Pales
Registered
Hi everyone,
Before making this thread, I reached out to several experienced modders in our community as well as 397/MSG staff to gather their insights and advice. Understandably, they're all very busy with their own commitments and projects. So, I've decided to create this central thread for all of us to hopefully communicate and contribute our knowledge and experiences.
The goal of this thread is essentially to delve into the physics parameters necessary to achieve consistent AI (which is what I believe plays a significant role in the reduction of oversteer and understeer for them). In essence, I am hoping by pooling our collective expertise, we can hopefully develop a understanding and possibly a set of universal parameters that can be applied (should it work that way).
I've noticed, as I’m sure many of you have, that a lot of the rFactor 2 modded content falls short due to the AI, and if we can retrospectively apply these fixes, it would restore a lot of life for the single-player modding scene.
What I know so far:
- When observing the GTE/GT3 official content, the AI do not suffer with understeer or oversteer issues (in the sense that they do not turn the wheel 360 degrees for small corners). This ultimately means that they're more consistent and do not suffer time loss. When I approached Coutie at Studio 397 to see how this is achieved, I was advised that it is largely down to the physics of the car but at this stage, it is unclear what parameters play a significant role for the AI. Apparently the physics guy may release a blog, but given how busy they are with LMU, I am not sure if this will ever materialise
-I reached out to Nick9320 regarding his workshop content that was recently released (noticeably the Ferrari and Porsche), both of which are insanely well optimized for the AI. No understeer or oversteer issues, very consistent, and he also implied it was largely physics related as well
- I am well aware that adjustments to the .JSON file can rectify this to some extent, as well as adjustments to the .TBC for the relevant car, but again, I think the nitty gritty of the physics parameters is largely where the consistency comes from, and it is trying to understand where that is.
Here is an example of poor AI (it is to the extreme, but observe the wheel rotation). In this example, it causes the car to slide but in other mods, they may not suffer as badly as this, but still suffer significant timeloss as result of the understeer /oversteer behavior.
I would appreciate any guidance on this.
Before making this thread, I reached out to several experienced modders in our community as well as 397/MSG staff to gather their insights and advice. Understandably, they're all very busy with their own commitments and projects. So, I've decided to create this central thread for all of us to hopefully communicate and contribute our knowledge and experiences.
The goal of this thread is essentially to delve into the physics parameters necessary to achieve consistent AI (which is what I believe plays a significant role in the reduction of oversteer and understeer for them). In essence, I am hoping by pooling our collective expertise, we can hopefully develop a understanding and possibly a set of universal parameters that can be applied (should it work that way).
I've noticed, as I’m sure many of you have, that a lot of the rFactor 2 modded content falls short due to the AI, and if we can retrospectively apply these fixes, it would restore a lot of life for the single-player modding scene.
What I know so far:
- When observing the GTE/GT3 official content, the AI do not suffer with understeer or oversteer issues (in the sense that they do not turn the wheel 360 degrees for small corners). This ultimately means that they're more consistent and do not suffer time loss. When I approached Coutie at Studio 397 to see how this is achieved, I was advised that it is largely down to the physics of the car but at this stage, it is unclear what parameters play a significant role for the AI. Apparently the physics guy may release a blog, but given how busy they are with LMU, I am not sure if this will ever materialise
-I reached out to Nick9320 regarding his workshop content that was recently released (noticeably the Ferrari and Porsche), both of which are insanely well optimized for the AI. No understeer or oversteer issues, very consistent, and he also implied it was largely physics related as well
- I am well aware that adjustments to the .JSON file can rectify this to some extent, as well as adjustments to the .TBC for the relevant car, but again, I think the nitty gritty of the physics parameters is largely where the consistency comes from, and it is trying to understand where that is.
Here is an example of poor AI (it is to the extreme, but observe the wheel rotation). In this example, it causes the car to slide but in other mods, they may not suffer as badly as this, but still suffer significant timeloss as result of the understeer /oversteer behavior.
I would appreciate any guidance on this.
Last edited: