ZCT 268 - Penrose

[Frieshansen]

It seems to be good practice to start a thread for new tracks. So here it is

The intention was a rather slow track with two equally fast sections.

I'm looking forward to the replays and hope everyone enjoys the track!

[Cas]

Alright!  Just posted my first lap   I see several cars are competitive here!

[alanrotoi]

So jumper and tricky. I like it!

[Erik Barros]

a beautiful track, very good work with the terrain

[dreadnaut]

My bad for not remembering the issue with splits immediately after the start line. But hey, the discussion that came up was great! So, where do we stand?

The rules state: "A replay must be a complete lap of the circuit."  A complete lap of a circuit starts and ends at the finish line.

This immediately invalidates any replays which end by touching the initial split: the car loops back to the start of the track, it never reaches the end. Mark's "flying from the loop to the split" also falls into this category.

Now, the ending of Mark's replay could be changed slightly, so that the car flies over the end of the track. There are two interpretations here: the games considers the finish tile, while there is an argument for a more complex finish line definition, something like "moving from the half of the finish tile that comes before the gate, in the direction of the track, to the half after the gate".

For ease of reasoning, with my current understanding of how the games assigns penalty time, I would stick to the game's interpretation: if you reach the finish tile, you are done; if you do so from an odd direction, you might receive a penalty time, which must be accounted for.

In my mind then, a replay where the car flies from the loop straight to the finish tile would be valid. The existing rule quoted above is sufficient to discriminate.

For the current race, we can continue as is, removing all invalid replays that end with the car touching a tile that is not the finish line. If people are still uncertain, we can move the first split slightly, and restart the race.

[Argammon]

Thanks @dreadnaut. The clarification is good enough for me. 

[Spoonboy]

Finishing via the finish line tile makes sense to me as a defined rule.
And if you have the skills to finish your lap by hitting that tile from the air, at speed, without touching the tarmac - you deserve to keep it!! 

[Duplode]

That does clear it up, thanks @dreadnaut ! While I kinda like the crossing-the-line interpretation, it's indeed questionable whether the added complexity is worth the trouble.

[Frieshansen]

Thank you @dreadnaut for clearing that up! Also thanks to @alanrotoi and @Erik Barros - I'm glad you like it.

[dreadnaut]

As mentioned in the shoutbox, this replay by @alanrotoi ends on the split, not the finish line. Removed from the scoreboard, but archived here.

Alan Rotoi    Argentina    0:44.89    Lambo       1:01.50    Oct 28   17:25

[Argammon]

It is totally possible I am overthinking the split issue, but I can still see two potential problems arising due to this track design flaw.

1) You cannot view this attachment.

When you fly a trajectory similar to the one indicated in the picture above it is difficult to tell whether the finishing tile or the split triggers the end of the lap. A quick test is to change the trajectory so that the car clearly lands on the finishing tile. If that test triggers 6 seconds penalty time the lap is invalid. However, I am unsure if my test detects all instances of the problem. Is it possible for one of you coders to develop a simple tool that checks for how a replay ends?

2) Suppose pipsqueak A is leading during the private days and pipsqueak B takes the lead with an invalid replay. Pipsqueak C drives a lap that would be sufficient to beat player A but not to beat player B. Since pipsqueaks A and C are close competitors pipsqueak C decides to wait and post his lap during the quiet days. After the end of the race, pipsqueak B's replay is found invalid and is deleted accordingly. The problem is that player A receives player B's leading time (possibly several days) even though player C should have received it. I find that problematic since yesterday's incident and my example above, see (1), show that even experienced players may post an invalid lap by accident.

[Duplode]

@Argammon My two cents on the trajectories:

When you fly a trajectory similar to the one indicated in the picture above it is difficult to tell whether the finishing tile or the split triggers the end of the lap. A quick test is to change the trajectory so that the car clearly lands on the finishing tile. If that test triggers 6 seconds penalty time the lap is invalid. However, I am unsure if my test detects all instances of the problem.

Your test should be good enough in most situations. A typical scenario here resembles the contrast between the orange (legal, with penalty) and red (illegal, skipping penalty) lines in the picture below:



(Alan's deleted lap, by the way, is unambiguously a red line case.)

A potentially problematic situation, though, is the blue line one: a trajectory with enough swerve to catch the final corner (which is enough to avoid penalty) and then miss the finish tile. In such cases, the penalty check won't help, and it isn't easy to tell where the car is at the end of the lap with only the naked eye:



However, there is an in-game camera trick that makes verification easier. Stop the tape at the end-of-lap frame, switch to the F3 camera, place it fully overhead (up direction all the way), zoom out and then rotate the camera (left/right direction). If at some point while you rotate it the split element disappears from view, so that you only see the start/finish element, the car is crossing into the finish tile, and the lap is legal:



This is a slightly conservative test, as if you are very close to the boundary of the tiles (even closer than in the picture above) neither the split element nor the start/finish one will disappear when the camera is rotated, and so the test won't be conclusive. That probably won't be much of a problem, though, as if you're trying something like the blue line getting too close to the limit will likely cost you time (the more you bend the line, the longer the trajectory is).

Edit: Attaching the replay used for the screenshots above, in case you want to reproduce the test. (It is Alan's deleted lap with a different ending.)

[Argammon]

Thanks @Duplode. Your test is much more practical than mine.

For replays that do not use the detour (the right way) a tool that does the following would also suffice:

Before we start, we need a second version (SV) of the track with one of the pieces from the right way missing making it incomplete. Then the tool extracts the replay from the real replay file and copies it into a dummy replay file containing the SV tack.

Loading the dummy replay in stunts would reveal whether the replay is ok or not. Do you have too much time on your hands @Cas? 

[Cas]

Ha, ha... Actually, not 100% sure, but I think with Bliss you can write a track onto a replay. You could use this to make that test!

[dreadnaut]

Only a day left! Sneak in your quiet replays

Attached, @Mark Nailwood's early replays with an invalid ending, which I removed from the scoreboard.