This NO HANDS Shortcut Track Sort of Broke the Game Again...

[@JaegerSeven0]

Love the wholesome and instant response of both Kan and Kosmo at the casual mention of funding the game "buy our friends game right now"

[@R.B.]

You should use a timing event so that you would normally hit a hammer or other something similar if you are AFK, but if you brake for a second at the start it will clear the objects. It would still be an AFK map for driving, but it won't reveal the secret.

[@nmf0]

I don't know if this is still the case, but I rememer there was an issue that made that impossible: So, unless that's fixed, the hammers don't reset when you reset, they keep going, which means the hammers will be in a different position every time you reset.

[@soup-flavored-soup6613]

Use some kind of other falling object, like something manually placed

[@dibaterman]

Unity physics is deterministic-like... if you implement it in a way to be deterministic. Without going too deeply into it, the short story is you keep physics operations out of any kind of update loop, that includes the recommended fixedUpdate. I have gotten consistent results by either triggering physics through Events or through Coroutines if needed. But for some reason using anything in update loop is not going to work right off the top. Maybe if in the project settings a timing can be changed for fixed update... but there may be things you want on fixed. Unreal however comes optimized for replicating deterministic physics via its visual coding set up. So you don't need to know about using delegatess, events or coroutines to get the same (and arguably better) result from Unity.

[@progamers-ug4ik]

physics is conected to game fps.

[@dibaterman]

@@progamers-ug4ik Yeah, but again it's about how you code it. In my own project to get consistent physics results in a 3d space I used ray cast during update and execute physics through events. So on input event the person jumps, the physics related to the jump will be exactly the same all the time because all physics are calculated in that frame. There are no other calculations made until my conditions are met. Again it isn't 100% but it is 99% consistent. I send a ray out to find the distance of the player to its original position. (You can do this also by collecting the ground vector or if using navmesh getting the point on frame of jump where the agent was on walkable mesh. I then have a state machine which cares about jumping, in air and falling. at the point of in air, that means the player has reached their jump maximum and a ray is sent to the likely fall directions of the player to figure out how long until the player will reach the ground based on their trajectory. I made that for animations but it works well for picking the frames I want to add physics for. I hope that clarifies what I mean by deterministic-like.

[@upsidedownairline9388]

From my experience it's actually rather simple to understand - the Δt parameter used by the physics engine every update cycle determines how much objects move, rotate, and accelerate in that physics frame. Usually that parameter is the actual measured time since the start of the last update cycle, which has some random variation to it, and that results in tiny but compounding differences between runs. Using a constant value regardless of framerate will result in perfectly deterministic physics, because it removes those random variations.

[@nessa6859]

Ironically the first time I played Zeepkist online, I got your first track, was AFK and got the no hands finish the first time, first place finish. I've only finished first once or twice since lol

[@rikkitje]

Thank you guys for being friends and making such fun content together🤟

[@dannypipewrench533]

That is exactly what I wanted to say.

[@ApotheosisStone]

Maybe a good idea for a future track is "Anything you can do I can do better". Take a track concept someone else did and improve it.

[@Mice-stro]

Why not just track evolution, they did it in trailmakers

[@mammalianmolasses4724]

@@Mice-stro they’ve already said several times why they won’t do track evolution.

[@ammon46298]

@@mammalianmolasses4724 Would you be willing to give a quick summary for those of us who haven’t heard their reason?

[@mammalianmolasses4724]

@@ammon46298 passing the files back and forth through steam is a massive pain

[@KoolKid29092]

11:21 the TRUE HANDS FREE run!

[@xugro]

I think non deterministic physics is caused by different floating point units in different processors and you can mitigate this by using software implementation ( softfloats ) but it's a lot slower.

[@vibaj16]

non-deterministic physics is caused by the physics engine using the time passed since the last frame in its calculations. This allows the speed of things to not depend on the framerate

[@flabort]

Actually, this was the 4th uploaded version of the track. And the hammers worked for me in the second version. I got hands free until the inconsistent part, and then took over, got the wallride, and got my first ever first place. But, unfortunately, it's not the version that went into the video.

[@Chosen1Creator]

I guess one way to make AFK tracks work is to have normalization mechanisms like walls or channels that point the car in a specific direction and boosters that bring the car up to a specific speed. Having these placed after every obstacle should eliminate compounding variances.

[@airzocker3046]

0:49 small correction, the old unity physics (NVIDIA PhysX) isn't 100% deterministic, at least how unity sets it up, but the new physics package or havok physics, which are based on the somewhat new entity component system are (or at least can be pretty easily) completely deterministic :D

[@Cheesius]

I never got that it was a hands-free until watching this video. I just tried it, had to play about a dozen times before the hands-free worked, but that's neat.

[@randommm-light]

I love the hands free tracks.. I make them in Trackmania w only the accelerator and brakes, no steering. The trick is to build the hand free “yeet” path first, all the way to finish, then backfill with plausible track. Cuts down on all the replays. A huge aerial jump w multiple height landing zones is a nice gimmick.

[@jmval81]

I wanna see u guys build tracks , to look as nice as possible , I know u all are creative, I know itd take way more time but itd b sick theme

[@amusedapple4933]

?

[@amusedapple4933]

1 year later i know but im curious

[@crodolog4134]

Unreal is generally worse than unity for non shooters due to lack of tuning. The reason most dev teams use UE over unity is because it’s cost is a % of the profit while unity is per seat which quickly ramps in cost.

[@10thletter40]

The joy when you got it made my day 😂

[@tylerhouston69]

Building deterministic physics is nuts. It's why a lot of people just license microsoft's physics (Havok) which is about $50,000-ish. TLDR; deal with janky physics cause slightly chaotic outcomes make it more entertaining.

[@vibaj16]

Making deterministic physics is easier than making non-deterministic physics. Anyway, Havok is deterministic? I wonder if Rubikon is deterministic...

[@tylerhouston69]

@@vibaj16 For people with the funding to build an engine, sure. For people who've never done it before and would have to spend years tweaking it, nah. Ask Valve.

[@vibaj16]

@@tylerhouston69 For anyone, deterministic physics is easier than non-deterministic physics. Non-deterministic physics is a consequence of a feature that makes things go the same speed regardless of the fps. It's much easier to cap it at like 30 fps and just let physics be slower if your computer can't handle it.

[@tylerhouston69]

@@vibaj16 that still requires the average indie dev to develop their own physics which is not happening lol.

[@Voriki2]

Track idea: co-op, everyone works on 1 map. Each designer gets their own quadrant of the map to work on. Say there are 3 designers(Kan, Kosmu, Dapper... but I hope Scrapman or other designers also partake. Each designer gets a blank map template where the start and finish are placed beforehand. Designer 1 gets Start A and Finish A. Designer 2 gets Start B and Finish B, but Start B is at the exact same location as Finish A. So each designer makes their track one after the other. Let's say each has to build a 30 second-ish track. But there is no discussion on color or themes, so you'd go through 1 big track in 3 vastly different styles. Then another innocent person(such as Yannic) takes the 3 text files of the 3 small tracks, and merges the elements. And then it will be a surprise for all 3 designers what the end result will be until it gets unveiled in the lobby. Each deleted start/finish also becomes a checkpoint.

[@SpeedyVibes27]

love your vids

[@slomopanic]

That was a great watch Kan :D

[@mrprotscg]

Best episode yet

[@christopherrogers532]

It might be fixable with some extra roads and props to help with the physics problems.

[@MikeHarrison3266]

The proof, the way to a win is no touching and playing with yourself.

[@ihiminen3393]

today in zeepkist -KaN 2022 (soon 2023)

[@ITpanda]

Would be interesting to have a way to align people to an exact spot mid track and boost them at some stupid speed off something that normally people would avoid, so that it yeats them across the track in a hands free manner. So half hands free track. Maybe make the first part difficult so people would be less inclined to risk what would normally be a very stupid play. The rest of the track could just be a long back and forth canyon run.

[@Ricky-Davis]

According to Google, Unreal is not deterministic.

[@SpeedyVibes27]

yup

[@kennythegamer1]

The Physics engine that Unity uses most likely uses discrete collision detection, which gets worse with lag and the speed of the object and is most likely causing this issue of single player vs multiplayer inconsistency. Also, the collision response probably isn't made to balance out the variable depth an object could be into another object; that would cause the weird bounces.

[@vibaj16]

i think there's an option to use continuous collision detection

[@kennythegamer1]

@@vibaj16 If it's there, most games probably don't use it, though.

[@programmeralexgames4953]

i believe you get different results because the interval for physics check gets offset it isn't realigned when you start.

[@vibaj16]

it's because the engine bases its calculations on how long it's been since the last frame, so the speed isn't dependent on the framerate, but this introduces the randomness

[@kollinspencer847]

nice love it

[@DefenderTerrarian]

If this was filmed on a Tuesday, Tuesday is the day Steam's servers reset, it causes problems for connectivity for an hour or so.

[@frankraymondmichaels332]

I bought the game because of these videos. Still a noob but my favorite game.

[@jnnmlchr91]

If Hyce some how found out a way to do this with trains

[@mdpl_]

it's a shame that you didn't add into the video the quick look of the RoanC's ice campaign challenge when it was on X02, that would be funny

[@Kodiak01]

Actually had no idea it was an AFK map ^^ I was one of the first to complete but I did it the long way 😁

[@7610143]

5 minute track suggestion Dragon ball's "Serpent Road of the afterlife"

[@bread0237]

👍Bread

[@2dr1fty]

425K SUBS? Last I remember was 100k

[@darrianshaker7121]

Back to the past (future)

[@Saxquatch_]

Imagine if someone came on afk and got first while being afk

[@ImNotGam]

Was this recorded last Tuesday? Steam has maintenance every Tuesday around noon Pacific Time so you might want to avoid recording around then.

[@ArythNeon]

I looked up "does Unity have deterministic physics and the first result literally from Unity was "Unity Physics is a complete deterministic rigid body dynamics and spatial query system" It is deterministic

[@Skirakzalus]

Same concept as the other track. Same outcome as the other track. That's deterministic.

[@TBdoessomething]

Wonder if this is based on the automatic track i suggested

[@icendox]

I wanted to say something funny ... My sim racing setup is compatible with the game

[@proxithermoblast643]

... buy zeepkist

[@andeeznation]

Its a broken game because a full afk should always produce the exact same result. So dev needs to fix something there

[@toolng1798]

It might be better if you just steer and aim it

[@plx1212]

Number 28th, lol :P

[@claytonverheul9662]

12 sec

[@ihiminen3393]

third?

[@OpGamer-yk1dw]

6th

[@claytonverheul9662]

Second

[@plx1212]

Twentyeight

[@1234winnersplacenonoobsallowed]

CTR = crash team racing