stuttery.
Fine for a prototype, but we could do so much better. This is not a particular hard game to code up in an afternoon or even an hour if you’re experienced.
> 3. It’s literally impossible to catch Pac-Man? You do not move fast enough and the Pac-Man AI is programmed for perfection so it does not make deliberate mistakes for the human player to take advantage of.
It's not impossible. I just did. You just have to corner the man into an impossible situation. But I agree on the AI-slop or lack of quality production.
Interesting! I found them mostly ok, but I was playing with a keyboard. They definitely remind me of the controls of Pacman games I grew up with where (as sibling comment notes), you have to decide where you're going before you get there.
I wonder if tweaking the input buffering or adding some frames after the turn where you 'snap-back' would help, similar to ghost jumps in Mario.
The controls in this version are, for lack of a better word, sluggish compared to the tight responsiveness of the originals on a four-way joystick or using a keyboard with MAME. Even when you press an arrow key to "move" the ghost, there’s a noticeable delay, almost like it’s polling for the key-up event instead of the key-down.
Once I got that figured out, it was mostly a race to catch pacman along the long straight path at the bottom edge with the ghost's superior speed.
EDIT: Found it.
https://en.wikipedia.org/wiki/Pac-Man_Vs%2E
For reference - it's a Gamecube game where the other players play as ghosts released all the way back in 2003.
https://www.smbc-comics.com/index.php?db=comics&id=2736#comi...
1. Claude couldn’t do a proper fence algorithm for the walls?
2. Controls feel horrible.
3. It’s literally impossible to catch Pac-Man? You do not move fast enough and the Pac-Man AI is programmed for perfection so it does not make deliberate mistakes for the human player to take advantage of.
4. The tile based movement is not smooth, very stuttery.
Fine for a prototype, but we could do so much better. This is not a particular hard game to code up in an afternoon or even an hour if you’re experienced.
I caught him! So it's definitely possible. You move a little bit faster than he does so you need to anticipate where he's going.
It's not impossible. I just did. You just have to corner the man into an impossible situation. But I agree on the AI-slop or lack of quality production.