Simmunity pre-alpha testing + offline championship discussion

Well that's a question I've not looked in to.
My goal is to start near the back of the field and battle my way through. If I nail the race, there is a possibility for a win. This would be quite rare though, generally I'd be shooting for a top ten finish.
So the battle is where my focus is.

I'm tuning for 100% strength and am trying to eliminate those one or two corners where the AI gets confused or reduce their braking effectiveness. So far I've done 3 or 4 tracks and have it so a lap at qualifying pace behind AI does not hold me up.
This allows me to push in the race only thinking about racing, not what weird behaviour is going to jump out at me Of course this means more immersion, more enjoyment... very satisfying indeed.


I'm still new to all this but I'd say the following are grouped:
WorstAdjust=(0.8000)
MidAdjust=(1.0500)
BestAdjust=(1.2000)
AIRange=(0.1000)

so the example above is contradicting itself as the range does not allow for the full scope defined in the adjust parameters.
AIRange=(0.4000) would be needed to cover the full range.

On some tracks where the AI seemed too fast overall, I've used:
WorstAdjust=(0.8000)
MidAdjust=(0.9760)
BestAdjust=(1.0740)
AIRange=(0.3000)

But as I've said, I'm still new and learning. All results are subjective and completely non-scientific

 

Sorry for late reply. I was entertaining out of town guests. I don't see any reason why people wouldn't be willing to install an update in order to achieve improved AI performance.

At a minimum, it seems that either (1) the AI need to be spread out to give a broad range of drivers someone to battle with (i.e. not get left behind or too easily beat the AI) or (2) drivers need to be able to adjust the AI strength along with basing results on total race time.

My bet is that number 1 will be preferred by most drivers and any driver who is not willing to install an update is not really serious about wanting to compete anyway as requiring use of mods is not uncommon in league racing.

 

hey guys, please could you test something for me, I had a lot of fun in this race in the nissan GT4, having a 3 way battle for 1st until I spun out. Even then turned out to be a lot of fun fighting for 5th since I got the cars nicely spaced out with AIrange parameter. Matt for some reason it wouldn't let me do an update to the mod so i had to remod the whole thing. Could you guys have a go and report back what you think, is the AI too slow for u?

View attachment 16043

mod: https://mega.co.nz/#!ekpUDQhC!1FwkwTUKjYBeSL0zO55nyw-GkTUXwMWOiARbs4auoOM

 

Can you check the link mate, it seems to be taking me to my files.

 

doh sorry, new to the mega site, should work now. cheers

 

Default setup with 54L. Came in last place at +00:25.051. Went off track several times. Been forever since drove this car. AI fastest lap = 2:16.095. My fastest lap = 2:19.283. Hard to tell if could win with practice. Probably could. Interesting choice to start in 5th position.

BTW, Matt I got your PM and just haven't had time to give an intelligent response with back to back meetings all day and my car died today too Now have to get to bed as I will have long day tomorrow too but will try to respond asap...

 

BTW - I tested using tracks.exe in download directory and works fine if set directory correctly and that way don't have to copy/paste filtered_data.zip

 

Nice one. Don't stress about responses mate, I know you are busy and I'm just writing thoughts as they come.

Sent from my Nexus 4 using Tapatalk

 

Was the AI fairly widely spaced? On my race there was about 50 seconds between 1st and last place. With Matts help I'm trying a different version now, and it's just just a 200kb patch, not a whole mod download. Will post here when I've done some tweaking. And regarding 5th place, yeh I thought might be interesting to have a chance of 1st place but still needing to drive well to hold off the others. Overall though I still prefer qualifying for official races.

 

I wasn't happy with the AI's brake points since they were braking much deeper than a human can. Apparently the parameter AIstiffness controls their braking effectiveness, so I changed that to 0.8 and this seems to make much more realistic brake points. Matt I think I've figured out why you had a bunch of slow guys and then one ridiculously fast guy in front, and that's because since the start grid is random, the other fast guys are getting stuck behind some slow ones further back. For example HArri Hanninen's times in a random race were 2.22, 2.20, 2.18, 2.16, 2.16. Clearly he was getting held up until the end. It def wasn't real road making the difference as the winner was just a solid 2.17 the whole race.

I think therefore that qualifying is going to be an important aspect of every race to ensure the fast AI are in the front and everything is consistent for different people. If people cant be bothered to qualify, they can just start the quali session, and hit 'finish session' so it's not much pt.

In this mod update, the fastest AI are doing 2:16's and the slowest 2:24's. I personally feel this is a good range. My best time is 2:19 so I know I can't win, but perhaps some aliens are going to be able to come up with 2.16's and compete for the lead. Do you guys feels the same or do you think the average Joe should be given a chance to win by having the fastest AI in the 2.19 range? Only problem with this is the fast guys are all bored at the front on their own and you have a bunch of people with 1st place which will be problematic for the championship table.

Anyway enough rambling, here is the update.

View attachment 16048

 

Thanks for posting your thoughts mate, this is great info. As I've said I'm still finding my feet with all this and I've read about this problem of faster guys stuck behind elsewhere.
I'll put some attention into this but saying that, your are spot on about qualifying.
I enjoy battling through the field, seeing how high I can climb. It feels great when you get to around 10th and you have to push so hard to make more progress. Anyway, it's habit for me to do that, nothing more.
Simulating a race weekend is the way to go. People can skip if they want, as you've said so I don't see any downsides.

I'm a gamer and a dreamer. When I race against AI there is often a scenario involved, a little game made up or emulating an event.
For example, the normal driver of this car is injured and I'm replacing him for this one race. I've got to prove myself with this one opportunity.
The AI is then set as best I can, to fit with this scenario.

Is it a club race where they are undisciplined or is it a national championship where the drivers are desperate to prove themselves?

As this site matures, I can see scenarios as an introduction... an enticement to download the package.

So this is your championship, think of a scenario and start tuning. If you like the end result and can convey your ideas to others well, then those people will be sure to enjoy it, no?

 

Let me check...

 

Two other things. The only way to allow slower drivers a chance to win without faster drivers being bored is to allow drivers to race against different AI strength percentage and using total race time as race result. Having said that I don't expect to win but rather be mid-pack most of the time because I am a mid-pack racer online. Second thing regarding qualify and AI is that you might want to look at Tim's thread about his BT 1966 car set where he said it is important to let AI have at least [x] number of minutes qualify time.

 

Looks like only ~17 seconds ==> http://simresults.net/150311-OF4 <== See Race #2

 

+1

 

Not to hot, not to cold...
Last lap

Simresults

 

I've been thinking about this and have an idea but not an implementation.
Possibly it could be integrated into the app. As you load the filtered data, it could ask are you rookie, novice or pro.
Rookie being set to 90% strength, Novice 100% and pro 110% or similar.
Quite how to determine what people are capable of around a track is another thing but deal with averages and it can't go far wrong.

I guess the database will be able to help here over time by calculating submitted races.

 

Nice catch! I seriously love this track!

 

That's weird, got a feeling it wasn't using my version of the aiw.

 

I think this is a great idea. The only requirement would be enough people to race in your class to make it interesting. Of course time based results could be integrated into 1, but points tables would have have to be separated.