Topics

Here go a few thoughts about handling different versions of the same car. I'll begin with a @dreadnaut quote from another thread:

You might note a version number there. The discussion above made me realise that we need to keep that explicit, because an old competition might want to point to a specific version of the car, and not to the most recent one. Versions can "branch", so it's not simple ordering. I think I'm roughly proposing WXYZ/<version-id> as a versioned car identifier.

This is a good capsule summary of what we'll have to deal with. Compared to the initial draft of the design in the GitHub repository, it also points -- rightly so, IMO -- toward more explicit version management. I'll also quote the relevant bits of the repo docs, just for the sake of providing contrast:

For each car:

• a <WXYZ>.yaml file containing
• [...]
• a resources directory containing all the car files
• an optional history directory, containing subdirectories for previous versions of the car

Such a layout would likely suffice for the happy path, in which cars always have a linear version history, the most recent release by the author is the default choice of version by the community, and there are no replay-incompatible releases once a car has been used in competition. History, though, has a way of playing tricks on us when we least expect.

One particular scenario I see as important to consider is that of a car having an initial release which is used in competitions and then, for whatever reason, gets a replay-incompatible update at a later point. It is worth noting this is not a theoretical concern: a few of Ryoma's cars which were picked up by competitions while he was away in 2022 later had performance-changing updates, most notably the Ferrari 641/2 featured in R4K last year.

With these considerations in mind, here's what, at the moment, I'd feel like adding to the repository layout for the sake of version management. This is a rough sketch, for the consideration and criticism of y'all:

• There should be an INI file for individual versions. There is no need to duplicate everything from the main INI file of the car, as the point is just providing clarity on what the version is. Plausible fields include:
  
  • car-id, for cross-referencing;
  • version-id, along the lines of what dreadnaut has suggested in the quote at the beginning;
  • release-date;
  • distributor (optional), who originally prepared the release. Relevant, for instance, to distinguish KyLiE's bugfix releases;
  • comment (optional), a one-line summary of what distinguishes the version.
  Additional information we might want to keep track of includes replay compatibility across versions, and known races in which a version was used. These are trickier, though, and can wait until we get the basics settled.
  
• All versions, including the default one, should get their own subdirectory and version INI file, ensuring uniformity of handling and documentation. (This differs from the initial GitHub draft, in which a privileged default version would have files in resources, while old versions would get subdirectories of history instead.)
• The main/default/current version should be recorded in a field of the car's INI file, so that tools and/or users can find the way to it. For the sake of convenience, it's likely we'll still want to provide the files of the main version outside of its subdirectory (e.g. as a zip in the car's root directory). I think it will be fine to replicate the files for that specific purpose.
  
  • Given the possibility of the version history splitting into branches, it might make sense to distinguish, with separate fields in the car's INI file, between the latest version by the author and the current community default (at least one of the fields being optional). The "community default" concept is perhaps similar to the notion of a "homologated" version (i.e. one that has been used in competition) that I tried to bring into the Southern Cross collection some time ago.

[diffcar]

I'm happy to announce the release of diffcar, a command line tool for comparing and detecting changes in CAR*.RES files. diffcar currently offers two commands. The first one is diffcar single, which takes a pair of CAR*.RES files and prints a nicely formatted, Stunts-aware diff of them:



The second one is diffcar report which takes two directories and prints diffs of files with matching names (case insensitively) that exist in both of them. The output is printed to the terminal by default; you can specify a destination file instead with the -o option.

You can download Windows and Linux binaries from GitHub: https://github.com/duplode/diffcar/releases/tag/v0.1.0.0 . As ever, feedback is most welcome!

[Light]

In the parent thread, one issue that has come up is how to name and record paint jobs. Here are a few notes on the topic, which might be of some use in figuring out how much will we want to standardise the descriptions.

The original cars span eleven paint jobs, with almost all of them come in light and dark pairs. The table below shows them, along with some plausible single-word fancy names that may or may not be useful to adopt. (Suggestions of nicer or more accurate, fancy names are welcome!)

		Light		Dark
Grey		Silver		Graphite
Blue		Blue		Turquoise/Aqua
Purple		Violet		Indigo
Red		Red/Scarlet		Ruby/Carmine
Yellow		Yellow		Gold
Green		Lime

The omission of a darker green was readily rectified by the community (see eg. the "British racing green" Lotus paint job #1), to the point it has pretty much become the 12th car colour. Other examples of custom solid colour paint jobs are:

• Black (eg. Speedgate #1)
• White (eg. F40 #4)
• Pink (eg. Thunderbird #1)
• Beige (eg. 911 Turbo #7)
• Orange (eg. Skyline #7 -- though note the palette there can be improved a bit)
• Brown (eg. Torino #1)

Beyond these, we have a plethora of combination paint jobs, which can be described in all sorts of ways: with slashes (eg. BMW DTM #9, Black/Pink/Red), plain descriptions (eg. Melange #3, Yellow with Black Stripes) or fantasy names (eg. Monster Truck #5, The Joker).

Welcome to the Car Archive project sub-board! This space is meant for specific discussions about the project that might benefit from having their own thread. The original "Car archive" thread remains open for general discussion and free-form brainstorming.

Useful link(s):

• dreadnaut's project repository on GitHub
• "Custom cars" Wiki article (includes an extensive list of cars compiled by Alan Rotoi)

I have worked out a recipe for making scoreboard side pictures of cars with consistent looks and angles. While not as sleek or convenient as a special purpose tool would be, it is quick enough to do that even a large handful of pictures can be prepared painlessly. The secret sauce is tweaking Stressed so that the default viewing angle is the one we want to use (more on that at the end). I've gone ahead and updated the pictures on the site, so let me know how you like it after a hard refresh -- while consistency requires a few compromises, there might be some room for adjusting the values in the recipe.

So, without further ado, here are the steps:

• Open the car P3S/3SH in (tweaked) Stressed with a maximised window, select car0 and the chosen paintjob.
• Take a screenshot of the car (leave some white background around it; no need to crop it precisely) and open it GIMP (or Photoshop, etc.).
• Use selection by color with zero threshold/fuzziness to pick all the white areas, then use color to alpha to make them transparent (this works great because Stressed renders without blurring and uses #fcfcfc rather than #ffffff for in-game white).
• Crop the image to the content.
• Look at the cropped image dimensions and calculate the height : width ratio. (We're aiming for a maximum car size of 360 x 92, so this ratio will guide how we'll scale the image.)
• If the height : width ratio is smaller than 92/360 = 0.2555..., then proportionally scale the image to a width of 360. Otherwise, proportionally scale it to a height of 92.
• Enlarge (without scaling) the image/canvas size to 380 width, centering the content horizontally, then enlarge it to 100 height, placing the content 4 pixels above the bottom. (The current size of the images on the site is 380 x 100.)
• Export the resulting image as a GIF.

Below are the the tweaks in the Stressed code. I picked 10° as the roll angle because that felt like a typical value for the old pics; 15° is an alternative value that might work as well. You'll note I have also changed the y-axis aspect ratio correction factor in the rendering from 0.8 to a more accurate 0.8333... (that's something I probably should make into an actual patch for Stressed). Note that the line numbers in this diff don't necessarily match the current state of the Stressed repository:

Code Select Expand

diff --git a/stunts/stressed/src/shape/shapeview.cpp b/stunts/stressed/src/shape/shapeview.cpp
index a92c31f..b4ecfb6 100644
--- a/stunts/stressed/src/shape/shapeview.cpp
+++ b/stunts/stressed/src/shape/shapeview.cpp
@@ -135,7 +137,8 @@ void ShapeView::reset()
         0.0f,
         0.0f,
         -distance(VerticesModel::toInternal(bound[2]), VerticesModel::toInternal(bound[5])));
-    m_rotation.rotate(10.0f, 1.0f, 0.0f, 0.0f);
+    m_rotation.rotate(-90.0f, 0.0f, 1.0f, 0.0f);
+    m_rotation.rotate(-10.0f, 0.0f, 0.0f, 1.0f);
   }
 
   QAbstractItemView::reset();
diff --git a/stunts/stressed/src/shape/verticesmodel.cpp b/stunts/stressed/src/shape/verticesmodel.cpp
index f9b7db0..0877e58 100644
--- a/stunts/stressed/src/shape/verticesmodel.cpp
+++ b/stunts/stressed/src/shape/verticesmodel.cpp
@@ -21,7 +21,7 @@
 const int VerticesModel::VAL_MIN;
 const int VerticesModel::VAL_MAX;
 
-const float VerticesModel::Y_RATIO = 0.8f;
+const float VerticesModel::Y_RATIO = 5.0f / 6.0f;
 
 bool VerticesModel::m_weld = false;


This is a thread for sharing test reports of Stunts cars: all cars are fair game, and everyone is welcome to contribute. It is a spiritual successor to the "Trying out Ryoma's cars" thread from the 2022 preseason, and accordingly tests from there will be linked from here as well.

Index of tests:

• Alfa Romeo 164 Procar *
• Citroën XM V6 24 *
• Cizeta V16T *†
• Dodge Challenger R/T Hardtop (1970) *
• Ferrari 312T and Ferrari 312T4 *
• Ferrari 365 GTB/4 Daytona *
• Ferrari 412 Automatic *
• Ferrari 456 GT *
• Ferrari 512TR *
• Ferrari Pinin *
• Jaguar XJ220 *†
• Lamborghini Diablo VT *
• Lancia 037 *†
• Lancia LC1 and Lancia LC2 *†
• Lancia Stratos *
• Maserati Shamal *
• McLaren F1 *†
• Peugeot Oxia
• Pontiac GTO (1967)
• Porsche 959
• Vector WX3 and Vector W8 *†

*: Test from the "Trying out Ryoma's cars" thread.
†: Retest pending due to major changes in a newer version of the car.

Here's a report on a little experiment ran a while ago, with support from @Daniel3D and @alanrotoi . The first two CCC races of this season, Magic Lamp and Free The Genie were set up so that we could gauge how much difference powergear makes to the performance of powergear cars. The extent of powergear availability is a major confounding factor in the sort of bulk comparison I attempted during the ZakStunts preseason rules discussion (which is otherwise fairly accurate). In this experiment, a complementary approach was taken: Magic Lamp is a (mostly) PG-free track, while Free The Genie is a PG speedway, so the differences in relative performance between the tracks can give us a view of influence of PG.

To begin with, these are the results for Magic Lamp. Those were NoRH laps (except for redoing the finish with the flexible PG cars to avoid getting PG on the final loop) driven by me with roughly the same effort on each of them (laps and tracks are attached). For the flexible PG cars, powergear was intentionally avoided on the final loop. The final column is the bonus percentage, ZakStunts-style, corresponding to the laptimes. P962 (fixed at 0%) and LM002 laps have also been included for reference:

Code Select Expand

PMIN    1:35.85    -5.7
P962    1:41.30    0   
XCFX    1:42.90    1.6   
PFCR    1:47.30    5.6
FGTO    2:09.40    21.7
VETT    2:10.60    22.4
ANSX    2:11.35    22.9
LM02    2:19.95    27.6
PACK    2:33.55    34.0

(XFCX is the Xylocaine, PFCR is the Crown Victoria, PACK is the Packard 8.)

This comparison puts non-PG Indy at -6%, the other original PG cars fairly close to each other around 21-23%, the Xylocaine at a pretty low 2%, the Crown Victoria (which isn't quite a PG car) at 6%, and the Packard at 34%.

Next, the Free The Genie results. I drove this with RH so that a comparison of full PG laps would be realistic, again putting similar amounts of effort in each lap (though the Xylocaine one might be a touch stronger than the others):

Code Select Expand

PMIN    1:07.55    -28.6   
XCFX    1:12.45    -19.9   
PACK    1:19.45    -9.3   
FGTO    1:20.10    -8.4   
VETT    1:20.85    -7.4   
P962    1:26.85    0   
ANSX    1:29.20    2.6   
PFCR    1:48.85    20.2   
LM02    2:28.60    41.6   

Both the Crown Victoria and the LM002 lose heavily here due to not having been fast enough to cut the second u/d cork (even the P962 barely made it). As for the PG cars, we see gains of over 40% for the Packard, around 30% for the Vette and GTO, and around 20% for the Acura (as expected, gains are limited by the lack of powerslides), Indy and Xylocaine (which are a little less dependent of powergear thanks to better handling).

These are results obtained by one specific driver on two specific tracks, so they are certainly affected by idiosyncrasies and chance. Still, they should give a rough idea of what to expect when it comes to the influence of PG on e.g. bonus percentages.

Kevin Pickell's Westwood is a great retro pick! Between here and CCC, I also like seeing the Xylocaine get some regular use -- I feel we've kinda done it dirty back when it made it to ZakStunts years ago.

A bug report from @Erik Barros a while ago made me realise the language packs for the forum interface weren't working well, as they had been made for a version of the forum software much older than what we currently have. I have just reinstalled the language packs with up-to-date versions, which should provide a better experience if you ever feel like switching the preferred language in your profile. Below is a before-and-after comparison:



 

[The proposal]

2023-04-22 update: The proposal below is being updated to better reflect our current understanding of unstable replays. To avoid invalidating the next several replies, removed passages will be struck out, while additions will be in italics. For a rundown on when instability arises and how to work around it using the replay controls, see this post.


I'd like to start a discussion about the unstable replays rule, motivated by the replay invalidations at ZCT260 (see the shoutbox posts from April 8th and 9th for what happened there, and the end note at the bottom of this post for an explanation of what unstable replays are). I will open it by putting a rule change proposal on the table for your (and dreadnaut in particular, of course) consideration.

The proposal

Change the relevant passage of the rule which makes unstable replays invalid (rules page, section "The System", bullet point #2) from:

"[...] and [a replay] should play correctly when loaded in Stunts, from 'Options → Load replay'."

To:

"[...] and [a replay] should show as complete upon loading in Stunts from 'Options → Load replay', rewinding to just before the finish line, and continuing from there."

2023-04-22 update: While checking how the replay ends upon loading from Options should be the primary means of validation, the current procedure of watching the lap from the beginning, with multiple cameras if necessary, should be kept as a fallback. That being so, it might be appropriate to concisely describe the validation procedure and how it is supposed to work in a separate item.

Rationale

My proposal tries to strike a compromise, keeping replay viewing and validation reasonably easy while excluding as few valid-as-driven replays as possible. To expand a bit on that, I'll consider the two goals which, as far as I understand it, motivated the 2021 rule change:

• Ensuring competition management can validate replays by following a simple procedure with reproducible steps.
• Avoiding confusing replay watchers who play laps from the competition archives.

Goal #1 is fully addressed by my proposal. In fact, as far as manager convenience goes it even improves upon the status quo, as it is no longer needed to watch the whole lap to be sure it is valid -- just loading from Options, rewinding a little bit, and continuing to the evaluation screen is enough. (This streamlined procedure is what I have always assumed managers short on time would default to on race closing, which partly explains my mix-up when replying to Friker at the shoutbox on April 2nd.) Besides that, unstable replays which are valid according to this procedure (i.e. the vast majority of them) can be watched to the end on normal speed through the well-known replay controls trick of rewinding a little around shortly after the point of divergence and resuming play, and are reproduced correctly by the repldump tool, and usually, with an adequate choice of starting frame, when played at double speed as well. (And conversely, Overdrijf's extraordinarily hard to verify replay from ZCT232, which triggered the rule change, fails the streamlined verification.)

2023-04-22 update: Considering what we have learned over the past two weeks about unstable replays -- in particular, that divergent timelines are system dependent, and that replays such as Overdijf's ZCT232 one aren't as rare as previously thought -- it makes sense, as suggested by Friker, to keep the current, play-from-the-beginning validation as a fallback method. It is fairer to drivers to accept as many unstable replays as we reasonably can, and keeping the current method as a fallback won't noticeably increase admin workload. The main caveat is that, since divergent timelines are system dependent, if a replay is only finished on a divergent timeline and not on the load-from-Options/fast-forward one there is no guarantee that the manager will be able to watch the lap being completed. That being so, it seems appropriate to keep the new wording of the rule as it was in the original proposal, recommending pipsqueaks to check if their laps show up as complete upon loading from Options -- if they don't, there is a nonzero chance of them being ruled invalid, even though the fallback validation might rescue them.

As for goal #2, there are two aspects to consider. Firstly, there being a reproducible way for managers to validate replays means every one else can check them in some way, and as pointed out above that also translates to reasonable ways of actually watching the laps. Secondly, I feel we should not penailse, in multiple ways, pipsqueaks in the here and now for the hypothetical benefit of a future visitor who might be confused by archival replays -- specially considering that the pipsqueaks have valid-as-driven laps and are at no fault, and that we can provide easy access to information that clarifies the matter to visitors (through Wiki articles, replay package readmes, and so on).

If need be, I can go into further detail on the points above, and in particular on what makes it so taxing to verify replays in the way specified by the current rule. But I have talked for long enough for an opening post, so it's time to give the floor to you. 


End note: what is an unstable replay?

An unstable replay is a lap, usually a RH powergear one, which plays differently in-game depending on how the replay controls are used. In the most common scenario, the player completes the lap normally without noticing anything strange, and the replay shows up as complete upon loading from the Options menu and looking at the end of the tape. However, rewinding to the beginning and playing at normal speed shows the car as crashing or veering off track.

[Setting it up]

If you looked at the ZakStunts shoutbox on the ZCT258 deadline day, you may have noticed that @Erik Barros , @Overdrijf and me have been doing a few test with playing Stunts on alternative controllers. I'll open this thread with a report on using a Xbox 360 controller.

Setting it up

First, here is how I (with some help from Erik) have set up the controller on DOSBox. I'm currently using DOSBox Staging 0.80.1 on Linux. The operating system shouldn't matter much; however, DOSBox Staging has improved support for controllers, including the ability to have more than two analog axes recognised as such (see the 0.75.1 release notes).

With the default configuration, DOSBox Staging will map the x and y analog axes (steering and accelerator/brakes) to the left analog stick, and joystick buttons 1 and 2 (shift up and down) to Xbox buttons A and B. While that is reasonably faithful to what you'd get with the kind of joystick originally supported by Stunts, it is also suboptimal in that it is not very comfortable to have steering, accelerator and brakes on a single analog stick, specially given the wealth of options offered by the Xbox controller:



Xbox 360 controller diagram. Source: https://github.com/dosbox-staging/dosbox-staging/wiki/Keymapper

Changes to the controller layout are done through the DOSBox mapper, available by pressing Ctrl + F1 in DOSBox. DOSBox Staging, allows you to remap the controller however you like, removing some limitations found with plain DOSBox (please reply if you run into difficulties with that!). My favourite mapping so far, which should work with no further configuration in both plain DOSBox and Staging, is a hybrid one: accelerator and brakes on buttons (as analog makes no difference for them), and steering on the left stick:

• Joystick axis x-/+ (steering): left stick x-/+
• Joystick axis y- (accelerator): A button
• Joystick axis y+ (brakes): X button
• Joystick button 1 (shift up): RB bumper
• Joystick button 2 (shift down): LB bumper
• Esc key (menu): Start button, or Y button

With the mapping done, select joystick input in the Stunts option menu, and perform the in-game calibration as usual.

Compared to the standard Stunts input devices, this hybrid setup is rather like an improved version of mouse controls (steering on x axis, accelerator and brakes on the buttons), free of the jerkiness that makes using the mouse so unwieldy.

Test drive

A natural thing to ask about using analog controllers in Stunts is whether the input is actually analog. After all, replay files store input in a digital format, a direct transcription of keyboard key presses. In the case of accelerator and brake input, there is indeed nothing analog about the input, even if you map an analog stick or trigger to the joystick y axis. As for steering, however, Stunts has a trick up its sleeve. In joystick mode, steering without moving the analog stick all the way left or right will keep the car wheel in an intermediate position as well. The game achieves that by sending key presses in the appropriate rhythm to keep the wheel roughly at the same place. That is demonstrated in the attached JOYTEST5.RPL, in which I spent more than a minute moving in circles, with the LM002 wheel held at one fifth of the way left.

What about actual driving, though? Though effective driving with non-keyboard controls will surely take a lot of getting used to, the first impression was far better than expected. After some warming up and acclimatising, for instance, I got the attached XBOX4.RPL: Default, Indy, classic line, 1:07.40. Slower than on keyboard, but a very normal lap, all things considered -- and the potential to improve is certainly there.

One difference between joystick and keyboard steering in Stunts is that with the joystick mode the movement range of the car wheel is only about half of what we get with a keyboard, presumably so that the analog steering wouldn't become too sensitive. That will probably mean a disadvantage in full powergear tracks and other scenarios which require extreme sliding tricks. On the other hand, I expect that it will be possible to put analog steering to good use in different situations, specially in OWOOT + NoRH driving.

(A practical detail worth mentioning: there seems to be a minor Stunts bug such that accelerating with joystick controls doesn't interrupt the leaving-the-truck animation. One way of skipping it without losing any time is holding the accelerator and then pressing shift down.)

Space and Enter

On a final note, these experiments made me think of a (perhaps obvious) hypothesis for why the keyboard controls have two pairs of keys for shifting (A/Z and Space/Enter). It sounds plausible that A/Z were originally meant as the "proper" shifting keys. However, as the two joystick buttons must handle both shifting and menu confirmation, it makes sense for a joystick button press to be translated to both. Given a reversible mapping, we'd end up with Enter (and Space) playing both roles on keyboard as well.

2025-05-08 update: monthly updates spun off from the general thread into this separate topic.

The ratings have been updated for ZCT258! This is the first public update of the rankings. This round, we have seen two pipsqueaks returning to the ranking, Friker and Ryoma, and Argammon further improving his personal best from ZCT257. No new entries in the rankings quite yet, though Erik needs just one more counting race entry for that.

I'm happy to announce the opening of the Arrabona project: an archive for previously inaccessible parts of the UnskilledStunts records. Right now that includes track-and-replay packs for all 137 races of UnskilledStunts competitions (USC, USL and Christmas Cup) so far, as well as the scoreboards for UnskillesStunts 2021. Thanks to CTG for green-lighting this project!

Now that I'm finally doing the sorely needed behind-the-scenes cleanup of the Southern Cross site, it is probably a good thing to decide what should happen to the car collection there -- and your input will be very helpful for doing so.

For several years, I had tried to maintain Southern Cross as the primary place to download custom cars. However, except for adding the 19th Anniversary Melange, I haven't updated the collection there since May 2021. To make matters worse, many of the Ryoma cars there aren't in their final versions, and I'm not even 100% sure if the downloads of them are replay-compatible with the ones used at CCC.

As things stand, the best maintained car collection around is clearly @KyLiE 's (see the Stunts Custom Cars page at https://www.markceccato.com ), and in fact I have recently added a note to the Southern Cross page recommending it. That, however, probably means I have to find some other role for the Southern Cross collection, as I'm not sure if there's much point in me (re)creating a second collection of individual car downloads with basically the same scope. Some of the alternatives include:

• Shutting down the main part of the Southern Cross collection, leaving only the old-school cheat cars and the historical curios.


• Switching the focus of the collection to car packs. (That would be a big U-turn, as the point of the Southern Cross collection originally was to offer individual downloads, so that people could easily pick just what they needed. However, the existence of Simple Garage increases a lot the usability of car packs, making this a more attractive option than it was ten years ago.)


• Figuring out a way to collaborate on a shared collection, set up so that it can be easily made available on different sites. (This is wild speculation, as right now I have no clue about what would be a good way to achieve that.)


• Restoring the Southern Cross collection in something like its current format. (You'd probably have to persuade me there's a point in doing so, though.)

So, how should the Southern Cross collection evolve? Is there any other possible role for it that I have overlooked? Please let me know of your thoughts, as I don't think I'll manage to find a good solution without considering your views.

[ZakStunts Folyami ratings]

I'm delighted to finally release the ZakStunts Folyami ratings, an Elo-like rating system for ZakStunts! You can check the ratings right now at the Southern Cross site. Below is a quick Q&A about the ratings -- if you have extra questions, feel free to ask them!

Why are there two rankings?

Given that Folyami ratings are pretty dynamic, responding rather quickly to changes in form/current performance, it felt appropriate to have something the more permanent to go along with the ranking of current ratings. That being so, there is also a historical ranking, which lists the highest ever rating reached by pipsqueaks.

Why am I not showing up in the rankings?

There are basically two possibilities:

• Firstly, at least five completed races are necessary to be included in the rankings, so that ratings have at least a few rounds to stabilise.


• Secondly, pipsqueaks leave the current current ranking after four races of inactivity, and rejoin upon returning to the competition. (Note that there are a few rules for discarding race results that aim at removing unrepresentative ones, such as those reached with an obviously uncompetitive car. That being so, race entries might, in special circumstances, not be counted for the ratings.)

So don't worry: no one gets excluded from the rankings, and you just have to keep racing for (re)joining it 

How do the ratings work?

Here are links to a summary of the ZakStunts-specific aspects of the rankings and a technical overview of the rating system. (I have attached the latter here as a PDF, in case you find that easier to read.)

Can other competitions be included in the ratings?

The Folyami project began as an offshoot of my earlier investigations about race strength, which is why the ratings came into being as ZakStunts-only. Ideally, it would be nice to follow illustrious predecessors such as WRL and SWR and make Folyami an omni-rating covering all Stunts competitions. While I do want to explore ways of achieving that at some point, it can't help but be a project for the long term. Not only there are decades of competition results to be reviewed and formatted, but also harmonising the competitions into a single system could prove challenging, especially given how much the currently active competitions differ from each other.

Will there be an update of the race strength ranking?

Sure! I will add race strengths to the site as soon as I figure out a few details about how to best present the data. By the way, if you have any suggestions of additional features and visualisations for the ratings and the historical data thereof, please do let me know!

[Three rounds early]

When in the season are titles decided? That is an interesting question -- specially during the season itself! -- which isn't always easy to answer just by looking at the season scoreboard. Discards mean that pipsqueaks can have different amounts of points available from the final races of the season, so just looking at the Real score and the number of remaining races isn't quite enough to make projections. The experiments with point systems from a few weeks ago gave me a good opportunity to look at title decisions in past years.

Here goes a list of ZakStunts seasons, ranked by how many races were remaining when the title was mathematically decided. When it comes to tiebreakers, it makes sense to use the number of discards and the total number of races, in that order.

Three rounds early

• 2008 (2 discards from 11 rounds)
• 2010, 2016, 2017, 2019 and 2022 (3 discards from 12 rounds)

Two rounds early

• 2004 (2 discards from 13 rounds)
• 2009, 2013, 2018 and 2021 (3 discards from 12 rounds)

One round early

• 2007 (2 discards from 9 rounds)
• 2005 and 2006 (2 discards from 12 rounds)
• 2003 (2 discards from 13 rounds)
• 2011 (3 discards from 11 rounds)
• 2012 (3 discards from 12 rounds)

Decided on the final round

• 2001 (0 discards from 9 rounds)
• 2002 (0 discards from 12 rounds)
• 2014, 2015 and 2020 (3 discards from 12 rounds)

And here are the seasons sorted chronologically, with a few extra details. The Available columns show how many points were available to champion and opponent before the deciding round, and Gap is the difference between them. Note that, thanks to discards, the opponent in the strongest position isn't always who was in second place at the deciding race -- for this table, I picked whoever had the largest difference between their available points and their gap to the leader.



A few notes on specific seasons:

• Unsurprisingly, the most impressive early decision is 2008. After ZCT87, Ayrton would have won the title even with a non-discarded missed race.
  
  
• While five seasons were mathematically decided in the last race, due to various circumstances only 2015 and 2020 really had a final round battle.


• 2009 is a bit of a special case. While CTG wouldn't have been able to overtake me in the final two races, in the most favourable scenario for him we would end with the same score. Since there are no tiebreakers in the season scoreboard, I counted the season as being a two-rounds-early one.


• You might be wondering why have I listed 2001 as being decided in the final round if the points available to Ben Snel were smaller than the gap to Roy Wiegerinck. The answer is that in 2001 (and 2002) average points over races joined were a major component of the season scores. That being so, a pipsqueak might lose points by entering a race and getting a sufficiently bad result in it. While the title wasn't mathematically decided before ZCT9, Roy could guarantee the title by skipping the race, which is duly what happened.

Who created the Coconut cars? That has been a nagging question that I have been reminded of every few years. Stumbling upon the readme I once wrote for the Coronet Pulsar ("created by an unknown designer in a bygone era") made me realise I had never made a serious attempt at finding out the answer. Back in my newbie days, when I got to race with the Coronet at SDR, or even when I started the Coronet Wiki article a while later, it felt like those old school cheat cars had always been there. The distance of a handful of years now tells me I should have known better. Time for some digging, then!

The oldest source on the live Internet for the Coconut cars is probably the cheat cars package at ZakStunts' downloads page, which was added to the site somewhere around August 2001:



The package includes nine cars, helpfully displayed on this table (thanks Alan for adding it to the Wiki!):



We might divide those cars in three groups:

• The five cars attributed to Mr Eriksson, whose files in the zip have modification dates from March and April 2000;
• The two Coconut cars: Coconut Car Coronet Pulsar STi-R, and Coconut Car GTR 7.5 GT IMSA. Their files date from August 2000; and
• The two remaining cars. The Ditsch car in-game description includes both authorship and a dedicatory ("for Magnus by brainSteen"). Though the Wooden wireFire has neither, the file dates (May 2000 in both cases) and the use of camelCase ("brainSteen", "wireFire") suggest they belong together.

The cars themselves provide no further clues on who Mr Eriksson, Magnus or brainSteen are, or whether they have anything to do with the Coconut cars. To remedy that, we can take a ride on the Wayback Machine and turn to the who's who of people in the community around the turn of the century: Kalpen's directory of Stunts links! It doesn't take long to discover that both brainSteen and Magnus not only had Stunts sites, but also offered a common set custom cars for download:



Unfortunately, brainSteen's site wasn't archived, and neither was the car downloads page on Magnus' site. Wandering along the other sites linked from Kalpen's, we can find a surprising number of custom car downloads; however, in all cases I have looked at there is enough information on the pages, even when the car files themselves weren't archived, to figure out the Coconut cars aren't among them. That includes the Supercar .zip from the TSST (The Swedish Stunts Team) site, which likely packaged some of the cars by Royforever:



That's not the only relevant information in the TSST site, though:



So Magnus joined TSST in March 2000. But wait! Have a look at the site footer:



Now compare with the full name of the Coronet Pulsar 

The contact info on that footer points to Red Baron, team leader dictator of TSST:



And now, the finishing touch: the list of URLs from Magnus' site captured by the Wayback Machine includes several PDFs with paper models of cars and other vehicles. These models are credited to...



So, here is what I think the events were: Magnus, aka Mr Eriksson, is a Swedish pipsqueak and car designer. He joined TSST in March 2000, in between creating the Škoda Felicia and the Лада Нива. Several months later, someone in the team made a couple of СССР cars in tribute to their organisation.

Even after tying those threads, it's hard to identify a single author: it could have been Magnus, though the Coconut cars are very different in style to the Mr Eriksson ones; it could have been Red Baron, though the copyright footer on the site is clearly part of an in-joke rather than anything personal. In any case, it seems certain that the Coconut cars are a TSST project. I guess I finally have something meaningful to write in that readme! I can't help but wonder, though, if @zaqrack or any of you who were around at the time would confirm, or deny, my speculation

[Ferrari 456 GT]

In the Cars and rules for 2022 topic, there was a consensus for having at least one of Ryoma's cars in the next season. For that to work smoothly, however, we need to do some testing in order to make an informed choice. To get things started, I will (try to) test one car from Ryoma's Mega every day, and report the results here. If some of you join me in doing that, I'm sure we'll get to figure out suitable choices in no time.

Here are the cars tested so far:

• Alfa Romeo 164 Procar (report by Alan)
• Citroën XM V6 24
• Cizeta V16T
• Dodge Challenger R/T Hardtop (1970) (a CTG/Ryoma project)
• Ferrari 312T and Ferrari 312T4
• Ferrari 365 GTB/4 Daytona
• Ferrari 412 Automatic
• Ferrari 456 GT (at the end of this post)
• Ferrari 512TR
• Ferrari Pinin (report by Alan)
• Jaguar XJ220
• Lamborghini Diablo VT
• Lancia 037
• Lancia LC1 and Lancia LC2
• Lancia Stratos
• Maserati Shamal
• McLaren F1
• Vector WX3 and Vector W8

To begin with, here are some words on the Ferrari 456 GT (forum topic):

• Gears: 6
• Powergear: No
• Flat track top speed: 196 mph
• Real top speed: 218 mph
• 0-60 mph: 5.1 s
• Time to hill at Default (auto gears): 12.55 s (References: Countach - 12.40 s; Skyline - 12.45 s; Acura - 12.70 s)
• Default test lap (NoRH, classic line): 1:32.45
• First impressions: A moderately fast sports car, whose closest match is probably the Skyline (the 456 should have the upper hand on faster tracks, though). Next to other cars from that class, its handling feels nice and responsive, though there doesn't seem to be much extra grip, and it is easy to get overconfident on the corners.

Extracting the race positions from all ZakStunts scoreboards is not a straightforward matter, even with database access. The positions are computed from the replays rather than stored, and there are all sorts of corner cases to look out for -- name changes, draws, disqualifications, and so forth. For my ratings and race strengths project, I opted to review all the scoreboards and prepare the dataset by hand. A CSV file with the data is available at my project's GitHub repository (attached below is a version of the same file as of ZCT243).

The Track, Racer and Rank columns of the CSV contain what it says on the tin. pipsqueak aliases and name changes have already been unified. The Ghost column value is 1 if the pipsqueak is a known ghost, and 0 otherwise. Lastly, the Status column records the following occurrences about a race entry:

• DSQ: Disqualification, from a 00's ZakStunts race (back then, disqualifications were handed by setting the laptime to 9:99.99 rather than deleting the race entry outright).
• INV: Invalid replay, discovered upon review.
• MSC: CTG's 2014 races.
• EX1: Laptime beyond the "300% and 2 SD" cutoff.
• EX2: Lap driven under an alternative ruleset, such as GAR.
• EX3: Lap driven with a clearly uncompetitive car.

Edit, 21:53 ZakStunts time: I have updated the attached file after re-adding a few 2014 season status notes I had accidentally deleted.

In my previous thread on pipsqueak performance modelling, I reported optimistically on modelling lap times over repeated attempts with a gamma distribution. Back then, all I had for empirical evidence was a set of lap times driven in a minimal four-corners track. To better understand the model and the problem space, it would make sense to gather data on more realistic conditions. Soon enough, the perfect opportunity came up with Alpine, R4K's October race: great car, great (and not too difficult) track, and a real OWOOT NoRH race going on.

From October 11th to the deadline day, I had one (occasionally two) NoRH session on Alpine every day, and recorded all of my valid lap times. I tried my best to keep a consistent driving style, going for the most effective reproducible racing line I knew of, and not giving up on laps unless I crashed or left the track. The following set of box plots show what my lap times on each session looked like (for a closer look at the data, you can look at the attached CSV file):



I would divide my racing in three main stages:

• On sessions 1 to 5, I was still learning both car and track. In particular, I got under 83 seconds after figuring out a good line through the second and third corners, and under 82 seconds by  understanding how the second fast chicane should be approached.
• From sessions 6 to 14, there was largely stability, with perhaps some very slow improvement, culminating with a 81.00 on session 14.
• On sessions 15 to 18, there was modest but marked improvement, which I attribute to either realising I could be a touch more aggressive with my lines upon rewatching the laps I had posted to R4K, or to being in better shape by no longer doing late night sessions. I drove my final R4K time of 80.55 early on session 15. After that, I only managed one more lap under 81 seconds: an 80.75 on session 18.

With the lap times at hand, the next step is attempting gamma fits on stretches in which I had consistent performance. A good place to start could be sessions 15-18, in which I was perhaps closer to my best. Here is a first attempt:

Code Select Expand

Fitting of the distribution ' gamma3 ' by maximum likelihood
Parameters :
        estimate Std. Error
shape  5.5682015 1.79934095
scale  0.3520632 0.07724456
thres 80.2265133 0.24903711
Loglikelihood:  -115.9133   AIC:  237.8267   BIC:  245.612
Correlation matrix:
           shape      scale      thres
shape  1.0000000 -0.9492797 -0.9041436
scale -0.9492797  1.0000000  0.7526068
thres -0.9041436  0.7526068  1.0000000

A recap on what the gamma parameters mean in our context:

• thres, short for threshold, is the predicted ideal time, the one the model assumes it is impossible to go below.
• shape indicates how hard it is to improve as you get closer to the ideal time. It is associated with track length and track/line difficulty.
• scale reflects how consistently the pipsqueak manages to drive. Smaller values make for a narrower gamma curve, squeezed closer to the ideal time.

While the diagnostic plots show the gamma fit is pretty good, the parameter values are all over the place, as the following bootstrap confidence intervals indicate:

Code Select Expand


Nonparametric bootstrap medians and 95% percentile CI
          Median      2.5%      97.5%
shape  5.2026795  1.965190 10.9580942
scale  0.3649339  0.222748  0.6145803
thres 80.2609610 79.615182 81.0369308

Ultimately, is is a bit much to try fitting those three parameters at once with the available amount of data; there is too much wiggle room. From the three parameters, the one we are in a better position to estimate by other means is thres. The sum of the best sectors over the fifteen laps I had saved for posting to R4K is 80.05; that being so, 80.00 is a reasonable, if conservative, ideal lap estimate. Fixing thres to 80.00 results in the following gamma fit over sessions 15-18:

Code Select Expand

Fitting of the distribution ' gamma3 ' by maximum likelihood
Parameters :
       estimate Std. Error
shape 7.0919246 0.98618134
scale 0.3083916 0.04444141
Fixed parameters:
      value
thres    80
Loglikelihood:  -116.1671   AIC:  236.3341   BIC:  241.5244
Correlation matrix:
           shape      scale
shape  1.0000000 -0.9650934
scale -0.9650934  1.0000000

The 95% confidence intervals look a fair bit tamer now:

Code Select Expand

Nonparametric bootstrap medians and 95% percentile CI
         Median      2.5%      97.5%
shape 7.2104854 5.4284800 10.1456150
scale 0.3034129 0.2069173  0.4084368

While thres is the easiest parameter to estimate through different means, the one it would be arguably more interesting to keep fixed is shape, as it is supposed to be the one most closely associated to the nature of the track. Given the more believable estimate of shape we have just obtained for sessions 15-18, it is worth trying to fits for other sets of sessions with shape fixed at 7.1. Here is a fit for sessions 6-9...

Code Select Expand

Fitting of the distribution ' gamma3 ' by maximum likelihood
Parameters :
        estimate Std. Error
scale  0.4312479  0.0320152
thres 80.1501134  0.1939291
Fixed parameters:
      value
shape   7.1
Loglikelihood:  -142.3025   AIC:  288.605   BIC:  293.6915
Correlation matrix:
           scale      thres
scale  1.0000000 -0.8532844
thres -0.8532844  1.0000000

Code Select Expand

Nonparametric bootstrap medians and 95% percentile CI
          Median       2.5%      97.5%
scale  0.4276147  0.3703891  0.4924684
thres 80.1722529 79.8464586 80.5086957

... and 10-14:

Code Select Expand

Fitting of the distribution ' gamma3 ' by maximum likelihood
Parameters :
        estimate Std. Error
scale  0.3832479 0.02554952
thres 80.0632570 0.15585001
Fixed parameters:
      value
shape   7.1
Loglikelihood:  -171.2663   AIC:  346.5326   BIC:  352.1242
Correlation matrix:
           scale      thres
scale  1.0000000 -0.8591264
thres -0.8591264  1.0000000

Code Select Expand

Nonparametric bootstrap medians and 95% percentile CI
          Median       2.5%      97.5%
scale  0.3800133  0.3260546  0.4442169
thres 80.0765083 79.7532830 80.3593560

Keeping the shape fixed, the fits point to a noticeable reduction of the scale parameter (suggesting more confident and consistent driving) across the sets of sessions. The difference in thres (which might be ascribed to refinements of the driving line) is too small to be clearly meaningful.

Parameter estimation for my proposed model is challenging: pining down those three significantly correlated values without huge amounts of lap time data is nontrivial, and  if it can be tricky to extrapolate from one day to the next, let alone doing so across different pipsqueaks or driving lines... On the flip side, the results of the experiment do suggest the gamma distribution is a reasonable model for lap times, and that it is worth it to keep pursuing this idea.