Stk-code: New physics engine causes a little too much bounciness near the end of snow peak

[psypherium]

New physics engine causes a little too much bounciness near the end of snow peak

You will sometimes fall straight off the track

[auriamg]

a video capture would help a lot so we know exactly where/how

[psypherium]

https://youtu.be/gAlW_lv77aU?t=27

Also, the "back to race" button is tiny?

[auriamg]

Which version of STK is this? I just tried a few times using latest git version but was unable to reproduce this issue

[psypherium]

git

You must be drifting for the bounce to occur. The new physics tweak causes some kind of a readjustment of descent speed every second or so, causing drifts to temporarily "stick" in the air. Try jumping off the mountain in the story mode overworld while drifting, you will see what i am talking about.

Video capture helps a lot so you know exactly where/how, Video capture involves many extra annoying steps for me until this is fixed: https://github.com/supertuxkart/stk-code/issues/2995

with that bug fixed I could have initially just uploaded video for all 3 problems very easily without having any confusion.

I'm not complaining, I'm just explaining why I will generally be explaining bugs instead of posting video.

(also, I only noticed the "back to race" button bug through watching the video, more bugs are fixed more quickly with more video.)

(Another edit: with that bug fixed, it would be much easier for me to record 4k footage to donate to your video editor to cut into trailer footage for the next version.

In any case, you could hook me up with whoever manages that if you like. I could record some more footage of the donation tracks and send to them. Something like this but with GUI disabled: https://www.youtube.com/watch?v=rHZsP7R07b8 YouTube says the video is CC-Attribution, but that's only because that is the only option they have, You are totally free to use any of my STK footage without attribution, download it from youtube with this https://github.com/MrS0m30n3/youtube-dl-gui most of my youtube videos can be downloaded with open codec formats, usually vp9/opus)

[Alayan-stk-2]

I can reproduce the issue in Snow Peak at the spot shown in the video.

[hiker]

I will try to reproduce the issue, but tbh, I am very certain that this is more a track design problem. What (very likely) happens is:

the kart is too fast to stay on the track, so it starts flying forwards with high speed. At the same time it starts falling. After a while, the downward speed increases, till it falls fast enough to hit the ground - at which time it would very likely crash, which is bad in terms of game design ;)

So at this stage either because our extremely long suspensions kick in, or because we detect the upcoming crash after a fall and slow the vertical speed down to avoid the crash, the kart's downward speed is reduced (to close to 0). But the forward speed is still there - so the same thing happens again: the kart will move forward fast, starts flying, till the downward speed (gravity) is high enough for it to hit the track, ... and repeat.

So, there are a few options:
1) We could accept a crash. This might cause the kart to suddenly rotate (if one side/edge hits the ground first); and certainly to slow down - which is not fun at all :). And the kart will certainly slow quite abruptly - which isn't fun either.
2) We could artificially slow down the forward speed as well when we slow down the downward speed - while we avoid the crash compared with 1) above, it is still not fun to suddenly see the kart slowing down.
3) We modify the track to work as expected ... which would be either:

• keep the kart connected to the track (so the track should start with a gentler slope, which then increases as the kart accumulates more gravity induced falling speed). This way the karts should be able to drive, which sounds fine; or
• make the kart even steeper, with a flatter landing area. This way the kart would clearly fly, and only on the landing area becomes controllable again - which sounds fine as well.

So imho track modifications is the way to go - what would be better? I don't know ;) Esp. considering being able to go in reverse, i.e. up the hill :)

[Alayan-stk-2]

I don't agree it's a track issue.

The core issue here is quite clear : assuming that 0 is the good vertical speed to slow down to is wrong when there is a slope.

[auriamg]

Fixed in svn with track changes