Misc True to Life AI 2013 Season 1.4

[snthennumbers]

OK, so we will wait for it!

Updated database will come first in v1.1 to reflect the most recent performance of the drivers and teams in real life after the Malaysian grand prix. I'm going to finalize testing on this within the next day and will post v1.1 asap. (Perez being so fast is annoying isn't it?)


v1.2 will include any AI improvements I've managed to do since posting v1.0. I'm still testing a few. Check the full description of this mod to see the parameters that still need testing and confirmation...


Everyone is welcome to test and report back what they find

 

[Deleted member 137812]

What time does the AI normally clock on Melbourne in Q1? mine are spread from 1:25 to 1:30/31 is that normal?

 

[manos426f]

@ snthennumbers

Unfortunately I cannot download your mod, for some reason it does not start....

 

[snthennumbers]

Sounds about

What time does the AI normally clock on Melbourne in Q1? mine are spread from 1:25 to 1:30/31 is that normal?

Sounds about right to me...Melbourne is one of the tracks I just can't keep up with the AI on, so I had to adjust my track_difficulty for it from 1 to 0.98 :\

But 1:25/1:26 sounds about right...I can't seem to manage anything faster than a high 1:26 there personally

 

[snthennumbers]

@ snthennumbers

Unfortunately I cannot download your mod, for some reason it does not start....

Some other people are experiencing this too...seems there's issues based on what country you're in. I'll PM you with another link

 

[Deleted member 137812]

Finally got the fix, be sure you dont make the same mistake as I did. Here is my post in the questions thread.

"Funny thing guys, make sure you even up the tiers in the database, earlier I said making all the teams tier 3 and 4 causes them to qualify on the same tyre. My new finding is this: Every car in tier 3 and 4 go out before the cars from tier 1 and 2. So that means when you put all the teams in the same tier they just ran into eachother and post horrible laptimes."

 

[snthennumbers]

Finally got the fix, be sure you dont make the same mistake as I did. Here is my post in the questions thread.

"Funny thing guys, make sure you even up the tiers in the database, earlier I said making all the teams tier 3 and 4 causes them to qualify on the same tyre. My new finding is this: Every car in tier 3 and 4 go out before the cars from tier 1 and 2. So that means when you put all the teams in the same tier they just ran into eachother and post horrible laptimes."

Well isn't that interesting...
In my current database (which will be released with v1.1 of this mod within the next day) I actually changed some teams tier levels in the "f1teams" table.

For instance I made Mercedes and Lotus both tier 1 in order to make them as competitive as we're seeing in real life. I haven't noticed anything negative as a result from this change.
During testing though, at one point I did set all teams to tier 1. This resulted in Caterham, Williams, etc qualifying within a couple tenths of a second from the #1 spot. At times the backmarker teams would out perform Ferrari or Red Bull. But I didn't see them running into each other or posting stupid slow laptimes. AI contacting each other would be handled in the AI files anyway...

But so far I've been running with modified tier teams (Mercedes & Lotus = tier 1 instead of 2, Force India = tier 2 instead of 3) as well as modified horsepower values for each team in the "ai_engine_power_per_team table" and I haven't noticed any issues...

 

[snthennumbers]

I ran some tests with the v1.1 database and here's the results (call this a sneak preview?):


Results
China quick qualifying ran at full speed (8 simulations):
Vettel:5, 1, 1, 2, 3, 2, 3, 2
Webber:1, 3, 3, 6, 2, 5, 6, 4
Alonso:2, 2, 2, 5, 6, 4, 5, 6
Massa:7, 6, 4, 4, 7, 8, 4, 7
Hamilton: 4, 5, 7, 3, 5, 3, 2, 1
Rosberg: 3, 4, 5, 7, 1, 1, 7, 5
Raikkonen:6, 7, 6, 1, 4, 6, 1, 3
Grosjean: 8, 9, 8, 8, 9, 7, 8, 9
Button:9, 10, 11, 9, 8, 11, 10, 13
Perez:13, 8, 10, 13, 10, 9, 9, 12
Gutierrez:15, 16, 16, 15, 15, 15, 16
Vergne:14, 14, 12, 14, 14, 13, 12, 14
Sutil: 10, 11, 9, 11, 11, 12, 11, 8


Results
Brazil quick qualifying at full speed (4 simulations):
Vettel:6, 1, 4, 3
Webber:4, 2, 6, 6
Alonso:1, 4, 1, 5
Massa:7, 6, 5, 2
Hamilton: 2, 5, 2, 4
Rosberg: 3, 7, 3, 7
Raikkonen:5, 3, 7, 1
Grosjean: 8, 8, 8, 10
Button:9, 10, 9, 8
Perez:13, 12, 10, 12
Gutierrez:16, 15, 15, 16
Vergne:14, 13, 14, 13
Sutil: 11, 9, 11, 9


I think this is solid enough evidence I have the drivers acting as they should for the most part...anyone in disagreement based on the above results?

 

[snthennumbers]

snthennumbers updated True to Life AI with a new update entry:

Database changes

v1.1 release notes:
Database update only, no AI changes.
Updated ai_driver, f1teams, and ai_engine_power_per_tier tables in the database to reflect the most recent 2013 performance stats after the Malaysian Grand Prix.
Updated damage settings (loose_parts_settings.xml) to be more realistic. Installing this file is not mandatory if you already have personal preferences for damage.

***PLEASE NOTE***
The included database.bin file should be installed...

Read the rest of this update entry...

 

[snthennumbers]

snthennumbers updated True to Life AI with a new update entry:

Mistake corrected

Sorry everyone I just realized I uploaded my personal database in v1.1 with my own custom track_difficulties levels and track_tyre_wear_level values.

Please re-download to v1.1a which has the default values for these tables.

Read the rest of this update entry...

 

[snthennumbers]

Also added description of database.bin file changes to the Overview

 

[manos426f]

First feedback, after very few race starts and attempt on restarting a new career.

Unfortunately in the career mode I suffer from the EXACT same problem I had when installing Realism mod + all the other mods I mentioned!
I cannot pass the hairpin test, since the car cannot reach the required speed on time and before hitting the brakes.
It is just as it is shown in the following video:

The difference is that in the previous case I could pass the test BEFORE I put the mods and continue afterwards.Now I cannot do that, since the new database.bin file is not compatible with the original one.
So how can I get through that huge problem that blocks me from starting a new career?

Other than that, I tried some race starts in Quick Race mode in Australia and China. I noticed that in Australia Turn 1 AI drivers are behaving really well! Same goes for China Turn 1, since now they are so damned slow as they used to be!

I also noticed that Perez scores very high qualifying positions, Alonso is very aggressive, constantly trying to gain positions (just like in reality) and Mercedes drivers are really good. I will give more feedback on these in the following days.

The question is how do I pass the hairpin test in career mode....?

 

[snthennumbers]

I just had the same problem you did actually. I did manage to pass the hairpin test (I use a gamepad), but you need to get a really good exit from the corner. You'll get to 185mph just before the red braking zone. Slam on the brakes and downshift fast to bring it to a stop and pass the test.

Otherwise you'll need to revert, and install all mods except the (edit 4.11.13) revised car performance mod included CTF files...get through the driving tests, save, exit game, install revised car performance mod, resume career at race 1.

 

[snthennumbers]

I also noticed that Perez scores very high qualifying positions, Alonso is very aggressive, constantly trying to gain positions (just like in reality) and Mercedes drivers are really good. I will give more feedback on these in the following days.

Version 1.1a database corrects this issue, Perez is qualifying between 8th-13th now.

 

[Daniel Alexandru]

I just had the same problem you did actually. I did manage to pass the hairpin test (I use a gamepad), but you need to get a really good exit from the corner. You'll get to 185mph just before the red braking zone. Slam on the brakes and downshift fast to bring it to a stop and pass the test.

Otherwise you'll need to revert, and install all mods except the revised car performance mod...get through the driving tests, save, exit game, install revised car performance mod, resume career at race 1.

This problem is caused by revised car performance.I eliminate thsi mod of my game.

 

[d4rK]

ok i fix my problem with slow ai time in q3 i've instaled mr'pibb 2013(without track) and now evrythink is correct the ai is the festes in q3( i tested only in quick race mode) but it still damm good in rain;/

I had q1 q2 dry best time was 1:30 somethink and forecast for q3 was dry. I drive out and it was heavy rain. My best time was 1:47(on setup from q1 and q2) at the and of sesion and the ai 1:40 did the ai keep the setup block rule ??

Will the future update reqiued reset carrer ??

 

[Deleted member 137812]

How did you edit the car performance? You know that changing tier performance doesnt work for the driver's pace right? only ai

 

[Kristian Nenov]

What time does the AI normally clock on Melbourne in Q1? mine are spread from 1:25 to 1:30/31 is that normal?

On my Co Op career (short weekend) I did a 1.25.8xx and it was a pole. Hamilton was just behind me with 1.25.9xx or 8high or smth and then was schumacher 5 tenths slower. So this is good. It's faster than original I guess.

 

[snthennumbers]

This problem is caused by revised car performance.I eliminate thsi mod of my game.

You don't need to eliminate the revised car performance completely if you don't want. Just don't install the revised car performance mod until you've completed the young driver test...or make sure to get a really good exit from the hairpin during the hairpin test (it is doable...you might need to turn on assists just for that test. I've completed it with the revised car performance mod. It took me 3 tries but I passed)

 

[snthennumbers]

How did you edit the car performance? You know that changing tier performance doesnt work for the driver's pace right? only ai

There's a number of things to affect car performance.
I've been using the CTF files from the revised car performance, but with increased grip levels to counteract the reduced downforce. This was to achieve realistic straight-line speeds the revised car performance mod gives, but without the imbalance the revised car performance mod brings.

Then, I changed tier levels for Force India from 3 to 2, Mercedes from 2 to 1, and Lotus from 2 to 1.
From there I edited the ai_engine_power_per_tier table to make slight adjustments to teams' horsepower levels so as to replicate real world results in 2013 like we've been seeing.

From there I edited the ai_driver table to make the drivers perform as we've been seeing in real life in the 2013 season.



I've also been making slight grip level adjustments to the Player's CTF files on tracks where I've found the Player is faster than the AI seemingly for no reason. See my post here: http://www.racedepartment.com/forum/threads/true-to-life-ai.67176/page-2#post-1407443

I was looking to get feedback from everyone to see if they had the same feeling about there being some tracks where the Player is notably faster than the AI (even with a setting of "1" in the track_difficulties table).

So far the tracks Ive noticed this on are:
Canada
Hockenheim
Abu Dhabi


Anyone else have input on this?