Polyphony have low key made the best F1 game of the 90's

[Voodoovaj]

I think this needs to be highlighted way more than it is, but making a custom race with the F3500's results in the best F1 game I have played in a LONG time. It might even be better than EA's F1 games, but I haven't played those for more than a decade.

Sorry for the camera work. It's 21 minutes, but this was the best single player race I have ever had in this game in 30 years.


This is the regular AI (because Sophy does not officially support the F3500B nor the tracks that I want to use. With regular cars, Reggie is terrible. With the F3500, it's somehow really good. Yes, it still does some strange things here and there, but it's a night and day difference to the AI with everything else.

Now - to be fair - this was the best racing game of the early 90's

So, I didn't have far to go to make something better, but still I wanted something where I worried more about beating my teammate than winning the race.

What did I do? Not much. The cars are modelled after their real world counterparts in terms of power, and I have changed transmissions to match. H-pattern cars in real life get the semi-race transmission. Sem-auto in real life gets the normal transmission, and I am reserving full race transmissions for later years. This results in a 30pp difference from lowest rated car to highest rated, but that doesn't seem to matter much. I used a random number generator on the anti-roll bars. Sounds strange, but it works to alter the behavior of each car ever so slightly. Same with the initial torque (random from 5-10).

I made a spreadsheet to track race results, and I start each AI based on the finish of the last race. My start position is a random number from 1-20. I started the season's first race based on the official final results for the year I am emulating. From then on, the only rule I have is no restarts. What happens, happens. Boost is weak (we all know about this). Damage is HEAVY - this is super important because it mixes up the results and it forces you to be careful. Tire wear is 5X and fuel is 2X over a 15lap race, so no need for a stop, but there is a need for fuel management at some events. Weather is a random number from 1-32 (because there is a total of 32 varieties). For the first year I let the game choose random weather, but it was always sunny, so I decided to change it up. The video above is 30 out 32.

I choose my team at the start of the year with a random selection as well. I made an alphabetized list of teams and chose a random number from 1-10. For 1992, I drive for Minardi.

Nothing is predictable. I can't explain what is happening. For some reason, these guys do all the things that we wished the entire game would do all the time. I tried this exact same set up with the other cars, and the results are nowhere near as good, so I stick to Sophy for everything else. That said, I am going to use what I've found here and apply it there. I've already found it helps improve my DTM custom race.

I don't know if this is something that is intentional, or something emergent from the combination of factors, but either way, this is the best single player racing I have ever had.

 

[Niko-H]

Fun stuff! I recently finished a pair of F1 90s collections - one for the A and one for the B. The weekly challenge race inspired me to do a few custom races. Totally agree on using HEAVY damage - it's a perfect proxy for the old school F1 chaos when half the field DNF'ed.

Did you use a mix of the engine swaps? If so, how did you balance the V10 with the V12 and V8?

 

[Speedster911]

I distinctly remember Psygnosis made an F1 game on the PS1 I think it was 94' or 96' I can't be sure, but it was downright incredible, with very realistic sounding engines, photorealistic graphics, superb physics and Murray Walker as the commentator. Too bad they never made any more games after that, could be wrong, but I miss that one to this day. Had lots of Joe Satriani tracks on the OST too! Fell in love with "summer song" after playing that F1 game in fact.

 

[Voodoovaj]

Fun stuff! I recently finished a pair of F1 90s collections - one for the A and one for the B.

You can mix them with no issues. They work great. Honestly, you don't need Sophy for this. I can't explain why, but this works great with the regular AI.

The weekly challenge race inspired me to do a few custom races. Totally agree on using HEAVY damage - it's a perfect proxy for the old school F1 chaos when half the field DNF'ed.

Did you use a mix of the engine swaps? If so, how did you balance the V10 with the V12 and V8?

Yes I did.

For the V10, I use it sparingly because it's so OP. Again, I used the real world stats and specs to set up the cars. With the V10, I start using it in the FW14 with it tuned down to about 720hp in 1991 and 750 in 1992. For 1992, I also swapped the V10 into the McLaren Mp4/7 with 768 HP, because the MP4/7 still had an H pattern box. That gearbox has cost Senna 3 championships so far. Prost won my 1990 series, Alesi won 1991, and Mansell won 1992

As I move forward through the years, I will use that V10 for the top teams for those years.

I distinctly remember Psygnosis made an F1 game on the PS1 I think it was 94' or 96' I can't be sure, but it was downright incredible, with very realistic sounding engines, photorealistic graphics, superb physics and Murray Walker as the commentator. Too bad they never made any more games after that, could be wrong, but I miss that one to this day. Had lots of Joe Satriani tracks on the OST too! Fell in love with "summer song" after playing that F1 game in fact.

Yep, that inspired me. With what PD has provided us, an F1 Series is MUCH better than those games

 

[Voodoovaj]

I finished the 1992 season. Here's the final results. I got 2 wins across 18 rounds, a far cry from the way any current in game championship plays out. No points means that I finished outside the top 10.

I find it fascinating how it played out. At first, Alesi was in control. Then Patrese started to come on strong. Senna had some rough races, but he was always there. It's Mansell who went from 3 non-scoring races to a dominant end of season.

Maybe this was always there and we never noticed, but I'm taking notice now.

For 1990 and 1991, I was with Ligier (again through random choice). Now, after 1992 with Minardi, random selection has given me...

Minardi again

 

[Voodoovaj]

FYI, as I begin this 1993 Season, just a reminder of what 1993 Formula 1 video games looked like.

Seriously, I HIGHLY HIGHLY recommend making this set up...just ONE. You can make and entire field of 1990 or 1991 cars for the cost of one 20mil car that you'll never use. The excel sheet set up is easy as well. This is what they should have done with the power pack, there would be FAR fewer complaints.

 

[KAlex122]

Isn't there a remake of some old F1 games coming out? Forgot the name.

 

[Famine]

Isn't there a remake of some old F1 games coming out? Forgot the name.

Yes - the one in the image:

https://www.gtplanet.net/geoff-crammond-return-steam-2026-20251217/

 

[ZEE]

I think this needs to be highlighted way more than it is, but making a custom race with the F3500's results in the best F1 game I have played in a LONG time. It might even be better than EA's F1 games, but I haven't played those for more than a decade.

Sorry for the camera work. It's 21 minutes, but this was the best single player race I have ever had in this game in 30 years.


This is the regular AI (because Sophy does not officially support the F3500B nor the tracks that I want to use. With regular cars, Reggie is terrible. With the F3500, it's somehow really good. Yes, it still does some strange things here and there, but it's a night and day difference to the AI with everything else.

Now - to be fair - this was the best racing game of the early 90's

So, I didn't have far to go to make something better, but still I wanted something where I worried more about beating my teammate than winning the race.

What did I do? Not much. The cars are modelled after their real world counterparts in terms of power, and I have changed transmissions to match. H-pattern cars in real life get the semi-race transmission. Sem-auto in real life gets the normal transmission, and I am reserving full race transmissions for later years. This results in a 30pp difference from lowest rated car to highest rated, but that doesn't seem to matter much. I used a random number generator on the anti-roll bars. Sounds strange, but it works to alter the behavior of each car ever so slightly. Same with the initial torque (random from 5-10).

I made a spreadsheet to track race results, and I start each AI based on the finish of the last race. My start position is a random number from 1-20. I started the season's first race based on the official final results for the year I am emulating. From then on, the only rule I have is no restarts. What happens, happens. Boost is weak (we all know about this). Damage is HEAVY - this is super important because it mixes up the results and it forces you to be careful. Tire wear is 5X and fuel is 2X over a 15lap race, so no need for a stop, but there is a need for fuel management at some events. Weather is a random number from 1-32 (because there is a total of 32 varieties). For the first year I let the game choose random weather, but it was always sunny, so I decided to change it up. The video above is 30 out 32.

I choose my team at the start of the year with a random selection as well. I made an alphabetized list of teams and chose a random number from 1-10. For 1992, I drive for Minardi.

Nothing is predictable. I can't explain what is happening. For some reason, these guys do all the things that we wished the entire game would do all the time. I tried this exact same set up with the other cars, and the results are nowhere near as good, so I stick to Sophy for everything else. That said, I am going to use what I've found here and apply it there. I've already found it helps improve my DTM custom race.

I don't know if this is something that is intentional, or something emergent from the combination of factors, but either way, this is the best single player racing I have ever had.

Really love reading yur post to the OP!!! That particular Ayrton senna game was my first contact with Motor racing on a console, SEGA AYRTON SENNA's SUPER MONACO GP II, when I was weeeee little boy, playing this game terribly on my first ever console purchased by my Dad as gift, the Sega MEGAdrive! I played it for hours and hours as well as Sonic the hedgehog 1, which was also a "fast" game , and kept destroying F1 cars left right and centre to get a good qualifier lap!

At the time I though this game was very hard but also very rewarding. It felt so immersive and intense! I never got to finish it, but if GT7 can now, 30yrs later on PS5, recreate such quality F1 old skool racing then I will damn well give it a go! I am also fascinated at your AI results with each race and how that has turned out, very cool!

That game in context gave me a strong love & passion for all racing games to come soon after, like the EA F1 game series, NFS, GT, Forza, VIRTUA RACING, DAYTONA, SEGA TOURING CAR, SCUD RACE & SEGA RALLY series, and the list goes on and on!!! Without this game to spark my imagination and love for F1 & Motor racing in general, I wouldnt be where I am right now and playing GT7.

So I look forward to testing this custom races and sharing my results in the future!

 

[Voodoovaj]

At the time I though this game was very hard but also very rewarding. It felt so immersive and intense! I never got to finish it, but if GT7 can now, 30yrs later on PS5, recreate such quality F1 old skool racing then I will damn well give it a go! I am also fascinated at your AI results with each race and how that has turned out, very cool!

Here are the rules for making this all work, and I will explain why for each point. It's a combination of race settings and car settings. Some will sound strange, but trust me. What I want to focus on is the emergent nature of the settings.

1. The player is assigned a team through a random selection of the ten available teams - Why? Because otherwise you would always choose a top team, and that's less fun.
2. Each team can only be a 2 car team max, so the player must only have a maximum of 1 teammate. Keep it fair.
3. No restarts - Why? Because you have to treat every other driver as a real person, not just an AI. We often rage quit (or rage restart). Pretend this is Sport Mode. Of course, sometimes a restart cannot be avoided, but don't do it on purpose
4. 15 laps, 3X tire wear, 5X fuel usage - This combination results in unpredictable results from race to race. Some races eat tires, some races burn fuel. The car set ups will alter each driver's wear and fuel, again resulting in unpredictable results. You will have to decide, from race to race, what you are going to do and you won't really know what the others will do.
5. Heavy damage, strong short cut penalty, flag rules, pitlane penalty - Again, adds to the unpredictable nature. Also, heavy damage forces you to be careful and mindful of where you put your car. With no restarts, you could put yourself out of the points. Also, heavy damage shakes up the field like in real life.
6. Weak boost - We all know what this does.
7. Race start order is based on "fastest first". Top 10 drivers are ordered by their previous race finishing position, EXCEPT for the player.
8. Bottom 10 drivers are ordered by their points standings - This helps to provide championship continuity. A driver who had a bad race can fight his way back up the ranks, and they do.
9. Player start position is random - I use a random number generator - Don't underestimate this.
10. Get a random number generator to choose the weather instead of "random" in game. Random in game is unreliable and changes between several major settings. Each setting already has some unpredictability in it, so make it like real weather, UNPREDICTABLE.

Points:

Position	Points
1​	15​
2​	12​
3​	10​
4​	8​
5​	6​
6​	5​
7​	4​
8​	3​
9​	2​
10​	1​

Car Setup - As I mentioned, try to stick to the real world info for the car. Wikipedia is really good for this. But, beyond that, you have to make a few adjustments.

Get a random number generator and adjust the anti roll bars to a random number from 1-10 for the front, and again for the rear. Adjust each car individually. Why? This adjustment creates slight behavioral differences in each car, and those behaviors change from track to track, and as fuel load drops, and tires wear. This slight, almost imperceptible difference adds to the unpredictability.

Do something similar for the diff. Adjust initial torque in a range of 5-10, acceleration sensitivity from 20-25, and braking from 20-30. You'll barely notice it, but it's just enough.

Lastly, why this works (as far as I can see). The AI aggression/speed seems to be based on their on track position and their performance relative to the cars immediately ahead or behind them. So, 1st place will start off as the most aggressive/fastest on track and last will start as the slowest/least aggressive. However, when the unpredictability hits, you will see the AI fighting for positions. Most of that happens at the beginning, but those tire and fuel settings can make it happen at the end. Lastly, it seems that these behavior changes are maintained for a few laps. They don't change behavior right away. Lastly, obviously, the behave in different ways relative to you, and in trying to keep you behind the, sometimes they end up fight for track position with the others.

That's why in the OP, Johansson in that Footwork managed to come up from 10th. As he gained position, he gained aggression and speed, and with all the other factors, it created a cascade effect where he because the dominant driver.

Does it work with the other cars? It somewhat does, more so with Sophy than Reggie. I think it's because the other cars lack the grip and power to capitalize on the behavior changes.

I've been playing only this for days now. It isn't perfect, but it's really good.

 

[Salsamuffin]

Here are the rules for making this all work, and I will explain why for each point. It's a combination of race settings and car settings. Some will sound strange, but trust me. What I want to focus on is the emergent nature of the settings.

1. The player is assigned a team through a random selection of the ten available teams - Why? Because otherwise you would always choose a top team, and that's less fun.
2. Each team can only be a 2 car team max, so the player must only have a maximum of 1 teammate. Keep it fair.
3. No restarts - Why? Because you have to treat every other driver as a real person, not just an AI. We often rage quit (or rage restart). Pretend this is Sport Mode. Of course, sometimes a restart cannot be avoided, but don't do it on purpose
4. 15 laps, 3X tire wear, 5X fuel usage - This combination results in unpredictable results from race to race. Some races eat tires, some races burn fuel. The car set ups will alter each driver's wear and fuel, again resulting in unpredictable results. You will have to decide, from race to race, what you are going to do and you won't really know what the others will do.
5. Heavy damage, strong short cut penalty, flag rules, pitlane penalty - Again, adds to the unpredictable nature. Also, heavy damage forces you to be careful and mindful of where you put your car. With no restarts, you could put yourself out of the points. Also, heavy damage shakes up the field like in real life.
6. Weak boost - We all know what this does.
7. Race start order is based on "fastest first". Top 10 drivers are ordered by their previous race finishing position, EXCEPT for the player.
8. Bottom 10 drivers are ordered by their points standings - This helps to provide championship continuity. A driver who had a bad race can fight his way back up the ranks, and they do.
9. Player start position is random - I use a random number generator - Don't underestimate this.
10. Get a random number generator to choose the weather instead of "random" in game. Random in game is unreliable and changes between several major settings. Each setting already has some unpredictability in it, so make it like real weather, UNPREDICTABLE.

Points:

Position	Points
1​	15​
2​	12​
3​	10​
4​	8​
5​	6​
6​	5​
7​	4​
8​	3​
9​	2​
10​	1​

Car Setup - As I mentioned, try to stick to the real world info for the car. Wikipedia is really good for this. But, beyond that, you have to make a few adjustments.

Get a random number generator and adjust the anti roll bars to a random number from 1-10 for the front, and again for the rear. Adjust each car individually. Why? This adjustment creates slight behavioral differences in each car, and those behaviors change from track to track, and as fuel load drops, and tires wear. This slight, almost imperceptible difference adds to the unpredictability.

Do something similar for the diff. Adjust initial torque in a range of 5-10, acceleration sensitivity from 20-25, and braking from 20-30. You'll barely notice it, but it's just enough.

Lastly, why this works (as far as I can see). The AI aggression/speed seems to be based on their on track position and their performance relative to the cars immediately ahead or behind them. So, 1st place will start off as the most aggressive/fastest on track and last will start as the slowest/least aggressive. However, when the unpredictability hits, you will see the AI fighting for positions. Most of that happens at the beginning, but those tire and fuel settings can make it happen at the end. Lastly, it seems that these behavior changes are maintained for a few laps. They don't change behavior right away. Lastly, obviously, the behave in different ways relative to you, and in trying to keep you behind the, sometimes they end up fight for track position with the others.

That's why in the OP, Johansson in that Footwork managed to come up from 10th. As he gained position, he gained aggression and speed, and with all the other factors, it created a cascade effect where he because the dominant driver.

Does it work with the other cars? It somewhat does, more so with Sophy than Reggie. I think it's because the other cars lack the grip and power to capitalize on the behavior changes.

I've been playing only this for days now. It isn't perfect, but it's really good.

Hi,

I joined this forum after reading your post. I am also making a 90s F1 single player on GT7 but I am basing it on 1997. I bought the 10 F3500A cars. I changed the drivers names. I've been using Sophy AI with weak boost so far. I am interested in what you said about the regular AI being good enough. Is it as fast as Sophy with this car?

Also, I am curious how much difference in PP you make between the cars for each team. I've been adjusting power output and downforce to get a spread of about 900 - 993 PP for the 10 teams.

I'm going to try the regular AI and see how that goes next time although I guess I'll have to start again with the car and driver name edits.

Playing this F1 series also takes me back to playing Microprose GP2. Thanks for your tips and please share any more that come to mind!

 

[Voodoovaj]

Hi,

I joined this forum after reading your post. I am also making a 90s F1 single player on GT7 but I am basing it on 1997. I bought the 10 F3500A cars. I changed the drivers names. I've been using Sophy AI with weak boost so far. I am interested in what you said about the regular AI being good enough. Is it as fast as Sophy with this car?

Yes and no. Reggie still does some strange things, and it is more affected by the race on the whole rather than racing you specifically. You will have to be at your best to get wins, just as you would with Sophy. Just be aware that Reggie can make dumb decisions (like a person might).

The big advantage is that you are not restricted to the Sophy tracks and the 3500A.

Also, I am curious how much difference in PP you make between the cars for each team. I've been adjusting power output and downforce to get a spread of about 900 - 993 PP for the 10 teams.

I don't touch the down force. I ignore performance points.

I hit Wikipedia for the specs. Then, I adjust the anti-roll bars and differential to get slight differences. For instance, two cars with different ARB settings will react differently over the curbing, which is just enough to cause a change of position.

EDIT - I give a little extra to "better" drivers. So, for Senna, I give the McLarens more. For the 1991 season, I gave the McLaren a V10 swap to allow with a detune to compete with the Williams. For 1993, instead of a V8, I gave the McLarens a V12 with a detune because Wikipedia says the power was between 640hp-730hp. Since I cannot make the V8 more powerful, I turn the V12 down.

I am driving a Minardi M193, and I have tuned it down to under 700hp, as would be the case with a customer Ford V8

I'm going to try the regular AI and see how that goes next time although I guess I'll have to start again with the car and driver name edits.

Awesome. Remember, the magic happens when you track the races. I had two big wins with my customer engine Minardi (698hp V8). We had light rain at Spa and I got lucky, then I got lucky again with traffic issues at Monza (I use the no chicane version). Then, while feeling like I might have an easy ride to a championship, I finished 8th at Dragon Tail gardens.

The last two points are why I think Reggie has an advantage. It allows both cars, and all tracks.

Playing this F1 series also takes me back to playing Microprose GP2. Thanks for your tips and please share any more that come to mind!

Exactly!! I loved those games back then. I loved the Psygnosis series as well. With this, it's like making a new game without spending money!!

 

[Salsamuffin]

Yes and no. Reggie still does some strange things, and it is more affected by the race on the whole rather than racing you specifically. You will have to be at your best to get wins, just as you would with Sophy. Just be aware that Reggie can make dumb decisions (like a person might).

The big advantage is that you are not restricted to the Sophy tracks and the 3500A.

I don't touch the down force. I ignore performance points.

I hit Wikipedia for the specs. Then, I adjust the anti-roll bars and differential to get slight differences. For instance, two cars with different ARB settings will react differently over the curbing, which is just enough to cause a change of position.

EDIT - I give a little extra to "better" drivers. So, for Senna, I give the McLarens more. For the 1991 season, I gave the McLaren a V10 swap to allow with a detune to compete with the Williams. For 1993, instead of a V8, I gave the McLarens a V12 with a detune because Wikipedia says the power was between 640hp-730hp. Since I cannot make the V8 more powerful, I turn the V12 down.

I am driving a Minardi M193, and I have tuned it down to under 700hp, as would be the case with a customer Ford V8

Awesome. Remember, the magic happens when you track the races. I had two big wins with my customer engine Minardi (698hp V8). We had light rain at Spa and I got lucky, then I got lucky again with traffic issues at Monza (I use the no chicane version). Then, while feeling like I might have an easy ride to a championship, I finished 8th at Dragon Tail gardens.

The last two points are why I think Reggie has an advantage. It allows both cars, and all tracks.

Exactly!! I loved those games back then. I loved the Psygnosis series as well. With this, it's like making a new game without spending money!!

Cool, which tracks do you find work best with this set up? Any tracks that really didn’t work? So far I’m mostly sticking to the 7 or F1 tracks in the game.

Also what kind of tire and fuel use would I need to force a pit stop in a 15 or so lap race?

Thanks for all the info. I’m loving the 90s nostalgia but mostly really enjoying driving these F1 cars.

 

[KAlex122]

Great. Now i have to learn to perform in high downforce cars in GT7.

 

[Salsamuffin]

Spent a few hours today trying the F1 1997 season with Reggie. Lots of fun and plenty of challenge for me.

The AI is a bit less aggressive and less defensive than Sophy. Overall it’s very good though. Much better than with other cars as the OP said.

My only real downside is how crazy the grid starts are. The AI happily plows into the back of me while I’m in traffic waiting to get through the first corner, especially at Monza. I’m left wondering if the causing a crash penalty also applies to them?

So much fun and I can feel
My driving slowly improving as I race more and more with it.

 

[Voodoovaj]

Cool, which tracks do you find work best with this set up? Any tracks that really didn’t work? So far I’m mostly sticking to the 7 or F1 tracks in the game.

Also what kind of tire and fuel use would I need to force a pit stop in a 15 or so lap race?

Thanks for all the info. I’m loving the 90s nostalgia but mostly really enjoying driving these F1 cars.

Here is what I am using.

1 Tokyo South Clockwise
2 Interlagos
3 CGV
4 Blue Moon Infield A
5 Lago Maggiore Full Reverse
6 Le Mans No Chicane / 6 laps
7 Brands Hatch Full Course
8 Nurburgring GP Circuit
9 Red Bull Ring Full Course
10 Spa
11 Monza No chicane
12 Dragon's Tail Gardens
13 Catalunya No chicane
14 Suzuka Full Course
15 Autopolis Full
16 Fuji Short Course
17 Kyoto Driving Park Yamagiwa + Miyabi Forward
18 Yas Marina

 

[Voodoovaj]

This is what the AI can do in this event (again, why it's different with the F3500's, I can't say for sure).

I started the video so you can see the last three laps of me versus A.I. Senna - If only the entire single player could be like this.