XXX

[Shrekt]

I wish someone could have skills to code a perfect AI that wins 100% of games, to watch the replays and see how they play against each others.

This is as good as it gets so far:
https://www.youtube.com/watch?v=fpIe6lNM1NE

Here's the codes for this AI
https://github.com/Innuendo1975/haxball-ai


Here is a video explaining how Reinforcement Learning was used for the Haxball AI and the challenges involved. Skip to 2:30
https://www.youtube.com/watch?v=Ma_MopOXLCg&t=1543s

[Yannex]

i'll fix myself... nowdays no one says the mechanics are the same. Maybe in the beginning of HTML some people didn't realize it, but i dont believe this is the case now

html has the same mechanics.

i was one of those who said at the start of html that is wasnt different which i admit was stupid af since i played flash for years and html 5 min befor i said its the same...
Hannes if you really are that bad to understand by this time after years of playing html that its different idk what else to say

0 extrapolation totally different
0-999999999 extrapolation makes the game feel a bit more like flash maybe BUT its opens a whole lot of other problems that make the game still feel different

[Hannes]

Hannes if you really are that bad to understand by this time after years of playing html that its different idk what else to say

that would prpbably be the best

[saviola]

I think what Hannes is referring to is the answers of Basro and pilaf in this issue. If we believe basro (and I don't see why he would lie – it is of course possible that he missed something when porting the code from Flash to JavaScript), then the logic (the physics) are exactly the same, they were unaffected by the change from Flash to HTML (that would be easier to belive if the old replays could be viewed in the new engine and vice versa maybe).

The engine changes the rendering, so the stuff that we see is slightly different, and the engine determines how our inputs are treated and passed on to the game – and there is nothing Basro can change about either of those.

[B4D4SS]

I think what Hannes is referring to is the answers of Basro and pilaf in this issue. If we believe basro (and I don't see why he would lie – it is of course possible that he missed something when porting the code from Flash to JavaScript), then the logic (the physics) are exactly the same, they were unaffected by the change from Flash to HTML (that would be easier to belive if the old replays could be viewed in the new engine and vice versa maybe).

The engine changes the rendering, so the stuff that we see is slightly different, and the engine determines how our inputs are treated and passed on to the game – and there is nothing Basro can change about either of those.

The thing is that I know exactly why the gameplay is different. The movement on FLASH is slower, and its because for whatever reason the FLASH engine added a little (like 20msec or less) Input Delay to all actions (movements and shoot). This tiny alteration made a massive difference for the gameplay especially at higher levels, people with a keen eye could see this.

Now it must be extremely simple for Basro to just check the Input Delay number that the FLASH engine gives, and add the same number to HTML through a synthetic delay. This feature to me sounds like it would be very simple to add to the game, but you know much more than me about that so could you tell me if its possible in your opinion?

If its true, then at that point we just need to convince Basro of the logic behind doing it, and because the truth is on our side it shouldn't be too hard.

[saviola]

Adding a synthetic delay should be easy, convincing basro that it is good might be the harder part. Plus some people (especially on Windows 10) have an additional input delay caused by the browser / OS on top of that, so it will be hard to get an even playing field either way.