Nürburgring - a new track for Revs (testers wanted!)

[MarkMoxon]

Hi.

As part of my Revs documentation project, I've been trying to port the Nürburgring track from Revs+ on the Commodore 64 to the BBC Micro. It's such a shame that this track never made it back onto the original Revs platform, and it's time to right that wrong!

It's taken quite a while, but I've finally managed to get the track working, so here's a brand new variant of Revs that includes the five original tracks, plus the Nürburgring. I've called this variant Revs+, and it's essentially the Superior Software variant (with Computer-Assisted Steering), plus the new track.

You can download the SSD from here:

https://revs.bbcelite.com/versions/revs-plus.ssd

or you can try it directly in JSBeeb here:

https://bbc.godbolt.org/?autoboot&disc= ... s-plus.ssd

Unfortunately my driving skills are rubbish, so I'm looking for help in testing the new track. I have managed to crawl around the Nürburgring in first and second gear, and it seems to work OK, but I really don't have the chops to test it properly.

If you can help, I'd be really grateful for any feedback. The extra tracks in Revs aren't just data files, they contain an awful lot of code, and converting the C64-specific parts to the BBC has been quite a challenge. I've had to fix problems with exploding tracks, sheer walls, disappearing corners and ridiculously fast opponents, but I think I've managed to tackle all the obvious issues.

The only new part in Revs+ is the Nürburgring track, so that's the only part that needs testing. I'm particularly interested in knowing what it's like to drive the track at speed, whether the other drivers have sensible AI and speeds for the various race classes, whether CAS works as expected, and any other aspects that you need driving skills to test. Also, if anyone can test this with a joystick, then it would be great to know how that goes, as all the extra tracks have joystick-specific code in them that kicks in for hairpins, but I just can't seem to get the hang of joystick driving.

Here's the map that came with Revs+, for reference:

If anyone has time to have a play with this and can let me know how they got on, that would be really useful. You can reply to this message, send me a private message, or contact me via Twitter at @markmoxon.

Good luck! (Especially with that hairpin...)

Mark

[chrisn]

A new track for Revs, great work! And it's a fun one to drive too.

I have found one bug ... I spun off the track on one of the corners, but instead of triggering the fence and then restarting, the game got stuck, with the wheel screech sound playing but the joystick and keyboard unresponsive, so I had to press Break.

[MarkMoxon]

I have found one bug ... I spun off the track on one of the corners, but instead of triggering the fence and then restarting, the game got stuck, with the wheel screech sound playing but the joystick and keyboard unresponsive, so I had to press Break.

Thanks for the report. I’m going to have to try crashing lots, to see if I can replicate it! Luckily, I should be able to manage that.

Just so I know where to concentrate my testing, was this on a real BBC, and if so, a model B? And were you using keyboard or joystick?

Thanks for the feedback, it’s really useful.

Mark

[chrisn]

Yes, a BBC B with joystick. If it helps, I think the corner where I came off was Ford Kurve. I'll keep playing and see if I can get round again without crashing and improve my lap time.

[MarkMoxon]

Yes, a BBC B with joystick. If it helps, I think the corner where I came off was Ford Kurve. I'll keep playing and see if I can get round again without crashing and improve my lap time.

Thank you. That’s handily close to the start, so I’ll see if I can replicate the issue with some intentional spin-offs. Should be fun!

Mark

[chrisn]

Looking at the time in the previous screenshot, it was a earlier in the lap, maybe the first or second corner after the long straight at the start. I haven't been able to reproduce the problem so far.

Here's a new best lap time ... 1:45.3

[MarkMoxon]

Looking at the time in the previous screenshot, it was a earlier in the lap, maybe the first or second corner after the long straight at the start. I haven't been able to reproduce the problem so far.

Yes, I suspect it's one of the the Castrol-S bends, given the time. I've flung myself off there a few times without a game crash, but I'm going to keep trying. At least it isn't happening all the time, so hopefully it's a rare issue. I'll keep looking though - thanks for the update.

Here's a new best lap time ... 1:45.3

That's pretty great. For reference, 1:45 is the cut-off time between Amateur and Novice races for the Nürburgring, in that if you qualify for an Amateur race with a lap time longer than 1:45, then the game automatically bumps the class back to Novice. So you're only 0.3 seconds off Amateur class, which makes me look particularly lame! The Professional cut-off is 1:41, incidentally.

I'm not sure how well-tuned the AI times are, as that's a different part of the code, but at least the cut-off times seem to tally, which is good.

Thanks for the feedback, it's proving really useful.

Mark

[jms2]

I have had a good go at it and so far have managed 1:41.8. I'm sure I can do better as well...

[jms2]

Getting better...

[MarkMoxon]

Getting better...

That’s a professional-class time! Very impressive.

Have you tried racing against the other drivers yet? I’m interested to know whether I’ve ported the AI driver speeds correctly, as all the C64 tracks seem to contain scaled speeds compared to the BBC version, so I’ve tried to scale the Nürburgring speeds accordingly. It’d be good to know if they make good competitors, or whether they are far too fast (or too slow).

(Edit: Ignore that bit, it’s the gear ratios and powers that are scaled, not the driver speeds, and given that the lap times are fitting nicely into the class ranges, I think they’re about right. It would still be interesting to know how the AI drivers manage on the new track, though.)

I’m glad to see it working, though. That’s a relief!

Mark

[jms2]

Not encountered any bugs so far. I haven't tried Competition mode but I'll give it a go now.

[jms2]

Had a go in Novice mode. I did terribly, but it did remind me of a cool feature that I don't remember reading about in your disassembly - if you crash heavily into another car, it gets knocked out of the race and remains on the track (often in an awkward location).

[MarkMoxon]

I didn’t know that happened! But I do know how it’s implemented:

https://revs.bbcelite.com/source/main/s ... track.html

I wondered what that routine was all about, and now I know. Thanks!

I haven’t written any deep dives into crashes and collisions yet, but they’re on the list along with the driver AI, so I’ll cover them then. And I’ll fix the typo in the first comment too.

Mark

[jimmy]

A new track for Revs, great work! And it's a fun one to drive too.

I have found one bug ... I spun off the track on one of the corners, but instead of triggering the fence and then restarting, the game got stuck, with the wheel screech sound playing but the joystick and keyboard unresponsive, so I had to press Break.

I think I have also experienced this bug. Originally I was using jsbeeb when this happened, but I couldn't find a way of saving a snapshot. I therefore used B-Em and tried again. Eventually I managed to crash and get stuck in the same way again - it seems to be after the 3rd corner.

I've saved state from B-Em and attached it to this message - I hope this helps Mark debug the issue.
I have noticed that this save-state crashes later versions of B-Em, so you'll probably have to use the version I was using which is v2.2
I was using a BBC B + 8271 FDC. I was playing the game with the keyboard.

Finally I would like to say the disassembly of Revs is superb. (And Elite too!) Thank you @MarkMoxon for all the effort you've put into this and for sharing it with us.

[MarkMoxon]

I've saved state from B-Em and attached it to this message - I hope this helps Mark debug the issue.

Amazing! That's helped me fix the bug. Thanks jimmy, the saved state was just what I needed.

I've uploaded a new version here that hopefully doesn't crash in the same way:

https://revs.bbcelite.com/versions/revs-plus.ssd

I've had a quick spin and all seems well, but if anyone still has problems, do let me know.

For those interested, here's what was causing the infinite loop. One of the software hooks that the extra tracks use is this change to the way the car gets moved backwards along the track:

Code: Select all

.HookMoveBack

 BIT playerPastSegment  \ If bit 7 of playerPastSegment is set, return from the
 BMI HookMoveBack-1     \ subroutine (as HookMoveBack-1 contains an RTS)

 JMP MovePlayerBack     \ Move the player backwards by one segment, returning
                        \ from the subroutine using a tail call

I copied this hook straight from one of the other tracks into the Nürburgring file, but I had to shuffle the order of the hook routines to fit them in (all the track files are slightly different because of this - they truly are hand-crafted). As a result, the instruction before the HookMoveBack routine, which is an RTS in all the other track files, was no longer an RTS in the Nürburgring, so the BMI HookMoveBack-1 no longer jumped to an RTS, and instead created an infinite loop.

I've fixed it by changing BMI HookMoveBack-1 in the Nürburgring file to point to another RTS instruction, so the hook routine should now work rather than crashing.

You can see this fix in action by loading jimmy's saved state into B-Em, opening the debugger and typing:

Code: Select all

writem 53F6 67

This fixes the destination of the BMI instruction to point to an RTS, so then typing:

Code: Select all

c

should start the game up again, showing you the crash fence and returning you to the track, so you can pick up where things went wrong.

Anyway, a very satisfying result! Thanks for the feedback everyone; hopefully things will work properly from now on.

Finally I would like to say the disassembly of Revs is superb. (And Elite too!) Thank you @MarkMoxon for all the effort you've put into this and for sharing it with us.

My pleasure! I'm glad you like them. They're really fascinating programs to pick apart...

Mark

[chrisn]

Well done, thank you jimmy for capturing a save state, and Mark for tracking down and fixing the bug so quickly.

I'll try out your updated version. Other than that one issue, I haven't found anything else amiss. I tried a race, and the other drivers were much faster than me, and somehow don't crash all the time...

Perhaps we could start separate threads for our best lap times on each of the tracks?

[jms2]

I found the same you Chris. Although my lap times were OK in practice mode, in race mode the other drivers were massively faster in the corners and I couldn't keep up. Overtaking on the straights was OK though.

There's not a "friction level for the tarmac" that varies between tracks is there?

[MarkMoxon]

Although my lap times were OK in practice mode, in race mode the other drivers were massively faster in the corners and I couldn't keep up. Overtaking on the straights was OK though.

There's not a "friction level for the tarmac" that varies between tracks is there?

Well, there’s a base speed that’s used when calculating the speed of the other drivers, and that's pretty much the same thing as a friction level, so yes, I guess there is. I could easily alter this value, and that would slow down all the other drivers across the board while leaving our speed alone.

But if the issue is that drivers are going round corners too fast but are otherwise reasonably correct, then this might indicate an issue with the data that sets the optimum racing line and speed for each section - perhaps the corner sections are out of kilter in some way. The whole area of driver tactics is next on my list to analyse, as I've only scratched the surface; I'm going to look at it once I've finished documenting the track data.

It turns out that tuning a Revs track is a bit like tuning a classic car. That feels apt.

I'll post here when I have something that makes the drivers slow down around corners. Hopefully it won't take long, and thanks again for all the feedback - it’s proving really useful!

Mark

[jimmy]

Amazing! That's helped me fix the bug. Thanks jimmy, the saved state was just what I needed.

Glad I could help!

Mind you, playing Revs again has reminded me why I didn't enjoy this game - it's just too hard. Playing on keyboard I find the steering wheel every so often won't straighten up - naybe this doesn't happen on real hardware.

I find Psion's Chequered Flag on the Spectrum easier to play - there's more made up tracks on there that could be implemented in Revs? ("Psion Park", "Micro Drive" and "Cambridge Ring"), plus real tracks of "Monza","Paul Ricard","Osterreichring" and "Monaco".

[chrisn]

Although my lap times were OK in practice mode, in race mode the other drivers were massively faster in the corners and I couldn't keep up. Overtaking on the straights was OK though.

Yes, the other cars were just flying round the corners, where I'm having to slow right down.

Here's my qualifying time in Novice mode. I thought 1:40.8 was pretty good, but I qualified last as I'm just under the 1:41 threshold for Professional class. 1:26.2 seems to be quite a bit too quick though.

So I wondered how it compares to the C64 version. Here's a screenshot from a video I found. The times here seem more reasonable for Novice class.

Digging a bit further I found someone posted their best lap time on the C64 wiki. With a lap time of 1:41.0, this person also came last, but the fastest time was 1:37.2. That seems a lot more reasonable for Professional.

[jms2]

Those C64 times look like what Mark should be aiming to tweak the code towards. 1:26 is completely impossible!

Oddly though, the distances are more similar than I expected - Nurburgring seems to be 4.569 km for the GP circuit that looks like the one in the driving guide (https://www.racingcircuits.info/europe/ ... gring.html), whereas Silverstone was 4.718 km in 1984. Silverstone is a higher speed circuit though.

[MarkMoxon]

Those C64 times look like what Mark should be aiming to tweak the code towards. 1:26 is completely impossible!

Absolutely!

Happily, I think I've found the cause. There are a number of track stats that are scaled in the C64 version when compared to the BBC, and all by a factor of 1.44 - not sure why it's that value, but I suspect it's something to do with the relative speed of the game's main loop on the Beeb and C64.

I've had a look, and it turns out that the corner approach speeds are scaled in this exact same way, but I simply copied them from the C64 to the BBC, without reversing the scaling. This means the non-player cars are currently driving around corners 1.44 times faster than they should be.

I just need to work out which entries need scaling back, as this only applies to some sections, probably the ones that precede corners. Once these speeds are scaled back then hopefully the non-player drivers will slow down properly for corners, and their overall times will drop to fit in with Chris's examples.

Thanks for the clues - they've made it so much easier to pinpoint the cause. I should be able to come up with a fixed version in the next couple of days.

Oddly though, the distances are more similar than I expected - Nurburgring seems to be 4.569 km for the GP circuit that looks like the one in the driving guide (https://www.racingcircuits.info/europe/ ... gring.html), whereas Silverstone was 4.718 km in 1984. Silverstone is a higher speed circuit though.

I think that Silverstone is a higher speed circuit not only because it's a straighter track, but because it's much flatter, being on an old airfield. The Revs track file for the Nürburgring modifies the game code to remove a restriction on the track's y-coordinates, which are used to describe the elevation of the track. No other tracks have this modification, and as a result the Nürburgring is five times as hilly as Silverstone, so that will affect the speeds quite a bit.

Though not as much as getting the cornering speeds wrong! I'll be back with a new version soon. Thanks again for the testing - it's proving really useful, especially as I don't have the driving skills to flush out these kinds of issues...

Mark

[jms2]

That's sounding great Mark. I'm glad you're still finding out new things about the game.

Nurburgring is an especially difficult track as it has several corners with blind apexes and where you have to commit to winding on a lot of lock and basically get it right first time, otherwise you crash. This is a track where a proper analogue driving wheel would make it a lot easier. I did try to buy an old one off Ebay recently but I was outbid.

[MarkMoxon]

OK, I have new version that should tackle the cornering and speed issues. It's in the usual place:

https://revs.bbcelite.com/versions/revs-plus.ssd

I've applied the correct scaling to both the corner approach speeds, and to the optimum racing lines, as both of these are scaled between the C64 and BBC versions, and I'd missed that out. I've checked the rest of the track data, and I don't think there is anything else to scale, so hopefully this is getting really close.

I've also adjusted the overall driver speeds to try to get the BBC times in sync with the C64 times, so the other drivers should now slow down for corners, should take less extreme racing lines and should drive at the right speeds. Looking at their speeds when a race is running, they are definitely braking a lot more, so hopefully this has done the trick.

I've fired up my C64 emulator and extracted times for all three classes, so this is how they now compare.

For Novice:

and Amateur (for which you need to be 1:45 or better):

and Professional (for which you need to be 1:41 or better):

It also turns out that the two games use a different randomising routine - the Beeb has a much higher chance of generating really fast or really slow entrants in each class, particularly in Novice, while the C64 seems much more regular in its time tables. I've therefore tried to match the bulk of the middle table best, ignoring the first and last fews times as outliers.

This is certainly an improvement in terms of times and looking at the driver tables during a race, but whether it feels correct when driving is another matter! It would be great to know how you get on...

Mark

[jms2]

Well that was fun!

I had a go at a Novice race. I qualified sixth and finished fifth, having crashed in the race twice. So had I not crashed, I could have won it.

I got the second fastest laptime in the race, at 1:40.6. The middle-of-the-pack laptime was 1:45.0.

Congratulations, I'd say you've fixed it!

[MarkMoxon]

Congratulations, I'd say you've fixed it!

Excellent! Thanks for checking it out. I think we might finally have the Nürburgring properly backported to the original (and best!) Revs platform.

If anyone else manages to have a play, do let me know. The more the merrier!

Mark

[BigEd]

The Revs track file for the Nürburgring modifies the game code to...

Wow, that's an interesting feature!

Great work on this BTW. I see some of the Nürburgring curves are banked but I imagine the physics engine can't do anything about that.

[MarkMoxon]

The Revs track file for the Nürburgring modifies the game code to...

Wow, that's an interesting feature!

Actually, all the extra track files modify the game code heavily; the only one that doesn't is Silverstone. In fact, in the five extra tracks (including Nürburgring) there is more code in the track data files than data. The Silverstone track is 100% data, but all the others generate the bulk of their track coordinates dynamically, using code that is injected into the main game file after loading. It's really quite mind-bending.

I'm writing a short series of deep dives on the very subject. These extra track files have turned out to be a lot more interesting that I thought they'd be! I hope to release these in the next couple of weeks, as real life's a bit busy at the moment...

Great work on this BTW. I see some of the Nürburgring curves are banked but I imagine the physics engine can't do anything about that.

Indeed. All track segments in Revs are horizontally level (i.e. from left to right). Segments can slope up or down as you move along the track, but the physics engine doesn't support camber or banking, so that's why there are no banked curves in the Revs version.

Stunt Car Racer has to have something that Revs doesn't, after all.

Mark

[MarkMoxon]

Thank you for all the feedback on the Nürburgring track. In case anyone is interested, I’ve written up an account of the backporting process (it was way more involved than I thought it would be!):

https://revs.bbcelite.com/deep_dives/ba ... track.html

I’ve also documented how the extra track files modify the main game code, which is worth a read:

https://revs.bbcelite.com/deep_dives/se ... racks.html

Enjoy the new track!

Mark

[jms2]

Thanks for providing more fascinating material to read Mark!

One question regarding the dynamic track generation - doesn't this slow the game down a bit? Presumably it does, but this is not noticeable because the animation is still locked to a fixed number of frames per second and there was a bit of spare time available?