Evoweb Gameplay Lab, FIFA 20 (Frostbite FIFAs?)

[manmachine]

My only wishes are that, firstly, the "pass quality scalar" tweaks did more (as I'm not seeing much of a drop in pass accuracy), but the biggie:

Yeah, it doesn't affect accuracy, but it still looks good if you try a 180 degree pass without turning and have decreased the values substantially. The ball loses most of its speed and it looks like a miss hit:

PASS_QualityScalarVSAng (in kickpass_passing_runtime) is pretty simple. The X-axis is the pass angle from 0 to 180 degrees, and the Y-axis is the "pass quality". It affects the pass speed only and the default values are.... well, very generous to the fut kiddies. The pass quality ranges from 1 (0 degrees) to 0.8 (for 180 degree passes), I like it scaled from 1 to 0. Don't be afraid to go overboard, it's just a slower pass the more difficult the angle and it feels nice!

For accuracy look into the scalar variables (in the same file I think, and they are simple values, not curves). This is from memory, but the variables are called something like "LeftsidehitScalar", "OverhitScalar" and similar. Increasing these is key to humanising passing because you can get many passes that are not played perfectly in one's path. Passes are more loose and the receiver may have to adjust his movement accordingly. I particularly like the effect of passes played slightly to the left/right of the receiver. Not so much overhit or underhit passes as I think these look a bit unnatural sometimes.

@manmachine do you remember which files you adjusted to get more long shots from the AI?

File was called "shot_decision" something. In the cpuai folder I think.

 

[Anth James]

For accuracy look into the scalar variables (in the same file I think, and they are simple values, not curves). This is from memory, but the variables are called something like "LeftsidehitScalar", "OverhitScalar" and similar. Increasing these is key to humanising passing because you can get many passes that are not played perfectly in one's path. Passes are more loose and the receiver may have to adjust his movement accordingly. I particularly like the effect of passes played slightly to the left/right of the receiver. Not so much overhit or underhit passes as I think these look a bit unnatural sometimes.

Yes spot on with these. I spent so long yesterday going through heaps of those variables and found this to be the case. Left and right really makes it look natural, the under, lower, upper hit etc look pretty odd. It’s just a matter of perfecting the values which is frustrating! I’m making some pretty good progress, will post something soon when it’s at a decent level! Makes quite a big difference and really slows down the play.

 

[Anth James]

...I’m also looking at the idea of slowing down the dribbling running and turning speed as I thought it could be a little more natural and even up the 1v1 a little. Compared to 16 for example it’s much harder to get through a defence. You can really shut out or be shut out. On 20 it’s very difficult to really just close someone out. I think congestion has helped but I think think can still be more realistic. In theory it should allow for defenders to just get closer to attackers if they have to take slightly longer when on the ball.

 

[Anth James]

Ok guys, give this a try for passing error. I've been having some good results with it. The challenge was to not introduce so much error that you didn't feel like you had control of the ball at all, but enough to make play more natural, random and varied.

I've found you really have to protect the ball more when moving out from defence, and trying a risky ball forward is now much more risky. With long balls, this was more touchy and it could quickly feel like you had no control, so this is toned down compared to ground passes, but seems to have a subtle impact now (perhaps too subtle, need to keep testing).
I've seen some really natural, organic play so far using this. One thing I did notice was the AI can seemingly just pass you the ball on occasion (or directly out of bounds). I need to test that some more. When I studied the replays of those times, they did miss their intended pass (left or right) but it obviously came straight to me as a result (again, as forward passing through the lines is harder to execute now). It might need some refining, as there seems to be a little trade-off with certain mistakes here and there so let me know what you think and hopefully we can improve further.

Pass error trial - Deleted. Try pass error 2.0 instead in the below post.

 

[Anth James]

Sorry again for the spam, try this version instead. It's more refined. Better long ball error and more natural ground pass error. There’s still a few too many easy giveaways which I’d like to get rid of, but in general I’m really happy with how this is playing out at the moment (only tested on legendary 20 min matches using v1 ball physics).

Pass error 2.0: https://mega.nz/file/8VMQDaQS#u7QsWNLAVLX-rdFV4If5oyWm0EWsgqlGRNUhBVlddGw

 

[Anth James]

Does anyone know what 'UCC' refers to in the files under gp_player? I feel like it might be relating to marking or pressing...

 

[papinho81]

Does anyone know what 'UCC' refers to in the files under gp_player? I feel like it might be relating to marking or pressing...

No idea, but @Rafter84 recently shared a locale.ini line with UCC_chasingplayers... and said it related to players activity.
Could be!

 

[tufkar84]

Does anyone know what 'UCC' refers to in the files under gp_player? I feel like it might be relating to marking or pressing...

In all honesty, I have no idea what UCC stands for.

In the locale files, UCC is generally used in relation with Unlocking Content: unlocking all boots, all stadiums and so on... or unlocking features... So maybe Unlocked Content Code?

In Frosty Editor, under gp_player, it seems all those fields are related to Goal Celebration behavior... so in this case, it could be Unlocked Celebration Content?

🤔

The only way to know is to test it, I'm afraid... and I hope the lines shared previously have effects on the gameplay rather than celebrations... 😅

 

[Anth James]

I found a different way to increase the left and right error range for passing, shooting, heading etc. It's in gp_physics, gp_physicstouchcontactpass, shoot etc. Then find the values called 'error_left_miss_range' and 'error_right_miss_range'. Default is set to 12 for most values.

 

[papinho81]

@Anth James great job with your last two mods, they really add a new layer of improvement to the game. I particularly like the pass error one. It is so refreshing to see more mistakes made by the user and the cpu. It makes for mor 50/50 balls ence more battle throughout the pitch.
Do you want me to add them to the list?

 

[Anth James]

@Anth James great job with your last two mods, they really add a new layer of improvement to the game. I particularly like the pass error one. It is so refreshing to see more mistakes made by the user and the cpu. It makes for mor 50/50 balls ence more battle throughout the pitch.
Do you want me to add them to the list?

Thanks man. Yeah the pass error 2.0 you can add for sure.

Do you think the runpathblock and passing lanes has an impact? I’m still not 100% convinced! I quickly tried changing the same values which I had set quite low to 100 just to see the opposite effect and noticed nothing really obviously different. Didn’t try it for long but nothing was clearly changed...

 

[papinho81]

Thanks man. Yeah the pass error 2.0 you can add for sure.

Do you think the runpathblock and passing lanes has an impact? I’m still not 100% convinced! I quickly tried changing the same values which I had set quite low to 100 just to see the opposite effect and noticed nothing really obviously different. Didn’t try it for long but nothing was clearly changed...

Not 100% sure as I tested it with others changes too. At least it didn't make it worst
All right I will add the pass error mod.

 

[papinho81]

This a link toward a cheat engine table that allows you changing and testing gameplay values without quitting the game:
http://www.mediafire.com/file/yv1jt7stlrfdevr/FIFA20.CT/file

How to use it: Open your game (can be with frosties), double click on the ct table, it will attach automatically to the game.

The original idea, the design, the code and many variables are from @manmachine and some variables have been added by myself strictly recycling the code provided by manmachine.
It is very much a work in progress and anyone addition to the table is welcomed. Early added variable have a quite simple design while more recent ones comes with a script entry that allows quickly resetting default values (very much welcomed).
If you want to use the table and see values different from the default expected values, open the script (double click) of the variable you wan to edit and uncomment the line define... with the frosty offset and comment the other define... with the non frosty offset or vice et versa (like here:

Spoiler

)

The CT was designed on the 12.05.2020 version of the game and there are chances that offsets are differents in different game versions. Let us know if that's not the case.

 

[Anth James]

Guys, I think this could be a good one! I've been looking at marking/pressing/defensive covering and how it can be improved.

I have one mod I'll post below and there's another which I'm still looking at

So for the first, I've made a few changes to different values to achieve what I think is a decent improvement to the game. Some changes are to 'minperceptionforaccuratemarking' and 'minperceptionforaccuratemarkingwhenoutofposition'. I've also changed some of the marking speed to get the defenders across a little faster based on ball distance and attacking threat. Finally I changed quite a few values in the CPUAI Marker Reactions. The result is what feels like much better awareness for marking across the pitch which results in far less gaps opening up. It still needs more testing and it isn't perfect as getting the midfielders to track into the box is still infuriating and I haven't figured that out quite yet! I think it almost certainly needs to be used with @papinho81's defensive line fix too, otherwise defenders can track and break their defensive line. With the fix it seems like they still hold shape!

Marking and Marking Perception - https://mega.nz/file/RdF3XYqJ#v-38Zft_8l7r1tpUB3Fs3BD1rh1kXycdvzcqNQCnxY4

 

[papinho81]

Guys, I think this could be a good one! I've been looking at marking/pressing/defensive covering and how it can be improved.

I have one mod I'll post below which makes improvements to defensive marking/awareness and there's another which I'm working on but think I've made some progress.

So for the first, I've made a few changes to different values to achieve what I think is a decent improvement to the game. Some changes are to 'minperceptionforaccuratemarking' and 'minperceptionforaccuratemarkingwhenoutofposition'. I've also changed some of the marking speed to get the defenders across a little faster based on ball distance and attacking threat. Finally I changed quite a few values in the CPUAI Marker Reactions. The result is what feels like much better awareness for marking across the pitch which results in far less gaps opening up. It still needs more testing and it isn't perfect as getting the midfielders to track into the box is still infuriating and I haven't figured that out quite yet! I think it almost certainly needs to be used with @papinho81's defensive line fix too, otherwise defenders can track and break their defensive line. With the fix it seems like they still hold shape!

Marking and Marking Perception - https://mega.nz/file/RdF3XYqJ#v-38Zft_8l7r1tpUB3Fs3BD1rh1kXycdvzcqNQCnxY4

Holly Molly! That sounds good!

 

[Anth James]

Holly Molly! That sounds good!

If you think it’s not that good/no impact I won’t get offended either so feel free to tell me haha. I was convinced the high/low pressure was working as well (the other mod I’m working on) but now I’m not so sure. I will keep trying with it as I do think there’s something in there. I wonder too if teams like City (and probably others) have an additional code in the database maybe which makes them play a certain way because just using their tactics on a different team isn’t the same result..

 

[papinho81]

If you think it’s not that good/no impact I won’t get offended either so feel free to tell me haha. I was convinced the high/low pressure was working as well (the other mod I’m working on) but now I’m not so sure. I will keep trying with it as I do think there’s something in there. I wonder too if teams like City (and probably others) have an additional code in the database maybe which makes them play a certain way because just using their tactics on a different team isn’t the same result..

I will try to give it a go during my lunch break. This also what I am really after, trying to improve the marking distance/awareness, getting players being more involved in a team effort rather than in an individual effort. For the tactics it is quite a mistery as there is really few variables that refers to tactics in the files we can see with frosty..

Testing would be easier if we could be in the files developpers brain. It still very much a mistery what the curves points means for many variables and I feel we could make faster progress if we could break it.

 

[thirdeyekidman]

Hey everyone.

After years of playing a modded PES17, I recently got back to FIFA with 20 and currently using FIFER's realism mod.
For the past couple of days I've been reading through this thread and I admire what you guys are doing. I'll try to contribute by testing your files and suggestions out etc

 

[Anth James]

Oh and I forgot to add (which isn't in the posted mod), I've also been using Defender loses his attacker rate -all at 0.02 and duration all at 1. I'm about to test if upping the 'marking importance' in conjunction with the rest of of the marking mods can get this midfielders to track into the box. I've got them tracking quite well right up until the penalty box and then just dead stop. Drives me crazy.

 

[papinho81]

Oh and I forgot to add (which isn't in the posted mod), I've also been using Defender loses his attacker rate -all at 0.02 and duration all at 1. I'm about to test if upping the 'marking importance' in conjunction with the rest of of the marking mods can get this midfielders to track into the box. I've got them tracking quite well right up until the penalty box and then just dead stop. Drives me crazy.

It must a line the code that tells midfielder to not go into the box. I ve tried to change the maxdistfrom best position from the position folder to try to acheive that but with no much success yet even though I haven't ried all the variables yet.
@Topaz some time ago said he managed to have them in the stands, I asked him by PM if he remebered what he changed but either he forgot to look for or what he changed lol.
Topaz if you are around

Also I have just noticed something bad, every time you hit the the shield button the cpu react in consequence. If you do that while sprinting it will not affect your player sprint but the cpu will take one or two step of hesitation that allows you to outsprint it even more,

 

[Anth James]

It must a line the code that tells midfielder to not go into the box. I ve tried to change the maxdistfrom best position from the position folder to try to acheive that but with no much success yet even though I haven't ried all the variables yet.
@Topaz some time ago said he managed to have them in the stands, I asked him by PM if he remebered what he changed but either he forgot to look for or what he changed lol.
Topaz if you are around

Yeah I was reading that today and was going ask him if what he was doing worked. Marking Importance did not work haha. It must be a line somewhere because it's just too obvious. I was just playing before with the mod I posted and the midfielder was properly tracking this run for ages, sprinting right with him and as soon as he entered the box 100 - 0 just like that.

 

[Anth James]

I ve tried to change the maxdistfrom best position from the position folder to try to acheive that but with no much success

what effect did maxdistfrombest have on the game? I figured it’s to do with the length of attacking runs but haven’t got to it yet.

 

[papinho81]

what effect did maxdistfrombest have on the game? I figured it’s to do with the length of attacking runs but haven’t got to it yet.

They are 3 of them (3 rundistfromformation files) a general one (?) a X one (for the position along the width? Smaller X than Y) and a Y one (position along the length?).
I only tested the first varibale of the general one and the first variable of the Y one.
By defaut the Y value of the general one are all null. When increasing the Y value (are you still with me here?) the players are tracking the players they are marking or chasing along much longer distance (cdm will still stop at the box entry though). It is like it's telling the player and more specifically the active player, your anchor spot is here and your allowed to move from your anchor up to here. I tested it with all the 8 entries of the general one with Y=100 and it makes everything more dynamic at least less lethargic but still not enought to wake up all the players through out the pitch.

I didn't test enougth the Y file one but I supsect it does the same but in one dimension. I didn't try to change X values though but it could be intresting to try that too.

Those 2 curves variables are in the cheat engine table, so if you have CE you can quickly try it.

 

[Anth James]

They are 3 of them (3 rundistfromformation files) a general one (?) a X one (for the position along the width? Smaller X than Y) and a Y one (position along the length?).
I only tested the first varibale of the general one and the first variable of the Y one.
By defaut the Y value of the general one are all null. When increasing the Y value (are you still with me here?) the players are tracking the players they are marking or chasing along much longer distance (cdm will still stop at the box entry though). It is like it's telling the player and more specifically the active player, your anchor spot is here and your allowed to move from your anchor up to here. I tested it with all the 8 entries of the general one with Y=100 and it makes everything more dynamic at least less lethargic but still not enought to wake up all the players through out the pitch.

I didn't test enougth the Y file one but I supsect it does the same but in one dimension. I didn't try to change X values though but it could be intresting to try that too.

Those 2 curves variables are in the cheat engine table, so if you have CE you can quickly try it.

I gave a few quick extreme variables a try. I could be wrong, but my feeling is the distance we're increasing with these are impacting the length of attacking runs rather than the marking. Did you find that?

 

[papinho81]

I gave a few quick extreme variables a try. I could be wrong, but my feeling is the distance we're increasing with these are impacting the length of attacking runs rather than the marking. Did you find that?

It does, but I think not only. Attacking players are tracking much deeper in their own half (with the general one at least). Against Chelsea I had William tracking me all the way from his wing to the other side of the pitch.

I played earlier with your pass error mod and I really liked how it brings a slower tempo, more midfield play and fouls. It is like it does not only impact passing. I have seen more touch errors and inaccurate challenge by the cpu. The only thing is it seems too much for the cpu. Teams like Bayern have 75-+ % pass accuracy and can't really keep possession. As a user I know if I do a pass with a bad angle it will go wrong but the cpu doesn't adapt and still try them too often.

 

[Anth James]

It does, but I think not only. Attacking players are tracking much deeper in their own half (with the general one at least). Against Chelsea I had William tracking me all the way from his wing to the other side of the pitch.

I played earlier with your pass error mod and I really liked how it brings a slower tempo, more midfield play and fouls. It is like it does not only impact passing. I have seen more touch errors and inaccurate challenge by the cpu. The only thing is it seems too much for the cpu. Teams like Bayern have 75-+ % pass accuracy and can't really keep possession. As a user I know if I do a pass with a bad angle it will go wrong but the cpu doesn't adapt and still try them too often.

Are you playing Legendary? I haven't had a team like Bayern down at 75%, the good teams are usually 83-88% from what I've played, with lower rated teams a little lower in the mid to high 70s. I agree though it can be a little too much for the CPU at times. I've tried to compensate by lowering their pass error using sliders, but I'm not 100% if that still has an impact if the core values have been changed. I think the added pass error might exacerbate the impact of first touch as the passes aren't being received perfectly (which has a knock-on effect of loose challenges). The more I play with it and adjust, the more I like it too.

 

[papinho81]

Are you playing Legendary? I haven't had a team like Bayern down at 75%, the good teams are usually 83-88% from what I've played, with lower rated teams a little lower in the mid to high 70s. I agree though it can be a little too much for the CPU at times. I've tried to compensate by lowering their pass error using sliders, but I'm not 100% if that still has an impact if the core values have been changed. I think the added pass error might exacerbate the impact of first touch as the passes aren't being received perfectly (which has a knock-on effect of loose challenges). The more I play with it and adjust, the more I like it too.

Playing on legendary yes, but using other things together including ball physics things and also supposely first time pass error increase. Without your mod Bayern had 85+ %. I play on normal speed and default sliders.

This is what I used with it: http://www.mediafire.com/file/pg6zrk5f6ahue74/Gameplay_All2.fbproject/file

 

[Anth James]

Playing on legendary yes, but using other things together including ball physics things and also supposely first time pass error increase. Without your mod Bayern had 85+ %. I play on normal speed and default sliders.

This is what I used with it: http://www.mediafire.com/file/pg6zrk5f6ahue74/Gameplay_All2.fbproject/file

I'll check it out. I wonder if the first time pass error increase might impact too. Try knocking the pass error on sliders down a touch and see if that helps too. It's a tough balance with the pass error, I feel like it has to be quite extreme for ground passes to really impact the play, but like you said it can be frustrating for the cpu side at times. The default error for left and right is 1. I increased ground passes to just over 13! You can try lowering that in the project file too and see if you can find a sweet spot you're happy with. I tried adjusting so many times and landed at that particular combination.

 

[papinho81]

I'll check it out. I wonder if the first time pass error increase might impact too. Try knocking the pass error on sliders down a touch and see if that helps too. It's a tough balance with the pass error, I feel like it has to be quite extreme for ground passes to really impact the play, but like you said it can be frustrating for the cpu side at times. The default error for left and right is 1. I increased ground passes to just over 13! You can try lowering that in the project file too and see if you can find a sweet spot you're happy with. I tried adjusting so many times and landed at that particular combination.

Yes I will try to adjust it to fit me (If it only it could be that easy with all the other parts of the game). The thing is if it could force the decision making of the CPU to incite it to do less risky pass, such as back heel, passes with poor angle, it would be fantastic!

 

[Anth James]

I've made a couple of new additions and I'm actually really happy with how the game is playing now. I'm finding it very difficult to create chances as the AI defence is really doing its job. I could probably back it off slightly even. Playing with 20 min matches I'm averaging really low shot numbers, but the games are playing out really well and I'm finding it far less frustrating. The passing error (possibly combined with the ball physics I'm using) is really separating the good teams from the bad. I played vs Man City and the had 88% pass completion and then I can play Burnley and they might have 74% or so. I've also just increased the shot decision distance as well so this should help from now on to get the AI shooting from a little more distance but I've only tested on a game or so with good results.

A couple of the changes I made today was very slightly upping the defensive workrate and slowing down the turning/change direction speed at full pace. The big ones for today though - I significantly reduced 'bad support range', 'defender loses his attacker' and 'defender bad commit to ball'. I think these values are really impactful as they're programmed for very quick, arcade matches where big holes in the defence open up.

I haven't quite figured out the midfielder into the box as I'd like, and the pressing from teams like City is still a bit stupid, but with the combination of adjustments I'm not finding it to be anywhere near as bad as the defenders are tracking much, much better and are generally more aware. I'm not finding I can just split them open at will (a combination of things I think including the pass error). So whilst it's not perfect I'm pretty happy with where it's at for now and I'll try and put as much as possible into one project for people to try soon (or I'll post 15 files and you can separate them how you'd like). I'll probably still try and improve it but I might actually start playing the game for a while now haha.