• Note for new users: Please keep your posts English or we will delete them. There's a "No English?" area in our Editing discussion forum for those who prefer writing in their native language.

Evoweb Gameplay Lab, FIFA 20 (Frostbite FIFAs?)

Anth James

League 1
22 September 2017
Just played 2 games in CM....I'm not sure I'm convinced by it all just yet. It's improved for sure but I can't shake the feeling there's something still at play. I could be wrong, it did feel better, and I didn't get that strange feeling towards the end of the match that something was going to happen, stats seemed quite good too, but I feel my players felt a little sluggish and made more errors than normal possibly. Definitely could just be in my head, I think I need to test some more first. I don't really know how form/morale works in CM so I wonder if that could be playing a part...I wonder if we can turn that off too!
 
Last edited:

papinho81

Premiership
22 March 2017
Chaumont Switzerland
Olympique de Marseille
You read my mind haha, I was just typing a post with this! That makes complete sense. I haven't even tried it yet, but match Intensity turned off is here -

Match Intensity Off- https://mega.nz/file/pV80HYwI#CYtctLLiSlh3BSJewe_zsku5SQOcPCZ6JMuNO9DVMSc
@Anth James by MatchIntensity table I actually meant this table from the db (the picture show demo but it is the same in ng):1596054432258.jpeg

The problem is if you use a patch such as FIP it has already its own db file so the file we might want to edit will be in conflict with the one from the patch. The only work around I can see would be to have access to the patch project files... Not anytime soon...

Too bad because there is also files like this one which could be interesting.
1596056240121.jpeg
 

Anth James

League 1
22 September 2017
@Anth James by MatchIntensity table I actually meant this table from the db (the picture show demo but it is the same in ng):View attachment 53608

The problem is if you use a patch such as FIP it has already its own db file so the file we might want to edit will be in conflict with the one from the patch. The only work around I can see would be to have access to the patch project files... Not anytime soon...

Too bad because there is also files like this one which could be interesting.
Ohh I haven't touched the db files. So do you think turning all of those to 0 will help also? Is that what you said you did in 16?
 

Anth James

League 1
22 September 2017
@papinho81 What did you use to slow animations down? I've been looking at reaction time, and I slowed that by 0.2 for each variable and I really like it. It seems to help let the animation play out a little better. I've attempted to try and slow the game down further with changes to acceleration speeds, base sprint speeds etc and it feels good at times, but just causes lots of other little issues and the whole gameplay breaks down. Very frustrating.
 

papinho81

Premiership
22 March 2017
Chaumont Switzerland
Olympique de Marseille
@papinho81 What did you use to slow animations down? I've been looking at reaction time, and I slowed that by 0.2 for each variable and I really like it. It seems to help let the animation play out a little better. I've attempted to try and slow the game down further with changes to acceleration speeds, base sprint speeds etc and it feels good at times, but just causes lots of other little issues and the whole gameplay breaks down. Very frustrating.
I don't know if the animations are actually slow in the mod I shared the other day (did you try it?). If anything it feels heavier than the default game and I would say it feels less skaty than the default game.
The only variable I used and I tested so far to slow down the animations is this one:
gp_actor_movement_runtime
ATTR_DribbleMinTouchInterruptPenalty
set to 6 instead of 2.

I think most of the weight and the animations slow down I "feel" comes from the the ball physics mod that uses an heavier and slightly bigger ball.
I have stopped using the perception mod as I feel it makes the cpu more direct in its approach (again, I feel, it worth what it worth).

Surprisingly with the combo I shared I also do have more fouls (not many though) without touching refs things.

It is still far from being great. The cpu passing is wayyyyyyyyyy too accurate for lower teams. It will be my next move, trying to make passing errors (and shooting) more attribute dependent (I am tired to look at those formation files, they are so frustrating).
There is a bunch of variables in the passerror file that have the attr terminology, they are likely (I hope) attribute dependent errors.
The first cpuai file (can't remember the name, ballcarry something?) looks interesting to change the cpu approach to the game. I think it can help making the cpu players eating space when it exist with the ball at feet. Making less one time pass (there is a one time pass modifier ). There is several time the same variables depending on players traits (one time passer, selfish, ect) that might help getting the balance.

Yesterday I played a cup game in my test Newcastle CM against Brentford FC. Even though I dominated the game and had more chances, they trashed me 4-1 because all their counter attack were executed with perfect passes. They ended the game with 85% pass accuracy. They might be a good Championship team but still...
 

Chris Davies

Chief PESsimist
Staff
14 May 2003
Tranmere Rovers
Yesterday I played a cup game in my test Newcastle CM against Brentford FC. Even though I dominated the game and had more chances, they trashed me 4-1 because all their counter attack were executed with perfect passes. They ended the game with 85% pass accuracy. They might be a good Championship team but still...
This is my biggest problem at the minute. I mentioned playing as England vs Kosovo (who are added in the Worldwide Football patch, but play any 5* team against any 1* team) and it's never a huge victory because their passing accuracy and general play is a much higher standard than it should be. Not PES levels (where the worse team is always better than you, because the difficulty levels are so broken)... But not good.

I wonder if that's just World Class, Legendary and/or Ultimate all overriding attributes, and that there's no way around it.
 

papinho81

Premiership
22 March 2017
Chaumont Switzerland
Olympique de Marseille
This is my biggest problem at the minute. I mentioned playing as England vs Kosovo (who are added in the Worldwide Football patch, but play any 5* team against any 1* team) and it's never a huge victory because their passing accuracy and general play is a much higher standard than it should be. Not PES levels (where the worse team is always better than you, because the difficulty levels are so broken)... But not good.

I wonder if that's just World Class, Legendary and/or Ultimate all overriding attributes, and that there's no way around it.
I hope we can do something about it. That would be really frustrating if all those variables we can edit had little to no impact on things like this. It is already frustrating to not have files with the tactics details in there.
 

Anth James

League 1
22 September 2017
I don't know if the animations are actually slow in the mod I shared the other day (did you try it?). If anything it feels heavier than the default game and I would say it feels less skaty than the default game.
The only variable I used and I tested so far to slow down the animations is this one:
gp_actor_movement_runtime
ATTR_DribbleMinTouchInterruptPenalty
set to 6 instead of 2.

I think most of the weight and the animations slow down I "feel" comes from the the ball physics mod that uses an heavier and slightly bigger ball.
I have stopped using the perception mod as I feel it makes the cpu more direct in its approach (again, I feel, it worth what it worth).

Surprisingly with the combo I shared I also do have more fouls (not many though) without touching refs things.

It is still far from being great. The cpu passing is wayyyyyyyyyy too accurate for lower teams. It will be my next move, trying to make passing errors (and shooting) more attribute dependent (I am tired to look at those formation files, they are so frustrating).
There is a bunch of variables in the passerror file that have the attr terminology, they are likely (I hope) attribute dependent errors.
The first cpuai file (can't remember the name, ballcarry something?) looks interesting to change the cpu approach to the game. I think it can help making the cpu players eating space when it exist with the ball at feet. Making less one time pass (there is a one time pass modifier ). There is several time the same variables depending on players traits (one time passer, selfish, ect) that might help getting the balance.

Yesterday I played a cup game in my test Newcastle CM against Brentford FC. Even though I dominated the game and had more chances, they trashed me 4-1 because all their counter attack were executed with perfect passes. They ended the game with 85% pass accuracy. They might be a good Championship team but still...
Yeah ball physics definitely is a big contributor to slowing the game down. Passing error is just as big as well and can have an influence on ball physics too somehow. What I really discovered is that it's often a combination of many elements that either click together or they do the opposite and cause an issue which breaks the gameplay down. I can see now how FIFA and PES often are terrible after small incremental updates as it really seems that a slight change of one thing can unbalance another and it can all unravel from there.

The cpu ai ball handler is great, you can increase many things which impact the ai behaviour like slide tackle probability, long shot probability and even for someone who has the trait of long shot shooting. I knocked down the one time pass probability in my mod to 0.1 but it could probably go further than that.

The passing error for me is the way to get differences between the teams but there's a couple of issues that make it difficult. So in the first mod I released I was getting really accurate pass % compared with real life and there was a BIG difference between good and bad teams from the games I played. I noticed big teams could really pin an poor team into their own half because they would struggle to play the ball out successfully. There were lots of variables but for example ground passes I lifted left and right error from 1 (default) to 13.2 This is how big of a change I needed to create that gap.

I was able to play 15 min halves and have some great games that were quite tactical. The problem though is that the error of passing isn't always as refined as we want it to be. So the AI can have moments of passing the ball straight to you, or out of play. So games were good, but had frustrating moments in them probably too often.

I like how playing good teams too, it became difficult to create scoring chances, but rewarding. You actually had to take time because the risk v reward becomes much more about finding the moment to make the right pass into the right position. I found if you played very slowly you could keep your pass % quite high, and the AI could as well. If you went all out you know for a fact you would lower your passing and give the ball away lots and it would open a game up. The problem again though - you had to live with those nonsensical moments and that was difficult because they were too frequent.

When I started to try and fix for the Sim Mod I tried to make a deeper system which didn't just have left and right error for different passes (you need different errors for different types of passes as they all seem to come out differently). Again for example, ground passes left and right error I dropped to around 7 (from 13 in the previous mod) and I made lots of changes to miss rate vs attribute, miss rate v difficulty, things like this. This would then either be over or under powered quite often. I honestly tried around 40-50 different combinations to find a balance that was good (but not perfect!). It's finding that right balance or living with the fact the AI will make errors they probably shouldn't at this point, but I really think the best way to get differences between teams is to widen the gap of passing mistakes as this is probably one of the biggest contributers in real football too. I might look at creating quite a wide gap between miss rate v attribute instead of scaling it down slowly and see how that goes.
 
Last edited:

Anth James

League 1
22 September 2017
I forgot to add in as well, passing error seems to impact first touch error and speed as well. When you have really low error (even in the difficulty or attribute setting) good passes are hit with more confidence it feels and they're quicker. Same goes the other way. It also impacts the error on control, more error seems to create some more erratic control of the ball.
 

Anth James

League 1
22 September 2017
@papinho81 @Chris Davies Sorry for the spam, I've been attempting to get a better result with this. I'm playing around with the miss rate v attribute for different passes within my mod trying to create a bigger difference between good and bad passers. Early tests have been good. After a few unsuccessful attempts I got it to a decent point. I played West Brom and after 30 mins they had 72% completion with 0 shots to my 8(5), I dominated them and it wasn't frustrating, a couple of 'odd' passes but nothing quite as ridiculous as before so far. Then I played Man Utd, 85% pass completion with 3(1) shots (after about the same time), they definitely held the ball better. More variety and unpredictability too. It needs some more refining, but I'm confident it can be done - just could get messy trying to get there haha.
 
Last edited:

papinho81

Premiership
22 March 2017
Chaumont Switzerland
Olympique de Marseille
Yeah ball physics definitely is a big contributor to slowing the game down. Passing error is just as big as well and can have an influence on ball physics too somehow. What I really discovered is that it's often a combination of many elements that either click together or they do the opposite and cause an issue which breaks the gameplay down. I can see now how FIFA and PES often are terrible after small incremental updates as it really seems that a slight change of one thing can unbalance another and it can all unravel from there.

The cpu ai ball handler is great, you can increase many things which impact the ai behaviour like slide tackle probability, long shot probability and even for someone who has the trait of long shot shooting. I knocked down the one time pass probability in my mod to 0.1 but it could probably go further than that.

The passing error for me is the way to get differences between the teams but there's a couple of issues that make it difficult. So in the first mod I released I was getting really accurate pass % compared with real life and there was a BIG difference between good and bad teams from the games I played. I noticed big teams could really pin an poor team into their own half because they would struggle to play the ball out successfully. There were lots of variables but for example ground passes I lifted left and right error from 1 (default) to 13.2 This is how big of a change I needed to create that gap.

I was able to play 15 min halves and have some great games that were quite tactical. The problem though is that the error of passing isn't always as refined as we want it to be. So the AI can have moments of passing the ball straight to you, or out of play. So games were good, but had frustrating moments in them probably too often.

I like how playing good teams too, it became difficult to create scoring chances, but rewarding. You actually had to take time because the risk v reward becomes much more about finding the moment to make the right pass into the right position. I found if you played very slowly you could keep your pass % quite high, and the AI could as well. If you went all out you know for a fact you would lower your passing and give the ball away lots and it would open a game up. The problem again though - you had to live with those nonsensical moments and that was difficult because they were too frequent.

When I started to try and fix for the Sim Mod I tried to make a deeper system which didn't just have left and right error for different passes (you need different errors for different types of passes as they all seem to come out differently). Again for example, ground passes left and right error I dropped to around 7 (from 13 in the previous mod) and I made lots of changes to miss rate vs attribute, miss rate v difficulty, things like this. This would then either be over or under powered quite often. I honestly tried around 40-50 different combinations to find a balance that was good (but not perfect!). It's finding that right balance or living with the fact the AI will make errors they probably shouldn't at this point, but I really think the best way to get differences between teams is to widen the gap of passing mistakes as this is probably one of the biggest contributers in real football too. I might look at creating quite a wide gap between miss rate v attribute instead of scaling it down slowly and see how that goes.
This is my 2 cents on how I think attribute should be set apart to better represents the difference in quality between players.
Intro note: I absolutely didn't dive into those ATTR files and I assume the difference between players is for the most part on a rather linear scale.

I think every skill related attribute should be set on an exponential scale (or a ^2 or a ^3) as it was the case for example for PES5. This is probably the only way to set a part good from bad players. PES5 is notorious for its player individuality and I think the attribute scale chosen was really for something (or was it is linear up to 90 then exponential?).
Concerning physical stats I think a linear scale is probably enough (except my be for strength?). For example I don't think increasing the gap for top speed is a good idea considering how speed can be OP in this game.
 

papinho81

Premiership
22 March 2017
Chaumont Switzerland
Olympique de Marseille
@papinho81 @Chris Davies Sorry for the spam, I've been attempting to get a better result with this. I'm playing around with the miss rate v attribute for different passes within my mod trying to create a bigger difference between good and bad passers. Early tests have been good. After a few unsuccessful attempts I got it to a decent point. I played West Brom and after 30 mins they had 72% completion with 0 shots to my 8(5), I dominated them and it wasn't frustrating, a couple of 'odd' passes but nothing quite as ridiculous as before so far. Then I played Man Utd, 85% pass completion with 3(1) shots (after about the same time), they definitely held the ball better. More variety and unpredictability too. It needs some more refining, but I'm confident it can be done - just could get messy trying to get there haha.
You are testing this with the pass error slider at 65 don't you?
 

Anth James

League 1
22 September 2017
This is my 2 cents on how I think attribute should be set apart to better represents the difference in quality between players.
Intro note: I absolutely didn't dive into those ATTR files and I assume the difference between players is for the most part on a rather linear scale.

I think every skill related attribute should be set on an exponential scale (or a ^2 or a ^3) as it was the case for example for PES5. This is probably the only way to set a part good from bad players. PES5 is notorious for its player individuality and I think the attribute scale chosen was really for something (or was it is linear up to 90 then exponential?).
Concerning physical stats I think a linear scale is probably enough (except my be for strength?). For example I don't think increasing the gap for top speed is a good idea considering how speed can be OP in this game.
That’s really good thinking and makes lots of sense. Definitely agree about physical stats too, gets very tricky with those. Nah, I test and use 50 error for passing. I haven’t looked at shooting error so I use 65 for that which seems to work alright for now.
 

Rafter84

Non-League
16 June 2020
I absolutely confirm that "slowing down" animations is not just about the players, but also about the ball.

With a heavier, or even just slower ball, you will already see the difference.
Tweaking acceleration values, in the sliders and in Frosty Editor, does help too, of course.

Based on my experience, the best results are combining both, but it's always a thin line to thread. Edit too much and it's robotic, edit too little and it's... FIFA... 🤣

In the end, we can tweak available values, but if an animation is done that way, there is no way we can change that. Going next-gen, I think that's what the main addition and difference-maker should be: not 8k graphics, but A LOT MORE and A LOT MORE REALISTIC animations (and of course, the right game speed to let those animations actually show...which is what PES does)
 

Anth James

League 1
22 September 2017
It's a bit of a nightmare to get a balance but I'm making good progress with the pass error adjustments. Trying to do what we spoke about yesterday - to get more of a difference between teams. If I'm understanding correctly, what I'm noticing is that EA have been a little lazy in some of the programming for certain pass error. So MissRate V Attributes some passes like ground passes, the 7 variables go up in small portions for the X axis (0, 0.03, 0.5, 0.6, etc, whereas lobs go 0=0, and then 1-6 X = 1 the whole way. I think this explains why lobs are extra sensitive to error as the passes seem to go well, or way, way off. There's no build up. Among other balancing, I've attempted to rewrite these types of errors to get more 'slight' errors. Having some good results with different teams, big v small and noticing much more of a difference between them whilst still keeping the core gameplay I've had. I'll keep trying to refine and hopefully share soon! Thanks @papinho81 for the tip of large increments between the attributes, that seems to be quite effective!
 
Last edited:

fernadez

Premiership
28 May 2003
E.7
Tottenham Hotspur
It's a bit of a nightmare to get a balance but I'm making good progress with the pass error adjustments. Trying to do what we spoke about yesterday - to get more of a difference between teams. If I'm understanding correctly, what I'm noticing is that EA have been a little lazy in some of the programming for certain pass error. So MissRate V Attributes some passes like ground passes, the 7 variables go up in small portions for the X axis (0, 0.03, 0.5, 0.6, etc, whereas lobs go 0=0, and then 1-6 X = 1 the whole way. I think this explains why lobs are extra sensitive to error as the passes seem to go well, or way, way off. There's no build up. Among other balancing, I've attempted to rewrite these types of errors to get more 'slight' errors. Having some good results with different teams, big v small and noticing much more of a difference between them whilst still keeping the core gameplay I've had. I'll keep trying to refine and hopefully share soon! Thanks @papinho81 for the tip of large increments between the attributes, that seems to be quite effective!
Doing the Lord's work.
 

papinho81

Premiership
22 March 2017
Chaumont Switzerland
Olympique de Marseille
It's a bit of a nightmare to get a balance but I'm making good progress with the pass error adjustments. Trying to do what we spoke about yesterday - to get more of a difference between teams. If I'm understanding correctly, what I'm noticing is that EA have been a little lazy in some of the programming for certain pass error. So MissRate V Attributes some passes like ground passes, the 7 variables go up in small portions for the X axis (0, 0.03, 0.5, 0.6, etc, whereas lobs go 0=0, and then 1-6 X = 1 the whole way. I think this explains why lobs are extra sensitive to error as the passes seem to go well, or way, way off. There's no build up. Among other balancing, I've attempted to rewrite these types of errors to get more 'slight' errors. Having some good results with different teams, big v small and noticing much more of a difference between them whilst still keeping the core gameplay I've had. I'll keep trying to refine and hopefully share soon! Thanks @papinho81 for the tip of large increments between the attributes, that seems to be quite effective!
Yesterday evening I have added the the missratevsattr variable for the kickerror_passshot general file and the ground pass one to the cheat table.
The variable from the general file didn't seem to do much even with extreme values.
The ground pass has a strong impact though and not just on passes it seems to me. I first tried with extreme values for the Y up to 1000 and the ball was bubbling out of every players foot. I then made a scale with low values for the Y of the point 6 and 7 (around 1), then extreme increasing values for the others (100 to 10000). The like of DeBruyne and David Silva didn't have much trouble passing (they still had the very odd one sometimes though but rarely), others were playing with an heavy and floaty beach ball.
My feeling is that the tails values might have also an influence on the head ones to some extent.
I ve seen passes I never seen before like a toe pock by de bruyne to set a shot for aguero.

A bonus of the error seems to be the extra time players takes on the ball before actually passing it. It quite significantly slow down the game.

Passing stats were lower but not extremely low though, in the middle 70s.
I think it is because AI players adjusted their way of passing, trying to take risks less often.
In a cpu vs cpu tes between Newcastle and City, city had 75% ball possession. I think because they didn't rush their passes and the opponents difficulty to escape the pressure and giving back the ball more often than not.


The values I used were extreme and gave a gameplay that looked nothing like professional football but it shown there is a lot of potential in this and that attributes can matter.
 
Last edited:

Anth James

League 1
22 September 2017
There's a particular pass that the AI seems to make that's really annoying me when you introduce error, where they essentially just pass you the ball either in midfield or coming out of defence when trying to play it forward through lines. I've noticed it happens a little too often, even with good teams and even with little error introduced. I'm looking at miss rate v difficulty to try and get rid of this. I think I'm figuring it out slowly, but getting the balance right again is the challenge. I’m quite confident that it’s not relating to difficulty level, but rather the difficulty of the pass. What’s strange is ground passes only has 4 variables of difficulty (0-3), but the general file and most other passes there are 8 (0-7). So I think what this is saying is there are 4 types of pass difficulty ‘types’ for ground passes and 8 for most others. Speculation but I think this could be why the AI plays more directly as well, as there’s less ‘safe’ passes programmed. I'm thinking a small jump in miss rate on the lower end might be the way to go with a bigger jump towards the higher end..

Edit: I realised you can add in more variables, so I created 0-7 for ground pass miss rate v difficulty (instead of 0-3) and I’m trying to create a slower progression so there’s more of a subtle error to the pass. I don’t have the balance right yet (surprise, surprise) but what I noticed pretty much confirmed my suspicion. The AI played much more ‘safe’, with far less risky passes, more sideways and conservative. This resulted in a much higher pass completion % though so again needs to be rebalanced. As @papinho81 said yesterday, the error rate slows down play and it seems to present itself in different ways and can give us more control in regards to how the AI plays. These pass error files really have a huge impact on gameplay.
 
Last edited:

papinho81

Premiership
22 March 2017
Chaumont Switzerland
Olympique de Marseille
Last edited:

final lap

League 1
21 August 2007
Miami, FL
So I reinstalled FIFA 20 yesterday after a long time. Played with no mods and noticed the AI defensive awareness seems to be much better. Did EA fix this? After that I installed Anth's mod and had fun games although not enough to really tell what was different. I just want to be sure that I am applying the mod correctly. You need to import and apply all 3 files correct? I have the "Worldwide Football Mod" as well. Should the gameplay mod be at the top or bottom of the list of applied mods? Thanks for the help.
 
Top Bottom