Ask the devs : AI/Ratings

[stepsix]

Re: Ask the devs : AI/Ratings

I'd like to know if the CPU will react to my defence appropriately this year. In Live 10 the CPU just did its own thing regardless of how I played D. I love FIFA 10 because the A.I. reacts to how I play then counters my moves. I didn't feel that in Live 10.

Depends on what level you're referring to. Do you mean ball carrier->defender, or team strategy?

[Majingir]

Re: Ask the devs : AI/Ratings

Will we see games fall apart near the end if the user is winning?

Many video games have that "AI come from behind wins" type of stuff within the final parts of the game where the AI catches up from out of nowhere.

[stepsix]

Re: Ask the devs : AI/Ratings

Just a few questions.

1. I have been hearing John wall will be a overall 81 in ratings? Please tell me that isn't true! Don't throw off the balance of the game...if he gets a rating of 81 I want Derek Fisher at 87
Ratings are controlled by our database guy...he's doing a first pass at tuning ratings right now given all the new implications of ratings this year.

2. Last year in Live 10 at the last 5 minutes of the 4th quarter, or when the AI was down by 20 it seemed like their team turned into a team full of Lebrons and Kobe's and they never missed a shot, while players could make nothing. Now I do understand this happens sometimes, but when it happens EVERYTIME it gets irritating. And I feel like...well like this..
I have a feeling this will be the first of many catch up logic questions I answer so I'll be brief. There is no ratings change for the cpu at any point in the game. If the cpu is down late in the game, they will run more offense through their superstars and increase their tempo which will result in more shots early in the shot clock. In addition they'll burn timeouts which decrease their fatigue levels. This year how well you stay on your man will have more of an impact on shooting %s, so if you keep your cool late in the game you should be able to maintain a lead better. I'll say it again for emphasis tho - there is absolutely no rubber band logic.
3. Will the AI be smart? PLEASE NO MORE CENTERS RUNNING TO THE 3 POINT LINE!

First question is too vague, but as for trees on the perimeter, I can't guarantee they won't be out there. What would you expect if Kobe is standing at the top of the key in an iso? I'd think you want everyone to clear out so you can go one on one...

[stepsix]

Re: Ask the devs : AI/Ratings

Question

Will there be "tendenices" as well as ratings in elite11... IMO dynamic DNA just doesn't seem to be working

Can you give me some examples of tendencies that you would like to see and how they aren't realized by DNA?

[gmix411]

Re: Ask the devs : AI/Ratings

Can you give me some examples of tendencies that you would like to see and how they aren't realized by DNA?

Can I give one? Billups pulling up for a three during a fastbreak.

[stepsix]

Re: Ask the devs : AI/Ratings

<hr style="color: rgb(57, 57, 57);" size="1"> <!-- / icon and title --> <!-- message --> What I want to know is will you guys try to keep the rating differential in check. I feel that after you account for team rating, momentum and player ratings some players were just gameplay breaking out of control. It would affect teams too strongly that things were happening on the court that would never happen in a real game. Some players became enormously inept while other players seem to go through an entire team singlehandedly without any resistance even if you see it coming and KNOW it's coming. In real life if Lebron James decides to just run straight at the basket and the team knew it, I mean KNEW it for a fact he would never score.

Basically I'm asking if you guys will keep the ratings within a certain range behind the scenes and if you must cater to the "fun" crowd at least have a lobby where the ratings could be marginalized to keep the gameplay tight and realistic. Where you can go off with a great player but you have to work for it still. Run some plays, burn an opponent in one fashion and then switch it up with the hot player in another fashion to score. I'm not asking for the players not to be able to do anything but I'm asking don't have the game help the souped up rated player out when the other guy has him figured out.

I'm here for you RandomBrother! I saw your other posts and apologize for not getting to them - rEAnimator & I are reading all the time and replying when we can...we were both online discussing alley oops at 2:30 this morning, so rest assured if we're not posting on OS we're probably improving the game.

ImInUrBase is the man for online features and he'll be talking about what we're doing this year in the online realm.

On to the ratings. I think there is a bit of a misconception emerging based on the hype that's been released up until this point. I'm referring to the idea that because there is more user control, the game is more of an arcade experience. rEAnimator made a solid post in the physics thread about the different types of sim experiences and what we're trying to achieve with Elite this year.

Physical attributes will be much more of a factor than they were in the past. The main focus of the physics changes this year have been in how players get around the court, and to be a true physical representation of this, we had to factor in height, weight speed and all the other things that affect how guys move and interact.

For more abstract ratings like 'offensive post footwork', in previous iterations, when we were playing out a two man scenario, this could easily be factored in when you had two canned animations with fixed outcomes. For example, when two guys were posting up on ball and a second defender came in to trap, it would enter a three man trap scenario. If the ball carrier didn't pass out in time, the ball would pop out. The timer would be controlled by some ratings. Thus, if the ratings related to how when the ball would pop out in this scenario, it would have a big impact on the outcomes, at times seeming imbalanced.

This year when the same defender comes over to trap, he is no longer tied to the ball carrier and the primary defender other than by proximity and the knowledge that he's double teaming. His hands will come up, but you can still contest, steal, and move around freely with the left stick. As such, his ratings for blocks, steals, weight & height will be what is deciding the outcome of the trap, not an abstract rating that had to be invented to match some artifical scenario that we created.

The end result is less confusing ratings that you're not sure what they accomplish, so you can focus more on the concrete ratings that make sense.

I'm not knocking previous iterations of Live (as you may have read I've worked on live for quite a while), but it's the new physics engine that allows us this new level of control.

[kumamae_33]

Re: Ask the devs : AI/Ratings

These are off topic questions but since there are no threads regarding these questions below, I think I'm going to ask it here... & maybe get some answers from you devs here.

1. Is it now possible to choose in Elite 11 what kind of timeouts we want? i.e. 20 sec or full TO? This is not doable in Live 10 since its only random in Live 10.

2. How about the coach strategy menu by pressing the Dpad? Did you guys keep it in Elite 11? If yes is it more friendly user now compared to Live 10? Coz for me it's not friendly user in Live 10. And it's annoying every time the EA logo cutscene appears coz the coach strategy menu exits every time the EA logo cutscene appears. Did you guys add more options to choose in the coach strategy menu? ie. "deny the ball D" in the defense strategy & etc.

3. About the user profile on Elite 11. Can we now save the game even in the halftime break or at the end of the quarter and load it from where we left it? Also will the user profile allow us to save and auto load our playbook online? By the way we cannot change the playbook online in Live 10. The only way to change/edit it is in offline.

4. How about the fatigue? Is it now effective in Elite 11? And is it effective even in shorter periods like 5 mins per quarter online?

5. Did you guys completely remove the turbo button in the game? Or is it only on the defense control?

6. About the players position and starting lineup intro. Will Elite 11 going to use the old format or the new format? Coz the NBA no longer uses the old format.

Old format
C - 5
PF - 4
SF - 3
SG - 2
PG - 1

New Format
C - 5
F - 4
F - 3
G - 2
G - 1

[randombrother]

Re: Ask the devs : AI/Ratings

I'm here for you RandomBrother! I saw your other posts and apologize for not getting to them - rEAnimator & I are reading all the time and replying when we can...we were both online discussing alley oops at 2:30 this morning, so rest assured if we're not posting on OS we're probably improving the game.

ImInUrBase is the man for online features and he'll be talking about what we're doing this year in the online realm.

On to the ratings. I think there is a bit of a misconception emerging based on the hype that's been released up until this point. I'm referring to the idea that because there is more user control, the game is more of an arcade experience. rEAnimator made a solid post in the physics thread about the different types of sim experiences and what we're trying to achieve with Elite this year.

Physical attributes will be much more of a factor than they were in the past. The main focus of the physics changes this year have been in how players get around the court, and to be a true physical representation of this, we had to factor in height, weight speed and all the other things that affect how guys move and interact.

For more abstract ratings like 'offensive post footwork', in previous iterations, when we were playing out a two man scenario, this could easily be factored in when you had two canned animations with fixed outcomes. For example, when two guys were posting up on ball and a second defender came in to trap, it would enter a three man trap scenario. If the ball carrier didn't pass out in time, the ball would pop out. The timer would be controlled by some ratings. Thus, if the ratings related to how when the ball would pop out in this scenario, it would have a big impact on the outcomes, at times seeming imbalanced.

This year when the same defender comes over to trap, he is no longer tied to the ball carrier and the primary defender other than by proximity and the knowledge that he's double teaming. His hands will come up, but you can still contest, steal, and move around freely with the left stick. As such, his ratings for blocks, steals, weight & height will be what is deciding the outcome of the trap, not an abstract rating that had to be invented to match some artifical scenario that we created.

The end result is less confusing ratings that you're not sure what they accomplish, so you can focus more on the concrete ratings that make sense.

I'm not knocking previous iterations of Live (as you may have read I've worked on live for quite a while), but it's the new physics engine that allows us this new level of control.

All good stuff there. Really like what I'm hearing and covered alot of what I was asking. My only last worry is what I was saying by the rating differential.

[jfsolo]

Re: Ask the devs : AI/Ratings

ADHD version- Will the A.I. "know" its abilities and skills and the abilities and skills of its opponents/teammates?


Every basketball game has had this same weakness-The A.I. has exhibited a very low level of basketball "knowledge"

For those of us who are Single player dynasty mode players, our enjoyment of the game is going to be largely dependent on how well the C.P.U. opponents and teammates do their job.

The A.I. in almost every basketball game hasn't been programmed to "know" its strengths and weaknesses. Its been bad on offence but worse on defense.

If Derek Fisher gets caught down low with Dwight Howard attacking the rim, he basically has 3 plays. Try to draw the charge-Flop, go for a strip down low if he's in position, or get out of the way.

In the past the A.I. would have him jump for a block most of the time. He wouldn't get the block, but the point is that the A.I. isn't cognizant of the fact that that jumping to try and block the shot is a bonehead play.

Its working off some rudimentary code of how basketball is played, unable to provide the illusion that every player in the game knows its own abilities and the abilities of all the other players as well.

With canned outcomes being gone, it more important then ever for the A.I. to be able to make good decisions because it seems that dice rolls, warping and morphing isn't going to be able pick up the slack for sub par basketball I.Q.

[Superstarshooter32]

Re: Ask the devs : AI/Ratings

If a big man gets switched off onto a quick perimeter player, will they sag off automatically? In addition to that, if the answer is yes, what if it's a big man who can actually play perimeter defense (they're very rare even in the NBA-- I know Kenyon Martin can, though)?

Say K-Mart gets switched off and is guarding a perimeter player, but not the quickest guy around, someone like... Ron Artest. Will Martin know that even though Artest is a perimeter player and K-Mart is a big man, he doesn't have to sag off?

EDIT: Also, jfsolo made a very good point. Even big men should let players dunk if they feel they (as the defender) rotated too late to have a good chance to affect the shot without fouling. It happens a lot more IRL than I think people realize.

Also, will the AI be aware of foul trouble? i.e., if an AI-controlled defender is in the game with 2 fouls in the 1st quarter, will he be less aggressive defensively?

I know this is advanced stuff, so if it's not in, I'm sure it won't be a deal-breaker.

[carnalnirvana]

Re: Ask the devs : AI/Ratings

while i agree with most of PM lenghty post, in IMO live 10 has the potential to play WAYYYYYYY better so i know elite can get it somewhere decent.......if there ratings were actually representative of players role in the offense.....

first the ratings:

lets use live 10: kobe has a mid range likelyhood of 99 , fisher is at 84, gasol is about 85 really and we wonder why the cpu cant run any type of offense.....kobe should be 3 times more likely to pull up compared to anyone on that team IMO, i am not saying fish wont take an open J, but with an 84 he is jackin anything..... kobe should be around 65 and fish around 20 at most that way there is some type of priority in the offense

this is live/elite biggest problem IMO all players share the same ratings kobe has a 85 off ball awareness and the members of his team are 10 points off, how can he really be that much better when he isn't rated that far off.......

[Playmakers]

Re: Ask the devs : AI/Ratings

while i agree with most of PM lenghty post, in IMO live 10 has the potential to play WAYYYYYYY better so i know elite can get it somewhere decent.......if there ratings were actually representative of players role in the offense.....

first the ratings:

lets use live 10: kobe has a mid range likelyhood of 99 , fisher is at 84, gasol is about 85 really and we wonder why the cpu cant run any type of offense.....kobe should be 3 times more likely to pull up compared to anyone on that team IMO, i am not saying fish wont take an open J, but with an 84 he is jackin anything..... kobe should be around 65 and fish around 20 at most that way there is some type of priority in the offense

this is live/elite biggest problem IMO all players share the same ratings kobe has a 85 off ball awareness and the members of his team are 10 points off, how can he really be that much better when he isn't rated that far off.......

I definetly agree that the ratings in years past have caused some of the issues on the court.....

I used to be really upbeat and patient when it came to editing ratings in Live. But after Live 10 i just lost the passion to edit the game last year.

I'm sure had i spent about a month tweaking ratings i could have seen more great things out of live 10.....but at some point i'd just like to buy the game fire it up and have the CPU players resemble their true tendencies and abilities out of the box without me overhauling the game.

Hopefully NBA Elite will once and for all solve that problem or atleast minimize the problem from what it was in years past.

[jfsolo]

Re: Ask the devs : AI/Ratings

I definetly agree that the ratings in years past have caused some of the issues on the court.....

I used to be really upbeat and patient when it came to editing ratings in Live. But after Live 10 i just lost the passion to edit the game last year.

I'm sure had i spent about a month tweaking ratings i could have seen more great things out of live 10.....but at some point i'd just like to buy the game fire it up and have the CPU players resemble their true tendencies and abilities out of the box without me overhauling the game.

Hopefully NBA Elite will once and for all solve that problem or atleast minimize the problem from what it was in years past.

As long as developers have to concern themselves with the Overall rating of certain players being at a certain number, ratings will never be strong out of the box.

[Jazza619]

Re: Ask the devs : AI/Ratings

Going off what Playmakers said about editing existing players, will it be possible to change hairstyles?

[Playmakers]

Re: Ask the devs : AI/Ratings

Going off what Playmakers said about editing existing players, will it be possible to change hairstyles?

I wish we could get an answer on the ability to create more than 50 players or edit over an existing player.....

But i sense the CAP situation will remain the same as always very limited.

It's kinda sad to for me becuase I've been wanting to creating a monster roster file with only Old School players.....that would actually be like a dream come true if they gave us about 300 CAP slots

I probably wouldn't even play a single game with the default rosters ever again......

It would be nothing but Jordan, Bird, Barkley and Magic all night long