[TAS] Trackmania A06-Obstacle 27.62

This run was found through the brute-force random algorithm through feeding the script the faster run it generated before. This one can be surely improved a lot, not only at the end but in the whole run too. I'm also feeling that with a longer search, this should get optimized as this is only 0.04 faster than the current WR by oli10. No manual changes were done in order to improve this run.

[@logictm]

cool, would love to see what my time would have been without the landingbug

[@j_ustanoob]

he didnt lose speed?

[@wavingbuddy3535]

@@j_ustanoob if you could read his comment would make sense to you

[@AlexanderB41]

Epic man, every wr is worth watching

[@lixea7618]

this man is a living legend

[@ora_veugle]

Hi, i think it would be a good thing to tell in a formatted way how the TAS has been done, like "Improve Run", or "Bruteforce", "Hand Crafted" or even "IA" in the future why not, or others I think it isn't the same job at all so it would be convenient to differentiate them from each others (btw i know that we can use them all in a same run, and i know that it require a lot of work for all, it's not about ranking TAS methods)

[@musti789456123]

damm sick what you done here. pls make a update video of your tool demo video!!!

[@LoL-pp5yx]

I was thinking we could try an uber at the first checkpoint towards the second, do we not have enough speed to realise an uber ?

[@chapstickbomber]

If it's TAS why not flip the other two ramps to end up with enough speed to jump the dip entirely and get a clean jump on the last ramp into finish?

[@ChessAnalysis161]

i think its impossible to do that cause even with the most speed i could gather i couldnt get past the middle of the pothole

[@Poutrel]

have you tried combining multiple runs, like the WR and a run that have an insane specific section? I can imagine the difficulty of fixing the alignment of both runs when switching from one to the other though

[@donadigotm]

This isn't really possible at the moment, aligning the cars to simple things like position, rotation and speed is very hard, let alone making most properties of the car identical. Even if you succed to some degree, the next section is almost guaranteed to fail due to the chaotic-ness of the simulation. I haven't tried this though so maybe it could work though, idk.

[@Nixed16]

wouldn´t a uberbug from second last cp jump into last cp be faster? by like hitting the top of the cp downwards back onto the road. Maybe even nosebug setup from cp bounce down idk

[@zxnull4397]

Link destroyed an old tas run thats crazy

[@MrCel1000]

Do you think a flip in the left side of the ramp is possible/faster?

[@AntonPogonin]

Yes, it's faster and possible

[@ilyaasbaderoen1898]

Link has a 27.50

[@KirbyTheKirb]

bullshit map, I hate that ending part. I can't get that weird bounce.

[@iocto._.]

is this a genetic algorithm you did?

[@donadigotm]

This isn't even a genetic algorithm. It's pure brute force of randomizing some steer values (with some %) of the run and just checking the resulting time.