Gran Turismo 7 Custom Race thread

[Theufcveteran]

Hey, do some of the changes discussed here affect "regular" (as in non-Sophy) custom races? Or haven't those been changed at all?

No change unfortunately. AI throttle is still 97%. I haven't checked the rubberbanding for Weak boost against regular AI though. That may have changed. I'll do some tests when I can and report back.

It would be great if the slowdown when behind by over 10 seconds and when near the end of the race is gone.

 

[DapperDan07]

I wonder why sophy isn’t available on the 24hr version of spa. Hopefully this gets added soon along with watkins glen

 

[Theufcveteran]

Did some testing. This applies both in the World Circuits menu when setting up a custom race, and in the pre-race menu.

---

If you assign the AI your car,

And then change any car settings, such as the ECU/power restrictor, and then go back into the custom race settings,

If you at all click X on "Start Race", even if you don't go back into the "Select from Garage" option, it will give the AI your current state of tune/car settings.

This also counts for cars which have been assigned to an AI in one specific state of tune, the event settings have been saved, then you leave the custom race settings menu in World Circuits, or the event itself, and change the car's settings, such as its power, outside of the event.

Then, you change back into a different car and re-load the event in World Circuits.

Once you click "Start Race" in the custom race settings menu after loading the event, the AI who had previous been using a different setup, now has the current stats of the car as it stands in your garage, even though when you saved the event, they had, for example, a more powerful tune.

---

More succinctly, if you confirm the custom race settings at all, any assigned AI cars via "Select from Garage" will be updated to the current state of tune as they stand in your garage.

This happens regardless of any changes to the car's settings since assigning the AI that car with that specific level of performance at the time of assignment.

On a related note, if you press circle to back out of the custom race settings, such as while in the pre-race menu, then no changes apply.

It's only once you press X on "Start Race", even if you didn't change any event settings.

---

To add on, when in the custom race settings, if you simply back out by pressing circle, in all cases related to the Rivals Selection/Select From Garage, no change appears for the AI car.

It will have the performance that the car had at the time of assignment.

The only function this serves is basically to know that, whatever tune the AI car has at the time of entering the custom race menu, backing out will not change it no matter what you do in the Rivals Selection category, even if you change it to Random and back to Select From Garage.

Because you have to press X to apply any changes, it doesn't get around the updated car performance when pressing X on "Start Race".

But you don't have to worry about any accidental or then changed custom race settings adjustments, assuming you back out and don't confirm them, altering the current performance of the AI car on your grid.

If you apply any changes in the custom race settings, all AI cars will have their performance updated to the same the cars stand in your garage.

 

[suomi1]

So I did a long testing session this afternoon at Le Mans, as I planned to do yesterday. 12 short races of 3 laps each to see if anything had changed regarding boost / rubberbanding.

I raced against few GT replica cars with my TS020 to be able to trigger AI's behavior without too much trouble.

The race setup is pretty simple, everything is set on realistic with tire / fuel on (1), mechanical damages off, professional difficulty, no penalties or flag rule, weather off.

On the table below, when I said I'm first, I'm in front of the pack and push as fast as I can (which is really easy with such a car). But when I'm last, consider I'm trying to stay in a range of 2s/3s down on the AI car just in front of me.

Note that the regular AI is less consistent than Sophy and can make mistakes, I did not seek to smooth out these errors by reproducing the experiment until I had an optimal result.
The first car in front of the player tend to be the slowest one, and here the Porsche starts last everytime. If the player is first, then the first car in front of him is... the last one too, so stays the slowest.
Also, take into account that even if I am driving a Toyota GT-One, I can set 4'11'xxx / 4'12'xxx laps with the same cars as the AI, so that you have a human point of comparison.

Here are the data :

You can draw your own conclusions, but Sophy has without any doubt the same virus as the base AI. It's even far worse when boost is set to Off.

Sophy is overall faster than PD's AI when boost is set on Weak or High (about 5s), but the gap between its slowest pace and fastest pace stays more or less the same as for the regular AI, somewhere between 17s to 19s which is still huge and atrocious...

Now let say I'm racing with the same car with boost weak. As I said above, my pace is around 4'11/12 :
PD's AI will be 13 seconds faster than me when I'm in front, and 6 seconds slower when I'm behind.
Sophy will be 16 seconds faster than me when I'm leading, and 3 seconds slower when I'm behind.

Considering both are plagued with RB, I still prefer Polyphony's AI over Sophy, it sounds more fair to me (especially for long races). Even more so when PD's AI sometimes makes mistakes that bring more fun.

I don't think I can continue to invest my time in this game anymore. At this point it sounds pretty clear that they are stuck with this vision and nothing will change. They can bring in all the Nissan R390 GT1 or Panoz Esperante they want, if I can't have a proper race there is no point. Sad, this update had given me some hopes...

 

[Theufcveteran]

So I did a long testing session this afternoon at Le Mans, as I planned to do yesterday. 12 short races of 3 laps each to see if anything had changed regarding boost / rubberbanding.

I raced against few GT replica cars with my TS020 to be able to trigger AI's behavior without too much trouble.

View attachment 1439274

The race setup is pretty simple, everything is set on realistic with tire / fuel on (1), mechanical damages off, professional difficulty, no penalties or flag rule, weather off.

On the table below, when I said I'm first, I'm in front of the pack and push as fast as I can (which is really easy with such a car). But when I'm last, consider I'm trying to stay in a range of 2s/3s down on the AI car just in front of me.

Note that the regular AI is less consistent than Sophy and can make mistakes, I did not seek to smooth out these errors by reproducing the experiment until I had an optimal result.
The first car in front of the player tend to be the slowest one, and here the Porsche starts last everytime. If the player is first, then the first car in front of him is... the last one too, so stays the slowest.
Also, take into account that even if I am driving a Toyota GT-One, I can set 4'11'xxx / 4'12'xxx laps with the same cars as the AI, so that you have a human point of comparison.

Here are the data :

View attachment 1439275

You can draw your own conclusions, but Sophy has without any doubt the same virus as the base AI. It's even far worse when boost is set to Off.

Sophy is overall faster than PD's AI when boost is set on Weak or High (about 5s), but the gap between its slowest pace and fastest pace stays more or less the same as for the regular AI, somewhere between 17s to 19s which is still huge and atrocious...

Now let say I'm racing with the same car with boost weak. As I said above, my pace is around 4'11/12 :
PD's AI will be 13 seconds faster than me when I'm in front, and 6 seconds slower when I'm behind.
Sophy will be 16 seconds faster than me when I'm leading, and 3 seconds slower when I'm behind.

Considering both are plagued with RB, I still prefer Polyphony's AI over Sophy, it sounds more fair to me (especially for long races). Even more so when PD's AI sometimes makes mistakes that bring more fun.

I don't think I can continue to invest my time in this game anymore. At this point it sounds pretty clear that they are stuck with this vision and nothing will change. They can bring in all the Nissan R390 GT1 or Panoz Esperante they want, if I can't have a proper race there is no point. Sad, this update had given me some hopes...

I would like to suggest that, it is possible the pace modifier is also dependant on the distance to the leader.

I did some runs at Dragon Trail earlier and got these results for Sophy:

With at least 5 laps out of 10 completed on both settings I saw:

Boost Weak fastest lap: 1:42.741 (I was 2nd at the time, AI was 7th)

Boost Strong fastest lap: 1:42.617 (I was 2nd, AI was 3rd, and AI had slipstream (set to Weak)

I also saw 1:42.814 after 2 laps on Weak in a 3rd run, I was 13th, AI was 2nd)

---

Slipstream probably brought things together but Sophy may shine more in group car racing where the packs are tight.

The difference in braking points and racecraft is so much better than the regular AI that it's hard to go back to even if the overall pace is not hugely different, at least at Le Mans.

I would say also that Le Mans is a track where the regular AI's issues may not shine through as much because of the long straights and relatively few corner complexes that leans into the AI's weaknesses.

I don't think Sophy is a write off but perhaps situationally the regular AI could be more beneficial overall given the higher instance of mistakes.

 

[A_Higher_Place]

I haven't seen the option to mandate the number of pit stops yet. Am I blind or is it only available at certain tracks?

 

[Theufcveteran]

I haven't seen the option to mandate the number of pit stops yet. Am I blind or is it only available at certain tracks?

It's not available at all, the trailer was a bit misleading. There are new custom race settings in that - Sophy now has access to them, but they're the same settings which were available in the regular AI custom races.

 

[pritchyTim]

So I did a long testing session this afternoon at Le Mans, as I planned to do yesterday. 12 short races of 3 laps each to see if anything had changed regarding boost / rubberbanding.

I raced against few GT replica cars with my TS020 to be able to trigger AI's behavior without too much trouble.

View attachment 1439274

The race setup is pretty simple, everything is set on realistic with tire / fuel on (1), mechanical damages off, professional difficulty, no penalties or flag rule, weather off.

On the table below, when I said I'm first, I'm in front of the pack and push as fast as I can (which is really easy with such a car). But when I'm last, consider I'm trying to stay in a range of 2s/3s down on the AI car just in front of me.

Note that the regular AI is less consistent than Sophy and can make mistakes, I did not seek to smooth out these errors by reproducing the experiment until I had an optimal result.
The first car in front of the player tend to be the slowest one, and here the Porsche starts last everytime. If the player is first, then the first car in front of him is... the last one too, so stays the slowest.
Also, take into account that even if I am driving a Toyota GT-One, I can set 4'11'xxx / 4'12'xxx laps with the same cars as the AI, so that you have a human point of comparison.

Here are the data :

View attachment 1439275

You can draw your own conclusions, but Sophy has without any doubt the same virus as the base AI. It's even far worse when boost is set to Off.

Sophy is overall faster than PD's AI when boost is set on Weak or High (about 5s), but the gap between its slowest pace and fastest pace stays more or less the same as for the regular AI, somewhere between 17s to 19s which is still huge and atrocious...

Now let say I'm racing with the same car with boost weak. As I said above, my pace is around 4'11/12 :
PD's AI will be 13 seconds faster than me when I'm in front, and 6 seconds slower when I'm behind.
Sophy will be 16 seconds faster than me when I'm leading, and 3 seconds slower when I'm behind.

Considering both are plagued with RB, I still prefer Polyphony's AI over Sophy, it sounds more fair to me (especially for long races). Even more so when PD's AI sometimes makes mistakes that bring more fun.

I don't think I can continue to invest my time in this game anymore. At this point it sounds pretty clear that they are stuck with this vision and nothing will change. They can bring in all the Nissan R390 GT1 or Panoz Esperante they want, if I can't have a proper race there is no point. Sad, this update had given me some hopes...

This is great analysis and testing but it feels like it is chasing a different hypothesis than what works for most of us. How fast DOES the AI drive when you drive the same car? It’s probably closer to your 4’11/12. I am grateful for what you’ve put together - you’ve proven there is indeed a rubberband beneath both versions of AI - but if it doesn’t disrupt good racing, I don’t know how much I mind.

Now, I’m a bit locked into old-AI habits - I avoid pit stops and long races lest I expose the rubber band too much. Can we loosen those restrictions because of Sophy? Maybe not. Still - when I take damage as a faster-class lapper dings me on Turn1 halfway through a two-hour race, I am relieved to not have to reset or drive an hour by myself in last place because I took an extra pit stop. I want racing action in my video game, and the RB helps with that, in those situations. And in the others - when I drive a similar-spec car at the same pace as AI - does it really matter there is trickery behind the scenes?

Ok, blind optimism over - shoot, it’d be nice if they gave us an actual ‘off’ switch.

I guess I hope you stick around and find some fun despite the flaw. Embrace the matrix! Your analysis has been so helpful over the years - thanks again.

 

[RCKakashi14]

Did a test with the Custom Race Sophy.

It was one hell of a race. And the gap between cars is much tighter than with the regular A.I.

However, not all cars are available to be driven by Sophy, so choices are limited.

 

[MisterWaffles]

Am I the only one who doesn’t care about weird rubber banding or pace differences when using the different settings?

All I know is setting the boost to weak and doing a 20-car Gr.3 custom race at Spa with BoP and 10 laps, that was the best darn race I’ve ever had against the AI on GT7 so far.

The moment to moment racing is far and away so much better than the regular AI I really couldn’t care about any weird cheats Sophy is still using compared to them.

Also, when the regular AI makes mistakes, it always seems to be them careening off into a wall or gravel trap and forgetting to brake because I “pressured” them or some baloney, they’ve given me races before like that and it’s so dumb and I can’t help but laugh. When there’s incidents with Sophy, it’s because two cars were actually battling and one got taken out, which is way more exciting and realistic.

 

[05XR8]

Am I the only one who doesn’t care about weird rubber banding or pace differences when using the different settings?

All I know is setting the boost to weak and doing a 20-car Gr.3 custom race at Spa with BoP and 10 laps, that was the best darn race I’ve ever had against the AI on GT7 so far.

The moment to moment racing is far and away so much better than the regular AI I really couldn’t care about any weird cheats Sophy is still using compared to them.

Also, when the regular AI makes mistakes, it always seems to be them careening off into a wall or gravel trap and forgetting to brake because I “pressured” them of some baloney, they’ve given me races before like that and it’s so dumb and I can’t help but laugh. When there’s incidents with Sophy, it’s because two cars were actually battling and one got taken out, which is way more exciting and realistic.

You’re not the only one.

 

[Tired Tyres]

Am I the only one who doesn’t care about weird rubber banding or pace differences when using the different settings?

All I know is setting the boost to weak and doing a 20-car Gr.3 custom race at Spa with BoP and 10 laps, that was the best darn race I’ve ever had against the AI on GT7 so far.

The moment to moment racing is far and away so much better than the regular AI I really couldn’t care about any weird cheats Sophy is still using compared to them.

Also, when the regular AI makes mistakes, it always seems to be them careening off into a wall or gravel trap and forgetting to brake because I “pressured” them or some baloney, they’ve given me races before like that and it’s so dumb and I can’t help but laugh. When there’s incidents with Sophy, it’s because two cars were actually battling and one got taken out, which is way more exciting and realistic.

I'm a lowly D driver and Boost off which is really on is giving me some good races.

 

[SPOOLMAN]

Guys. If you haven't done so already. Try the custom sophi race on the nurburgring. Absolutely epic races!

This is legit the sort of AI leap I want to see moving forward

 

[CodeRedR51]

Just did my first race with Sophy and went all out. 60 minute race at Interlagos with Gr4 cars and Boost on Weak. Started myself P16 out of 20 cars, and no matter how hard I pushed I could not get any higher than 11th before the pit stops. Me and a Ferrari went back and forth a few times over the course of a couple laps and it put a smile on my face. Managed to pass P9 on the last lap to secure the most hard fought 9th place I have ever had against GT AI. The leader was over 1 minute ahead...

And then I realized that I forgot to turn BOP on. (my car was manually set to BOP)

So I did a 20 minute race at Spa with all other settings being the same, and BOP On. Thankfully they actually take Eau Rouge with some actual speed now but their line is a bit weird so it still slows them down a little. Same with turns 9 and 18. Took the win with 2 laps to go and then the AI did the typical GT rubber band thing and they stayed right behind me and kept beating my fast laps.

Definitely much better overall. I think with my Gr4 races I might keep BOP off and then set the power/weight manually but make some tweaks to the BOP numbers to make it a bit harder. Oh, and Sophy doesn't hesitate for even one second. If you leave the door open it will fill it.

Edit: been setting up cars for a Mustang Challenge recently so did a 5 lap race at Road Atlanta. Started P5. Racing for me was great, not easy to pass. But since all of the cars are the same specs, there wasn't much passing. So I'll try adding some ballast and see how that goes.

 

[Skygrasper550]

Tried the updated Custom Race with Sophy AI just now, the tire settings is still broken. Any road cars that come stock with Comfort tires end up reverting back to Comfort tires even if you use Sports tires in your tune/setup. This is ridiculous. This problem had existed for 3 years now since launch and they still have not fixed it. Welp, back to my other games, then.

 

[MroczMR]

. When there’s incidents with Sophy, it’s because two cars were actually battling and one got taken out, which is way more exciting and realistic.

exactly



I like the shocked reaction of one sophy after mclaren crashed. It felt like he pooped his pants xD

 

[kjb]

I had a pretty good quick race in custom race with Sophy. I did a a 2 lap race with my 700 pp LeMans cars. My car was a little higher so I was able to finish 3rd. It was good racing and for the most part I used my higher top speed to get past most of the AI. I didn't have any issues getting hit in the back. I plan on doing some longer races now. I think I want to keep using the 700pp cars

 

[Bendido85]

thanks for all those first impressions, guys.

A full car list, supported by Sophy, would be a great asset.

 

[A_Higher_Place]

thanks for all those first impressions, guys.

A full car list, supported by Sophy, would be a great asset.

I'm curious which cars aren't? I saw someone post that electric cars and cars since January cannot be used. What others are there?

 

[Alfa Romeo GT]

I had the opportunity to race against the same car as me and the Boost: weak and it clearly showed me that it's buggy and that there's less of rubber banding than when the boost is off.

About the speed in boost : weak



We came out of the bend at about the same speed and it was the last lap.



For the next video (same race) I fought from start to finish against the same model (alfa romeo Giulia) as me and it was absolutely thrilling, proving (maybe ?) that AI doesn't cheat (at least in this race).


In boost off, even in difficult conditions, she gives up very quickly when overtaken.

In boost : weak she never let me beath if it has similar performance to my car.



Sorry for the ugly capture, I don't know much about the PS5 and its clips.

Oh and it was in hard difficulty with weak boost. (both video)

 

[LeanTec]

Having a blast with Sophy AI on weak boost, hard difficulty.
Ran a few races with "GT500" tunes spec'd to PP700, 500HP, 1200kg...


Lap of Nord. 24h, starting 20th and finished 14th
It's awesome having a little bit of dynamism in the AI opponents and seemingly less power fudging with boost weak.

Noticed a few patterns of divebombing and pile-ups on chicanes like the last corner of Suzuka and the Nurb. sprint first corner.

 

[suomi1]

I would like to suggest that, it is possible the pace modifier is also dependant on the distance to the leader.

[...]

Slipstream probably brought things together but Sophy may shine more in group car racing where the packs are tight.

The difference in braking points and racecraft is so much better than the regular AI that it's hard to go back to even if the overall pace is not hugely different, at least at Le Mans.

Yes, position / distance interfers with how rubberbanding works. I think that's because the game is set around the podium places in cafe mode, so most of the time you'll see AI trying to create a pack of 3 cars ahead of the field (the third AI will fall off if you're into the top 3), then another group of 5 cars which are the ones you can see on the HUD. The others are slower and slower the more you go down in the rankings, regardless of their capacities.

Slipstream may help but not that much, RB can whatsoever make AIs to ignore the theoretical top speed of their cars, by a lot.

Agreed that for a same lap time set by both AIs, Sophy is more eye candy to see.

1/ This is great analysis and testing but it feels like it is chasing a different hypothesis than what works for most of us. How fast DOES the AI drive when you drive the same car? It’s probably closer to your 4’11/12. I am grateful for what you’ve put together - you’ve proven there is indeed a rubberband beneath both versions of AI - but if it doesn’t disrupt good racing, I don’t know how much I mind.

2/ Now, I’m a bit locked into old-AI habits - I avoid pit stops and long races lest I expose the rubber band too much. Can we loosen those restrictions because of Sophy? Maybe not. Still - when I take damage as a faster-class lapper dings me on Turn1 halfway through a two-hour race, I am relieved to not have to reset or drive an hour by myself in last place because I took an extra pit stop. I want racing action in my video game, and the RB helps with that, in those situations. And in the others - when I drive a similar-spec car at the same pace as AI - does it really matter there is trickery behind the scenes?

3/Ok, blind optimism over - shoot, it’d be nice if they gave us an actual ‘off’ switch.

4/I guess I hope you stick around and find some fun despite the flaw. Embrace the matrix! Your analysis has been so helpful over the years - thanks again.

1/ Yes, AI does set laps around 4'11/12 when I'm using the same cars, so does it set 4'13/14 if I slow down on purpose, or does 3'55 (if let say) I cut the corners to cheat. It actually doesn't have its own pace. Now I wonder if AI is just able to race alone, since it only drives depending on what you do and where you are. May be this is why B-Spec still isn't here.
This is not good racing at all, a good entertainment at best but certainly not a race. A driver targets victory, AIs doesn't; its target is to stay close to you even if it means cheating or pretending to be worse than it really is.

2/ See? To enjoy what you call good racing, you find yourself avoiding things that define what a race is, which is sad. And yes, trickery matters, at least to me since I love realistic longer races with tire and fuel management, mechanical damages, pit stop strategy, multiclass and messing with cars from other classes, etc..
We have realistic physics, realistic environments, realistic sounds, very realistic cars, now a realistic AI, but a fake race theater.

3/ It would be the best of all possible worlds for every player indeed. I don't want rubberbanding to be totally disabled for everyone because I know some people still love it. An option to switch it off if I want, hell yeah!

4/ Your welcome.

 

[Gingerale]

this and a showcase feature where you can share your events and get events from others. It should also be possible to create series and not only single races. Like, in my example, I could share the Porsche cup with the schedule, rules, points system and even a set of liveries.

that, plus this AI that is capable of engaging racing, would be the endgame for me.

I would never need a new racing game when this happens lol

 

[E_Spori]

I tuned some compact cars for imagined 500PP Cup race. In car settings I use SH tires but in custom event with Sophy they are all reverted back to CS tires. Civic EG is set to CM tires which are stock ones I guess. I know it was asked already but I didn't see an answer. Is there a way to avoid this?

 

[Theufcveteran]

I tuned some compact cars for imagined 500PP Cup race. In car settings I use SH tires but in custom event with Sophy they are all reverted back to CS tires. Civic EG is set to CM tires which are stock ones I guess. I know it was asked already but I didn't see an answer. Is there a way to avoid this?

No, AI cars will either use the tyres equipped from the factory, or Racing: Hard if equipped by you.

The latter only applies if the factory tyres are Sports: Soft or below.

Racing: Medium/Soft factory tyres would stay that way for the AI even after giving them a car with Racing: Hard equipped.

If the AI car is a racing car and comes with Racing tyres, when pitting from a wet tyre to a dry tyre, they will always change to Racing: Soft.

I'm not sure if this applies to road cars that have Racing: Hard tyres equipped - my gut says they would switch back to road tyres for wet conditions, but I don't know for sure. I need to test this.

 

[KAlex122]

No, AI cars will either use the tyres equipped from the factory, or Racing: Hard if equipped by you.

The latter only applies if the factory tyres are Sports: Soft or below.

Racing: Medium/Soft factory tyres would stay that way for the AI even after giving them a car with Racing: Hard equipped.

Now i get it. Thanks! Posting saved.

 

[Theufcveteran]

Now i get it. Thanks! Posting saved.

No worries friend. I have also added to the post just now about wet weather.

 

[Meltac]

No, AI cars will either use the tyres equipped from the factory, or Racing: Hard if equipped by you.

The latter only applies if the factory tyres are Sports: Soft or below.

Racing: Medium/Soft factory tyres would stay that way for the AI even after giving them a car with Racing: Hard equipped.

If the AI car is a racing car and comes with Racing tyres, when pitting from a wet tyre to a dry tyre, they will always change to Racing: Soft.

I'm not sure if this applies to road cars that have Racing: Hard tyres equipped - my gut says they would switch back to road tyres for wet conditions, but I don't know for sure. I need to test this.

Strangely though, AI cars won't have rain tires equipped when the custom races starts with heavy rain even if you previously equipped them with rain tires, which doesn't make much sense IMHO.

That also might be the reason why PD removed the option to set the surface wetness at the beginning of the race, sadly.

 

[Mister_XFR]

I’m getting crushed by my own cars… 😅

 

[Theufcveteran]

Strangely though, AI cars won't have rain tires equipped when the custom races starts with heavy rain even if you previously equipped them with rain tires, which doesn't make much sense IMHO.

That also might be the reason why PD removed the option to set the surface wetness at the beginning of the race, sadly.

I believe racing cars will, but do you mean road cars? If so, thanks for the information - not surprising I suppose given the logic of how AI tyres currently work. Really seems like anything other than vehicle default and Racing: Hard doesn't apply for road cars.