Disclaimer: this guide reflects my personal opinions and is due to my passion for pCARS 2 as endurance player in VR. I am aware that what is reported may be different from the opinions of other simmers.
FFB
In this section I have no specific suggestions to make, other than avoiding the stock FFBs. In my opinion, the main problem with PC2 is not the physics engine. The SETA for the tires is one of the most complex and realistic systems and the goodness of the Madness Engine is clearly visible in AMS2. However, it is true that the control of the car is sometimes unnatural or imperfect.
This is particularly true when driving around corners: sometimes (not always, it depends on the car and the track) it seems to lose the sensitivity of the weight of the car and, more generally, of its control. An emblematic case occurs when driving at Le Mans, in particular when exiting the second chicane of the Hunaudieres. When the car passes from left to right to exit the chicane, there is no longer a feeling with the car, it seems that there is a gap in the FFB which resumes immediately after.
Another problem concerns the curbs. PC2 has a large variety of curbs, with specific characteristics. Yet, especially the larger ones, they can cause excessive / unnatural effects on driving. A classic case concerns the curbs on the inside of the turn, which if hit, determine an “anchor” or “glue” effect (the car seems to stick to the curb with the front wheel and it is violently pushed towards the inside of the turn, with a marked oversteer). Similar problem also with the curbs on the outside of the turn: sometimes a pendulum effect occurs, with the car being pushed out of the track instead of "bouncing" towards the track as expected.
All of these are quite serious problems, but they can easily be overcome using a custom FFB. These FFBs know how to adequately manage the different forces, ensuring that the car is always under control (eg. no more gaps in the FFB) and that the forces are well balanced with each other, without any conflicting effects such as those on the curbs. There are many and the quality depends a lot on the steering wheel used and obviously on personal tastes. I personally choose PopsRacerFFB and the feeling with PC2 is similar to AMS2.
Therefore, to evaluate the validity of the FFB I suggest to carry out a test in these conditions:
- LMP1, check the feeling at the second chicane of the Hunaudieres. Pay attention to the variation in weight distribution in the right-left movement of the car.
- LMP2, check the feeling at the turns 3-6 of COTA walking along the curbs. Evaluate the glue effect in the internal curbs.
- LMP2, check the feeling at the last turn of Road America. Evaluate the pendulum effect while traveling on the curb.
VR SETTINGS
The Madness Engine is probably the best graphics engine for VR management. Obviously, this does not mean that it is possible to set everything to the maximum: the general criterion is that a lot depends on the resolution of VR headset. We assume that at the time of the release of the latest DLC the most common reference VR headset was the Oculus Rift S and one of the most performing graphics cards was the RTX 2080. In fact, with this combination and using the native resolution without SS, it is possible to have a good framerate with settings close to the maximum. In my opinion it is useful to start from this consideration to adapt PC2 to your hardware.
Some general considerations
- Of course everyone has their own tastes, and there are those who prefer to lower the details a lot in favor of the resolution and vice versa. In my opinion the level of detail should always be kept quite high, as it represents the expected level for that game. For example, some say that one of the first things to do is to eliminate the detail of the grass. Personally I disagree, as if it is true that grass is not essential, it is still an element that can be seen (we are not in a flight simulator, where it would be visible only when departing and landing) and which increases the sensation of realism typical of PC2. Furthermore, its elimination is not decisive, as well as the strong reduction of detail of a single parameter, with the exception of environmental map and shadows. In general terms, only environmental map and shadows (and MSAA, headset and SS resolution) are, by themselves, decisive: for the other parameters there is an overall effect to consider.
- The setting should guarantee an acceptable framerate even in the worst conditions (for PC2 they are represented by dawn, rain and night). Of course, if the player only runs during the day without bad weather, it is possible to raise (even significantly) the video quality (e.g. increasing by 0.1 SS).
- While confirming that the goal is to maintain a high level of detail close to the maximum potential of the game, we must consider that the resolution perceived in VR is generally lower than the monitor. In other words, particularly high levels of detail may not be noticeable to the eye, especially when in motion. This is the case of two of the most impacting factors on fps: environmental map and shadows. The difference between ultra and high is minimal, practically not perceptible for environmental map, while it is visible for shadows, but this detail is lost in movement. On the other hand, further lowering causes an appreciable visual deterioration (eg. at medium level not only the detail of the shadows decreases, but their quantity and whether or not they are represented in the mirrors).
- Use VR headroom (see also following tips on ASW) as reference for your settings. During the starting it should be less than -40%, as in critical conditions (sunset, rain, night) this value could double.
Suggested settings
- Assuming that an RTX 2080 or a 1080Ti are the entry level cards for PC2 in VR, I suggest setting all the parameters at “high”. At the same time set the texture filter to triliner, and disable enhance mirror and motion blur. Render frame should be set to 1.
- MSAA is the best AA filter in PC2. It also fits very well with VRperfKit's NIS upscaling function (see later), resulting in an excellent balance between detail and softness. Thanks to the blur generated by upscaling, it is sufficient to set the MSAA to low, as higher values would only lead to excessive blurring. Furthermore, MSAA is particularly impacting on fps, so just set it to low.
- There are many conflicting opinions on retroprojection. Obviously if there are no fps / headroom problems, the best thing is not to use any settings that limit fps or act on ASW. However, it is highly unlikely that this will happen, as on very performing systems you will tend to push a little on the SS or have very high-resolution headset. My suggestion is to use OTT (or equivalent for other headset) and set the ASW to 45 (not forced). PC2 seems to be well optimized to work in this mode, ensuring excellent fluidity and consistency.
- Use VRperfKit, it is very useful even for high-end systems. This is particularly useful for nVIDIA users, thanks to NIS option. I suggest to leave upscaling as default (90%) and set NIS as method. Than, just put to 0 the sharpening parameters, otherwise you will ruin your AA setting. Another useful tool is the fixed FOV, even if it doesn’t work well in PC2, because it simply cuts polys at the edges of your screen. To avoid this, you have to set the inner circle to 1.1/1.2: obviously in this way the bulk of the benefit is lost, as practically all the view remains in native resolution. Despite this, the fps gains slightly, especially during sunset (perhaps there are out-of-view advantages on which the lighting effects impact equally).
- All the above should work fine on the reference system indicated (2080 + Rift S) with SS = 1. Of course, the scenario is different if you have a headset with a higher resolution (eg. Reverb G2), as the resolution is the most impactful parameter on the whole system. It may not be very good less intense conditions (eg. driving during the day in good weather) but the effects in critical situations (eg. starting, night, rain, sunset) are exponential. Therefore it is advisable to set the SS in such a way as to reach a resolution similar to that of the Oculus Rift S (e.g. about 0.7 for Reverb G2), and then gradually raise it according to your video card, remembering not to exceed -40 headroom at the start.
Suggested settings for Oculus users
- If you are an Oculus user, you must edit the file graphicsconfigoculusdx11.xml as follow:
<prop name="SharpeningStrength" sharpeningstrength="1.300000" />
<prop name="SharpeningClamp" sharpeningclamp="0.080000" />
<prop name="OffsetBias" offsetbias="1.300000" />
The visible result is a much higher level of image detail with blur removal, similar to SS but with a low impact on performance. I am not aware of similar settings for other types of VR headset.
- Set OTT as reported above.
MODES AND SETTING
Weather
As is known, the randomized time selection system does not produce realistic weather variability. In fact, the probability of rain is identical to the absence of rain. Therefore by selecting the random mode, you will find yourself running very frequently in the rain or in extreme conditions. The only solution is to use the CustomRandomWeatherAutomator tool. It is an external tool that automatically selects the weather in the various weather slots, making it less likely that rain will occur. The limit is that the player will know the progress of the weather before the race, but in fact puts him in the same condition as the AI.
If you select the weather manually or use the Automator tool, I suggest to avoid "overcast". It is known to cause a big drop in fps (use "heavy clouds" instead of "overcast").
Custom grids
The lack of custom grid is one of the main issues for single player racers. Fortunately, there is an external tool (Custom_Grid_Tool2) able to select every single car (even between different classes) that will make up the starting grid. Unfortunately it is not possible (as in AMS2) to edit the performance of the AI, so if you insert some identical cars (eg. five ORECA 03R) the performance differences between the cars will be random from race to race. Furthermore, it works only in single race mode. However, for those with a bit of time to waste, it is possible to save the custom grids and use them to build a championship by hand. In other words, since the custom grid only works with single races, it is possible to manually write down the results of race 1 on an excel sheet (e.g. the cars that scored points) and then proceed with a race 2 using the same grid. In this way it is possible to play a championship with the same cars on the grid.
AI SETTINGS
There are several interesting guides for setting up AI (eg. www.youtube.com/watch?v=LfCbw0TGbnY). In my opinion, I agree with those who say that the player must somehow adapt to the AI, as no game responds perfectly to every type of player. This guide is primarily intended for those who prefer the endurance side of PC2, and should allow for a good level of challenge and realism.
Aggressiveness is the key
- Set aggressiveness to 100. Aggression does not significantly affect AI errors (just a bit at starts). Even the actual tendency to overtake is not particularly encouraged. Otherwise, aggression represents the consistency with which the AI takes on all turns, braking, overtaking. If the aggression is around 50, the AI will do a few laps at maximum speed (depending on the strength setting) but most of the laps will be 1-2s (depending on the car) slower than reference time. Otherwise, at 100, the AI will always make constant times, very close to the maximum performance. In addition, the slowdowns due to overtaking lapped cars in multi-class races will be reduced, in line with the player (with 50, the AI will tend to waste a lot of time once it gets close to a group of lapped cars).
- Aggressiveness to 100 is also essential for rain conditions. During the rain, AI tends to be particularly cautious, and therefore slow. At 100, the speed is acceptable, although it depends a lot on the circuit and the intensity of rain.
- Strength represents the maximum possible performance of AI. It is very important to calibrate it properly. The cause of the ramming is closely related to the strength of the AI. If the AI is too fast (or if it is much faster than you) it will tend to brake better than the player or overtake in risky place, hitting the player from the back (ramming). In fact, if the AI is stably slower than the player (e.g. about 1s), it is highly unlikely that it will ram the player. The problem is that very often we tend to set low aggression to avoid ramming or risky moves, but this lowers the average AI times and traffic management. Therefore, the player tends to raise the strength of the AI, rebalancing the AI average times but making the AI too strong on the single lap (and on the single braking), increasing the risk of ramming. Ultimately, to avoid rear-end collisions, the AI must be more or less fast like the player, otherwise the braking point will be more advanced and the distance with the car in front (that of the player) will be less, making the rear-end collision likely. So, what values should the AI strength be set to? See qualify tips.
Ramming
Despite the adjustments for AI, ramming is not completely avoidable. However, it is strongly linked to specific turns, and it is advisable to give way to the AI if it is decidedly faster than us in those particular zones of the track. A positive element of ramming in PC2 is that it generally does not alter the player's guidance. In other words, it rarely causes off the track (eg. as happens right before the "Esses" of Indianapolis at Le Mans, a very critical point where it is necessary to pay close attention to the AI that follows us) and thanks to the different physics of the AI, it generally does not involve significant damage to the AI. However, it can happen, especially when the ramming is angular and depending on the shape of the car, that it determines the detachment of the rear wing, with obvious consequences. This problem, even in other simulators, is overcome by selecting the option of no damage to the car. This solution is found in numerous posts also for other simulators, as it solves the problem of playability at the root. Yet it may be unwanted, so an alternative may be to make only the rear wing indestructible. In fact, most of the accidents made by the player cause (or would cause) damage to the front of the car, while damage to the rear wing is quite rare (only following a spin and subsequent collision with the barriers if they impact on the rear and only if the rear wing protrudes from the car).
It is possible to make only the rear wing indestructible by editing the file ”componentdamage.xml”, increasing damage and remove parameters (eg. <entry substring="LIGIER_JSP2_REARWING" damaged="1.00" removed="1.00" />). To find/edit this file you need to have a modded version of PC2 (the patch 7 livery access). Here a link for an edited file for most prototypes and some GT3 https://mega.nz/file/HhhmiSBQ#_ErvTXDyWJXVuOqIW-Btbp4Iit8vbEWpnbLZNJ4CCl4
Qualify
- The AI uses a simplified physical system, and does not distinguish between qualification and race. Basically, it is as if he were using the race set-up also in qualifying. Therefore, the player should do the same, otherwise if the AI is in the same performance window of you in qualifying, then it will be too fast in the race.
- The qualify is essential for setting the strength of the AI. For an experienced player, the goal is to be classified (with a good lap, not a dream lap) within the top 3-5 positions. If you are faster than the best AI by 0.5-1 second, raise the AI strength by 5 points. If you are about 1.5-2 seconds, raise it by 10. Follow the same approach if you are slower than AI.
- If you drive the Audi R18 2016, AI strength should be always rise by 10 points, otherwise you will be faster than AI.
Race length
- Multiples of 35 minutes represent ideal periods for setting the duration of a race. A 1h10’ race includes 1 stop for the player and for the AI (or zero for both, in certain GT3 races, for example). A 2h20' race includes 2-3 stops for cars with large gasoline load capacity (eg GT3, old prototypes), or 3-5 stops for other cars.
Specific AI for tracks
For a player with many hours of experience, using non-extreme setups, the optimal AI should be around 100-110, but each one has its own reference, even in function of the cars. Different classes result in different AI performance, so it's about making small corrections around your reference value. There are, however, races generally classified as difficult (eg. Silverstone, COTA) and others as easy (eg. Laguna Seca, Sonoma). Here are some suggestions.
Barcelona
110 (LMP3)
90 (GT3)
Bathurst
90 (GT3)
Brands
120 (GrC)
Brno
90 (LMP900)
COTA
110 (LMP3)
100 (LMP1, LMP2)
90 (GT3)
Daytona
110 (GrC)
105 (LMP2)
100 (LMP3, LMP900, GTP)
Donington
105 (LMP900)
Dubai
110 (LMP1)
Fuji
120 (GrC)
110 (LMP1)
105 (LMP2)
Hockenheimring
120 (GrC)
100 (GT3)
Imola
110 (LMP2, LMP3)
100 (GT3)
Indianapolis
105 (LMP2)
Laguna Seca
115 (LMP2, LMP3)
110 (LMP900, GT3)
Le Mans
105 (LMP1, LMP900)
Le Mans Bugatti
115 (LMP3)
Le Mans classic
120 (GrC, Vintage Prot A)
100 (Gr6)
Long Beach
115 (GrC)
110 (LMP2)
105 (LMP900, GTP)
Monza
120 (GrC)
105 (LMP900)
100 (LMP2, GT3)
Monza classic
110 (Vintage Prot A)
100 (Gr6)
Mugello
110 (LMP3)
Nurburgring
120 (GrC)
110 (LMP1)
100 (GT3)
Nurburgring Nordschleife
90 (GT3)
Oschersleben
120 (GrC)
110 (LMP900)
105 (GT3)
Portimao
110 (LMP3)
100 (LMP2)
Red Bull Ring
100 (LMP2, GT3)
Road America
100 (LMP3, GrC)
90 (LMP2, LMP900, GT3, GTP)
Ruapuna
110 (LMP2)
Silverstone
100 (LMP1, LMP2)
90 (GT3)
Silverstone classic
110 (Vintage Prot A)
100 (Gr6)
Sonoma
115 (LMP2)
110 (GT3)
Spa
120 (GrC)
110 (LMP1, LMP2)
105 (LMP900)
100 (GT3)
Spa classic
100 (Vintage Prot A)
90 (Gr6)
Sugo
120 (GrC)
110 (LMP2)
Suzuka
120 (GrC)
105 (LMP2)
Texas Speedway
100 (GrC, LMP900, GTP)
Watkins Glen
105 (LMP2)
100 (GT3, GrC, LMP3)
95 (LMP900)
90 (GTP)
Willow Springs
80 (GrC, GTP)
Zhuhai
110 (LMP1, LMP2)
Zolder
120 (GrC)
115 (LMP3)
FFB
In this section I have no specific suggestions to make, other than avoiding the stock FFBs. In my opinion, the main problem with PC2 is not the physics engine. The SETA for the tires is one of the most complex and realistic systems and the goodness of the Madness Engine is clearly visible in AMS2. However, it is true that the control of the car is sometimes unnatural or imperfect.
This is particularly true when driving around corners: sometimes (not always, it depends on the car and the track) it seems to lose the sensitivity of the weight of the car and, more generally, of its control. An emblematic case occurs when driving at Le Mans, in particular when exiting the second chicane of the Hunaudieres. When the car passes from left to right to exit the chicane, there is no longer a feeling with the car, it seems that there is a gap in the FFB which resumes immediately after.
Another problem concerns the curbs. PC2 has a large variety of curbs, with specific characteristics. Yet, especially the larger ones, they can cause excessive / unnatural effects on driving. A classic case concerns the curbs on the inside of the turn, which if hit, determine an “anchor” or “glue” effect (the car seems to stick to the curb with the front wheel and it is violently pushed towards the inside of the turn, with a marked oversteer). Similar problem also with the curbs on the outside of the turn: sometimes a pendulum effect occurs, with the car being pushed out of the track instead of "bouncing" towards the track as expected.
All of these are quite serious problems, but they can easily be overcome using a custom FFB. These FFBs know how to adequately manage the different forces, ensuring that the car is always under control (eg. no more gaps in the FFB) and that the forces are well balanced with each other, without any conflicting effects such as those on the curbs. There are many and the quality depends a lot on the steering wheel used and obviously on personal tastes. I personally choose PopsRacerFFB and the feeling with PC2 is similar to AMS2.
Therefore, to evaluate the validity of the FFB I suggest to carry out a test in these conditions:
- LMP1, check the feeling at the second chicane of the Hunaudieres. Pay attention to the variation in weight distribution in the right-left movement of the car.
- LMP2, check the feeling at the turns 3-6 of COTA walking along the curbs. Evaluate the glue effect in the internal curbs.
- LMP2, check the feeling at the last turn of Road America. Evaluate the pendulum effect while traveling on the curb.
VR SETTINGS
The Madness Engine is probably the best graphics engine for VR management. Obviously, this does not mean that it is possible to set everything to the maximum: the general criterion is that a lot depends on the resolution of VR headset. We assume that at the time of the release of the latest DLC the most common reference VR headset was the Oculus Rift S and one of the most performing graphics cards was the RTX 2080. In fact, with this combination and using the native resolution without SS, it is possible to have a good framerate with settings close to the maximum. In my opinion it is useful to start from this consideration to adapt PC2 to your hardware.
Some general considerations
- Of course everyone has their own tastes, and there are those who prefer to lower the details a lot in favor of the resolution and vice versa. In my opinion the level of detail should always be kept quite high, as it represents the expected level for that game. For example, some say that one of the first things to do is to eliminate the detail of the grass. Personally I disagree, as if it is true that grass is not essential, it is still an element that can be seen (we are not in a flight simulator, where it would be visible only when departing and landing) and which increases the sensation of realism typical of PC2. Furthermore, its elimination is not decisive, as well as the strong reduction of detail of a single parameter, with the exception of environmental map and shadows. In general terms, only environmental map and shadows (and MSAA, headset and SS resolution) are, by themselves, decisive: for the other parameters there is an overall effect to consider.
- The setting should guarantee an acceptable framerate even in the worst conditions (for PC2 they are represented by dawn, rain and night). Of course, if the player only runs during the day without bad weather, it is possible to raise (even significantly) the video quality (e.g. increasing by 0.1 SS).
- While confirming that the goal is to maintain a high level of detail close to the maximum potential of the game, we must consider that the resolution perceived in VR is generally lower than the monitor. In other words, particularly high levels of detail may not be noticeable to the eye, especially when in motion. This is the case of two of the most impacting factors on fps: environmental map and shadows. The difference between ultra and high is minimal, practically not perceptible for environmental map, while it is visible for shadows, but this detail is lost in movement. On the other hand, further lowering causes an appreciable visual deterioration (eg. at medium level not only the detail of the shadows decreases, but their quantity and whether or not they are represented in the mirrors).
- Use VR headroom (see also following tips on ASW) as reference for your settings. During the starting it should be less than -40%, as in critical conditions (sunset, rain, night) this value could double.
Suggested settings
- Assuming that an RTX 2080 or a 1080Ti are the entry level cards for PC2 in VR, I suggest setting all the parameters at “high”. At the same time set the texture filter to triliner, and disable enhance mirror and motion blur. Render frame should be set to 1.
- MSAA is the best AA filter in PC2. It also fits very well with VRperfKit's NIS upscaling function (see later), resulting in an excellent balance between detail and softness. Thanks to the blur generated by upscaling, it is sufficient to set the MSAA to low, as higher values would only lead to excessive blurring. Furthermore, MSAA is particularly impacting on fps, so just set it to low.
- There are many conflicting opinions on retroprojection. Obviously if there are no fps / headroom problems, the best thing is not to use any settings that limit fps or act on ASW. However, it is highly unlikely that this will happen, as on very performing systems you will tend to push a little on the SS or have very high-resolution headset. My suggestion is to use OTT (or equivalent for other headset) and set the ASW to 45 (not forced). PC2 seems to be well optimized to work in this mode, ensuring excellent fluidity and consistency.
- Use VRperfKit, it is very useful even for high-end systems. This is particularly useful for nVIDIA users, thanks to NIS option. I suggest to leave upscaling as default (90%) and set NIS as method. Than, just put to 0 the sharpening parameters, otherwise you will ruin your AA setting. Another useful tool is the fixed FOV, even if it doesn’t work well in PC2, because it simply cuts polys at the edges of your screen. To avoid this, you have to set the inner circle to 1.1/1.2: obviously in this way the bulk of the benefit is lost, as practically all the view remains in native resolution. Despite this, the fps gains slightly, especially during sunset (perhaps there are out-of-view advantages on which the lighting effects impact equally).
- All the above should work fine on the reference system indicated (2080 + Rift S) with SS = 1. Of course, the scenario is different if you have a headset with a higher resolution (eg. Reverb G2), as the resolution is the most impactful parameter on the whole system. It may not be very good less intense conditions (eg. driving during the day in good weather) but the effects in critical situations (eg. starting, night, rain, sunset) are exponential. Therefore it is advisable to set the SS in such a way as to reach a resolution similar to that of the Oculus Rift S (e.g. about 0.7 for Reverb G2), and then gradually raise it according to your video card, remembering not to exceed -40 headroom at the start.
Suggested settings for Oculus users
- If you are an Oculus user, you must edit the file graphicsconfigoculusdx11.xml as follow:
<prop name="SharpeningStrength" sharpeningstrength="1.300000" />
<prop name="SharpeningClamp" sharpeningclamp="0.080000" />
<prop name="OffsetBias" offsetbias="1.300000" />
The visible result is a much higher level of image detail with blur removal, similar to SS but with a low impact on performance. I am not aware of similar settings for other types of VR headset.
- Set OTT as reported above.
MODES AND SETTING
Weather
As is known, the randomized time selection system does not produce realistic weather variability. In fact, the probability of rain is identical to the absence of rain. Therefore by selecting the random mode, you will find yourself running very frequently in the rain or in extreme conditions. The only solution is to use the CustomRandomWeatherAutomator tool. It is an external tool that automatically selects the weather in the various weather slots, making it less likely that rain will occur. The limit is that the player will know the progress of the weather before the race, but in fact puts him in the same condition as the AI.
If you select the weather manually or use the Automator tool, I suggest to avoid "overcast". It is known to cause a big drop in fps (use "heavy clouds" instead of "overcast").
Custom grids
The lack of custom grid is one of the main issues for single player racers. Fortunately, there is an external tool (Custom_Grid_Tool2) able to select every single car (even between different classes) that will make up the starting grid. Unfortunately it is not possible (as in AMS2) to edit the performance of the AI, so if you insert some identical cars (eg. five ORECA 03R) the performance differences between the cars will be random from race to race. Furthermore, it works only in single race mode. However, for those with a bit of time to waste, it is possible to save the custom grids and use them to build a championship by hand. In other words, since the custom grid only works with single races, it is possible to manually write down the results of race 1 on an excel sheet (e.g. the cars that scored points) and then proceed with a race 2 using the same grid. In this way it is possible to play a championship with the same cars on the grid.
AI SETTINGS
There are several interesting guides for setting up AI (eg. www.youtube.com/watch?v=LfCbw0TGbnY). In my opinion, I agree with those who say that the player must somehow adapt to the AI, as no game responds perfectly to every type of player. This guide is primarily intended for those who prefer the endurance side of PC2, and should allow for a good level of challenge and realism.
Aggressiveness is the key
- Set aggressiveness to 100. Aggression does not significantly affect AI errors (just a bit at starts). Even the actual tendency to overtake is not particularly encouraged. Otherwise, aggression represents the consistency with which the AI takes on all turns, braking, overtaking. If the aggression is around 50, the AI will do a few laps at maximum speed (depending on the strength setting) but most of the laps will be 1-2s (depending on the car) slower than reference time. Otherwise, at 100, the AI will always make constant times, very close to the maximum performance. In addition, the slowdowns due to overtaking lapped cars in multi-class races will be reduced, in line with the player (with 50, the AI will tend to waste a lot of time once it gets close to a group of lapped cars).
- Aggressiveness to 100 is also essential for rain conditions. During the rain, AI tends to be particularly cautious, and therefore slow. At 100, the speed is acceptable, although it depends a lot on the circuit and the intensity of rain.
- Strength represents the maximum possible performance of AI. It is very important to calibrate it properly. The cause of the ramming is closely related to the strength of the AI. If the AI is too fast (or if it is much faster than you) it will tend to brake better than the player or overtake in risky place, hitting the player from the back (ramming). In fact, if the AI is stably slower than the player (e.g. about 1s), it is highly unlikely that it will ram the player. The problem is that very often we tend to set low aggression to avoid ramming or risky moves, but this lowers the average AI times and traffic management. Therefore, the player tends to raise the strength of the AI, rebalancing the AI average times but making the AI too strong on the single lap (and on the single braking), increasing the risk of ramming. Ultimately, to avoid rear-end collisions, the AI must be more or less fast like the player, otherwise the braking point will be more advanced and the distance with the car in front (that of the player) will be less, making the rear-end collision likely. So, what values should the AI strength be set to? See qualify tips.
Ramming
Despite the adjustments for AI, ramming is not completely avoidable. However, it is strongly linked to specific turns, and it is advisable to give way to the AI if it is decidedly faster than us in those particular zones of the track. A positive element of ramming in PC2 is that it generally does not alter the player's guidance. In other words, it rarely causes off the track (eg. as happens right before the "Esses" of Indianapolis at Le Mans, a very critical point where it is necessary to pay close attention to the AI that follows us) and thanks to the different physics of the AI, it generally does not involve significant damage to the AI. However, it can happen, especially when the ramming is angular and depending on the shape of the car, that it determines the detachment of the rear wing, with obvious consequences. This problem, even in other simulators, is overcome by selecting the option of no damage to the car. This solution is found in numerous posts also for other simulators, as it solves the problem of playability at the root. Yet it may be unwanted, so an alternative may be to make only the rear wing indestructible. In fact, most of the accidents made by the player cause (or would cause) damage to the front of the car, while damage to the rear wing is quite rare (only following a spin and subsequent collision with the barriers if they impact on the rear and only if the rear wing protrudes from the car).
It is possible to make only the rear wing indestructible by editing the file ”componentdamage.xml”, increasing damage and remove parameters (eg. <entry substring="LIGIER_JSP2_REARWING" damaged="1.00" removed="1.00" />). To find/edit this file you need to have a modded version of PC2 (the patch 7 livery access). Here a link for an edited file for most prototypes and some GT3 https://mega.nz/file/HhhmiSBQ#_ErvTXDyWJXVuOqIW-Btbp4Iit8vbEWpnbLZNJ4CCl4
Qualify
- The AI uses a simplified physical system, and does not distinguish between qualification and race. Basically, it is as if he were using the race set-up also in qualifying. Therefore, the player should do the same, otherwise if the AI is in the same performance window of you in qualifying, then it will be too fast in the race.
- The qualify is essential for setting the strength of the AI. For an experienced player, the goal is to be classified (with a good lap, not a dream lap) within the top 3-5 positions. If you are faster than the best AI by 0.5-1 second, raise the AI strength by 5 points. If you are about 1.5-2 seconds, raise it by 10. Follow the same approach if you are slower than AI.
- If you drive the Audi R18 2016, AI strength should be always rise by 10 points, otherwise you will be faster than AI.
Race length
- Multiples of 35 minutes represent ideal periods for setting the duration of a race. A 1h10’ race includes 1 stop for the player and for the AI (or zero for both, in certain GT3 races, for example). A 2h20' race includes 2-3 stops for cars with large gasoline load capacity (eg GT3, old prototypes), or 3-5 stops for other cars.
Specific AI for tracks
For a player with many hours of experience, using non-extreme setups, the optimal AI should be around 100-110, but each one has its own reference, even in function of the cars. Different classes result in different AI performance, so it's about making small corrections around your reference value. There are, however, races generally classified as difficult (eg. Silverstone, COTA) and others as easy (eg. Laguna Seca, Sonoma). Here are some suggestions.
Barcelona
110 (LMP3)
90 (GT3)
Bathurst
90 (GT3)
Brands
120 (GrC)
Brno
90 (LMP900)
COTA
110 (LMP3)
100 (LMP1, LMP2)
90 (GT3)
Daytona
110 (GrC)
105 (LMP2)
100 (LMP3, LMP900, GTP)
Donington
105 (LMP900)
Dubai
110 (LMP1)
Fuji
120 (GrC)
110 (LMP1)
105 (LMP2)
Hockenheimring
120 (GrC)
100 (GT3)
Imola
110 (LMP2, LMP3)
100 (GT3)
Indianapolis
105 (LMP2)
Laguna Seca
115 (LMP2, LMP3)
110 (LMP900, GT3)
Le Mans
105 (LMP1, LMP900)
Le Mans Bugatti
115 (LMP3)
Le Mans classic
120 (GrC, Vintage Prot A)
100 (Gr6)
Long Beach
115 (GrC)
110 (LMP2)
105 (LMP900, GTP)
Monza
120 (GrC)
105 (LMP900)
100 (LMP2, GT3)
Monza classic
110 (Vintage Prot A)
100 (Gr6)
Mugello
110 (LMP3)
Nurburgring
120 (GrC)
110 (LMP1)
100 (GT3)
Nurburgring Nordschleife
90 (GT3)
Oschersleben
120 (GrC)
110 (LMP900)
105 (GT3)
Portimao
110 (LMP3)
100 (LMP2)
Red Bull Ring
100 (LMP2, GT3)
Road America
100 (LMP3, GrC)
90 (LMP2, LMP900, GT3, GTP)
Ruapuna
110 (LMP2)
Silverstone
100 (LMP1, LMP2)
90 (GT3)
Silverstone classic
110 (Vintage Prot A)
100 (Gr6)
Sonoma
115 (LMP2)
110 (GT3)
Spa
120 (GrC)
110 (LMP1, LMP2)
105 (LMP900)
100 (GT3)
Spa classic
100 (Vintage Prot A)
90 (Gr6)
Sugo
120 (GrC)
110 (LMP2)
Suzuka
120 (GrC)
105 (LMP2)
Texas Speedway
100 (GrC, LMP900, GTP)
Watkins Glen
105 (LMP2)
100 (GT3, GrC, LMP3)
95 (LMP900)
90 (GTP)
Willow Springs
80 (GrC, GTP)
Zhuhai
110 (LMP1, LMP2)
Zolder
120 (GrC)
115 (LMP3)
Last edited:
![Luca [OT]](/data/avatars/s/3235/3235113.jpg?1697196734){kind=avatar}
