On bikes they crash brutally on the high speed kink style corners on West Hill and the start finish at Blackwood on motorcycles for me. Seems as though they come in way too fast and start to steer way too late. If I lower them to level 2 or less skill they stay on track but they're hilariously slow elsewhere.
On bikes they crash brutally on the high speed kink style corners on West Hill and the start finish at Blackwood on motorcycles for me. Seems as though they come in way too fast and start to steer way too late. If I lower them to level 2 or less skill they stay on track but they're hilariously slow elsewhere.
OK, the pit stop issue is definitely fixed. Thanks for the reproduction but this one is too unreliable to use as a proof of fix. I couldn't reproduce it (using default setup). I would need something like 100% reproduction method, hopefully for a single car on a short race. Then, if it didn't happen in my test version we'd know for a fact it is fixed. But I'm not asking you to keep trying it, just let me know if you happen come across one where it happens every time.
I was asking specifically about problems joining and leaving the pit lane path, but I think you are talking about ordinary corners?
If you are talking about bikes failing at corners, please state the exact bike and which config, as all the bikes I have been testing with, do not crash on any configs I know about.
Yeah, I did use a custom setup, for AI users ( AS4 ). It is like they are on throttle too long and attempting for perfect limit, which they fail on certain instances.
Hmmh... Of course, removing whole pit speed limit completely would not cause issues, or removing penalties ( which you can do ), but that is completely different story.
I am certain I will find a proper proof ( not soon I believe ), as what I believe is purely settings oriented, where AI seems not to handle certain settings enough sufficiently.
EDIT: Just in case IF this MAY be true, I added a setup here, it is happening exactly on this setup. ( Dont mind the name, I have so many stuff and setups that I had to rename it like this ) ( And yes, same car, Formula US 93, on Combo of KY3R )
It may happen on different courses as well, as well as FE4 IIRC
Hmmh... Of course, removing whole pit speed limit completely would not cause issues, or removing penalties ( which you can do ), but that is completely different story.
I am certain I will find a proper proof ( not soon I believe ), as what I believe is purely settings oriented, where AI seems not to handle certain settings enough sufficiently.
EDIT: Just in case IF this MAY be true, I added a setup here, it is happening exactly on this setup. ( Dont mind the name, I have so many stuff and setups that I had to rename it like this ) ( And yes, same car, Formula US 93, on Combo of KY3R )
It may happen on different courses as well, as well as FE4 IIRC
Thank you, that does reproduce the issue reliably, even in my updated patch (add 4 AI to KY3R / 2 laps / mustpit). With your setup they always wheelspin and then it causes the issue described by Michal. So I should be able to fix it by making the AI use the pit lane speed limiter.
I've tested your bike on Westhill and they can't even make past the second turn without sliding off the road (out of 10, 2 crashed on the first turn). Must be issue with insane downforce on that bike.
(Viperakecske)
DELETED
by Scawen : not constructive
(turbofan)
DELETED
by Scawen : reply to post that is now deleted
Downforce on bikes is mostly "no go" I think. It's a problem because it will push not only down into the ground but out of the corner too. Basically it should be zero or in some rare cases there could be a very small amount of "front wing" downforce.
Slashpca, I've had a look at your R6 and R1 now, they have huge downforce, something like 6 times greater than the drag, which is in the region of the downforce achieved by an F1 car, so that's not reasonable. Also, it's impossible for an undertray on a bike to produce downforce. The undertray is really just a large flat area, often enhanced by a rear diffuser. Doesn't really make sense on a bike.
Until recent years I would have thought downforce on bikes was near impossible, but now some race bikes have small winglets at the front which can help keep the front down at high speed. I'm not sure how much benefit they can bring for cornering grip, as I pointed out their force is "out" as well as "down"...
Slashpca, I've had a look at your R6 and R1 now, they have huge downforce, something like 6 times greater than the drag, which is in the region of the downforce achieved by an F1 car, so that's not reasonable. Also, it's impossible for an undertray on a bike to produce downforce. The undertray is really just a large flat area, often enhanced by a rear diffuser. Doesn't really make sense on a bike.
Until recent years I would have thought downforce on bikes was near impossible, but now some race bikes have small winglets at the front which can help keep the front down at high speed. I'm not sure how much benefit they can bring for cornering grip, as I pointed out their force is "out" as well as "down"...
Well most bike are generating lift (like most cars) and it extends to racing motorbikes.
This paper give some real values for racing motorbikes. The little wings on new motoGP are reducing a bit the lift and cleaning the flow for drag removal... 32 pages but pretty interesting!
https://upcommons.upc.edu/bitstream/2117/393241/1/energies-16-04793.pdf
It is still a pretty complex matter with ground effect appearing on leaning bikes and Cl and Cd varying depending on angle of lean and wheel speed...
This paper give some real values for racing motorbikes. The little wings on new motoGP are reducing a bit the lift and cleaning the flow for drag removal... 32 pages but pretty interesting!
https://upcommons.upc.edu/bitstream/2117/393241/1/energies-16-04793.pdf
It is still a pretty complex matter with ground effect appearing on leaning bikes and Cl and Cd varying depending on angle of lean and wheel speed...
Now that I've fixed a lot of pitlane-related AI bugs ready for D47, I finally had a chance to look at (and fix) this command interpretation function.
Also the automatic replay naming codepage bug reported in your D41 bug report PDF.
Thanks for the detailed descriptions. D47 should be available this evening.
aware of this being off topic, but seems like a relatively simple request and i'm not sure how actively the improvements forum is monitored.
can we please have 'right-click, enter value' behaviour for the sliders in colors tab as all the other sliders in the garage?
thanks.
can we please have 'right-click, enter value' behaviour for the sliders in colors tab as all the other sliders in the garage?
thanks.
As the tougher parts of the AI were finished (for now) in D46, I was able to do another small change that had been planned for the incompatible version. The four character version of a mod's name now appears in the results table instead of "MOD". It's an imperfect system in that more than one mod may end up with the same 4-character name, but it is useful as an indication of a mod when space is tight. Of course the algorithm used in LFS was designed to be the same as the one used at LFS World, so hopefully the names are always the same!
After that, I looked back through this thread and found various issues related to AI and pit stops. There were a few reasons for AI to speed in the pit lane, miss pit stops and miss their garage when parking. Also I was able to update the pit speed limiter to be more realistic and more reliable at the same time. Fixed some rarely seen interface bugs too.
Thank you for all the testing and comments!
Changes in Test Patch D47
Interface:
Four-character mod names are shown in results table instead of MOD
FIX: /setlap command error if name coloured and number 4 chars long
FIX: Replays auto-named with special characters could appear wrong
Simulation:
Pit speed limiter now based on drive speed instead of world speed
- prevents wheelspins (e.g. at RO) pushing car over the speed limit
AI:
Avoid excessive downshifting when approaching speed limit zone
Approx 1 second safety margin entering pit lane to avoid speeding
Use speed limiter or throttle to avoid wheelspins causing speeding
Smoother transitions switching between main path and pit lane path
FIX: Some mods would brake too gently and miss the pit stop point
FIX: Some mods would overshoot their pit garage when parking
Download:
https://www.lfs.net/forum/thread/102117
After that, I looked back through this thread and found various issues related to AI and pit stops. There were a few reasons for AI to speed in the pit lane, miss pit stops and miss their garage when parking. Also I was able to update the pit speed limiter to be more realistic and more reliable at the same time. Fixed some rarely seen interface bugs too.
Thank you for all the testing and comments!
Changes in Test Patch D47
Interface:
Four-character mod names are shown in results table instead of MOD
FIX: /setlap command error if name coloured and number 4 chars long
FIX: Replays auto-named with special characters could appear wrong
Simulation:
Pit speed limiter now based on drive speed instead of world speed
- prevents wheelspins (e.g. at RO) pushing car over the speed limit
AI:
Avoid excessive downshifting when approaching speed limit zone
Approx 1 second safety margin entering pit lane to avoid speeding
Use speed limiter or throttle to avoid wheelspins causing speeding
Smoother transitions switching between main path and pit lane path
FIX: Some mods would brake too gently and miss the pit stop point
FIX: Some mods would overshoot their pit garage when parking
Download:
https://www.lfs.net/forum/thread/102117
May I ask what you mean when you reference incompatible version?
(Joe mama1256)
DELETED
by Scawen : no longer needed
I've already done most of the work for an incompatible version, with improved mod support. It's a relatively minor update, an intermediate version still using the old physics and graphics, but has better support for mods.
It's intended to become version 0.7E after a short period of testing. I hope the short period of testing can begin in about two weeks from now if I can finish a few things.
Some of the updates in it were listed here: https://www.lfs.net/forum/thread/104758
(Scawen)
DELETED
by Scawen : no longer needed
Scawen, I found a bug in this newest test patch. It's related to this:
It seems like with mods, their abbreviations on results table may change when someone improves a qualifying lap. I haven't tested, can this be reproduced on races, though. I've attached a SPR about this. You need to download following mods:
- POCHETE 118 GT3
- PROTECH 92 GT3
- LILLIA-X GTR
- RB GTR
- MRG V8 GTR
When you look at the replay, you can see, that at first, it works as intended. But when those cars drive that second timed qualifying lap, it changes another vehicle's mod abbreviation in that results table to something else, whenever somebody improves their best lap. Additionally, game renames those abbreviations to something else (in other words, to an abbreviation, that doesn't exist on any car on current grid).
Slightly OT, but this came to my mind, when I noticed this bug when testing: is there any reason, why that qualifying/best lap/results table is not shown at all during practice sessions (= when race length is 0 laps)?
It seems like with mods, their abbreviations on results table may change when someone improves a qualifying lap. I haven't tested, can this be reproduced on races, though. I've attached a SPR about this. You need to download following mods:
- POCHETE 118 GT3
- PROTECH 92 GT3
- LILLIA-X GTR
- RB GTR
- MRG V8 GTR
When you look at the replay, you can see, that at first, it works as intended. But when those cars drive that second timed qualifying lap, it changes another vehicle's mod abbreviation in that results table to something else, whenever somebody improves their best lap. Additionally, game renames those abbreviations to something else (in other words, to an abbreviation, that doesn't exist on any car on current grid).
Slightly OT, but this came to my mind, when I noticed this bug when testing: is there any reason, why that qualifying/best lap/results table is not shown at all during practice sessions (= when race length is 0 laps)?
Thanks, I'll check that out this evening. It's probably something simple.
(Joe mama1256)
DELETED
by Joe mama1256 : not where it should be
(Flame CZE)
DELETED
by Scawen : no longer needed
Can confirm this - I just run a replay,where many different cars were running in qualifying,the car abbreviations start to change eventually there was listed a car "RB4G" in a field with a single standard RB4.
If you need material with many different cars at same time on track to test,feel free to use qualifying mprs (mpr files ending with Q) from RandomCar - each round has one. https://www.lfs.net/forum/526-RandomCar
(Joe mama1256)
DELETED
by Scawen : no longer needed
Thanks, the D47 bug was very simple to fix. Just missed a point where one table was updated and the other wasn't.
D48: https://www.lfs.net/forum/thread/102117
D48: https://www.lfs.net/forum/thread/102117
Can confirm that bug has been fixed now
However, I found a very strange glitch and also a crash:
Here's how to reproduce it.
1) Start a single player race where you are racing alongside AI-drivers - you can't do this on multiplayer replays
2) Make sure you save a replay of it - it doesn't matter if race is finished or not
3) Start the replay.
4) Now, here's the first bug: you can use commands in SPR-replays. So, if you use command /spec [AI-driver name], you can spectate AI driver from a replay. Now, soon you will encounter Replay OOS Error.
5) If you just exit that replay normally, nothing happens. However, if you start playing that replay again by pressing 1 and then exit that replay, game crashes.
Crash report:
Now, here's how to get things a little bit ridiculous and hilarious to say the least.
Repeat the first four steps when trying to reproduce that crash. But after spectating that AI, press 1 to start that replay again. You will then encounter that Replay OOS Error. But instead of exiting, enter test drive mode.
What will happen is something really stupid. Game will merge you, and your closest AI driver together. What will happen, is following:
- If you're watching other AI-drivers, it runs normal speed. But if you're watching yourself, it enters double speed. Sounds are still enabled, though in extremely low quality. This means you can now increase game speed to 64X!
- Standings think you're watching both yourself and AI driver which merged with you.
- You can't use Shift+TAB when viewing yourself - in fact, one position is completely missing in the interface. For example, if game thinks you're in 2nd place, you can't Shift+TAB to view that merged AI whose in 1st place! And if you try to use TAB, it will skip that 1st place entirely!
And here's even more hilarious thing. If you pause that replay and then start spectating those AI drivers, you can spectate multiple AI drivers before you encounter Replay OOS Error. What will happen if you then enter test drive mode? Well, all those spectated AI's will be merged into you! What this happens that your car almost certainly starts flipping by itself and doing amazing stunts.
Also game speed increases even more!
And when you exit that test drive, game crashes. Here's a crash report to that:
While part of me wishes that this bug won't be fixed due to how ridiculous and amazing it is, I think it's very important to fix these crashes and bugs.
And luckily, I think there is a very simple way to fix this. Prevent using commands when viewing SPR-replays. I don't think there is any purpose to use those as it seems that every command causes that Replay OOS Error at minimum, if not crash the game. In test drive mode I think using commands is fine. I spectated AI during that by command and while it stopped recording replay, I managed to restart the test drive with no problems at all.
I've attached a replay where you can test this bug, though you may need to download those five mods I mentioned when testing D47.
However, I found a very strange glitch and also a crash:
Here's how to reproduce it.
1) Start a single player race where you are racing alongside AI-drivers - you can't do this on multiplayer replays
2) Make sure you save a replay of it - it doesn't matter if race is finished or not
3) Start the replay.
4) Now, here's the first bug: you can use commands in SPR-replays. So, if you use command /spec [AI-driver name], you can spectate AI driver from a replay. Now, soon you will encounter Replay OOS Error.
5) If you just exit that replay normally, nothing happens. However, if you start playing that replay again by pressing 1 and then exit that replay, game crashes.
Crash report:
Now, here's how to get things a little bit ridiculous and hilarious to say the least.
Repeat the first four steps when trying to reproduce that crash. But after spectating that AI, press 1 to start that replay again. You will then encounter that Replay OOS Error. But instead of exiting, enter test drive mode.
What will happen is something really stupid. Game will merge you, and your closest AI driver together. What will happen, is following:
- If you're watching other AI-drivers, it runs normal speed. But if you're watching yourself, it enters double speed. Sounds are still enabled, though in extremely low quality. This means you can now increase game speed to 64X!
- Standings think you're watching both yourself and AI driver which merged with you.
- You can't use Shift+TAB when viewing yourself - in fact, one position is completely missing in the interface. For example, if game thinks you're in 2nd place, you can't Shift+TAB to view that merged AI whose in 1st place! And if you try to use TAB, it will skip that 1st place entirely!
And here's even more hilarious thing. If you pause that replay and then start spectating those AI drivers, you can spectate multiple AI drivers before you encounter Replay OOS Error. What will happen if you then enter test drive mode? Well, all those spectated AI's will be merged into you! What this happens that your car almost certainly starts flipping by itself and doing amazing stunts.
Also game speed increases even more!And when you exit that test drive, game crashes. Here's a crash report to that:
While part of me wishes that this bug won't be fixed due to how ridiculous and amazing it is, I think it's very important to fix these crashes and bugs.
And luckily, I think there is a very simple way to fix this. Prevent using commands when viewing SPR-replays. I don't think there is any purpose to use those as it seems that every command causes that Replay OOS Error at minimum, if not crash the game. In test drive mode I think using commands is fine. I spectated AI during that by command and while it stopped recording replay, I managed to restart the test drive with no problems at all.
I've attached a replay where you can test this bug, though you may need to download those five mods I mentioned when testing D47.
Wow, crazy bug!
Yes, crashes must always be fixed.
I don't want to exclude all commands from use during replays. We don't know what people might use for various reasons. A more targeted approach would be better.
I looked in the code and found some commands excluded from use in training lessons. I wonder if these same commands could also be excluded from use during replays and that would be a good solution without unwanted side effects.
These commands are: aiskill / aiset / aiset_all / spec / join / ai
EDIT:
Now I found a whole lot more commands that are excluded from training lessons. These are the commands that can be used on the server, not only affecting the local computer.
axlist axload axclear axlaps track weather qual laps hours wind clear ck cv restart qualify end names
And all the host functions like maxguests, adminslots etc. I can't list them all now but could note ones that stick out:
pit_all zero_all setlap
I guess all these can reasonably be excluded from replays too.
Yes, crashes must always be fixed.
I don't want to exclude all commands from use during replays. We don't know what people might use for various reasons. A more targeted approach would be better.
I looked in the code and found some commands excluded from use in training lessons. I wonder if these same commands could also be excluded from use during replays and that would be a good solution without unwanted side effects.
These commands are: aiskill / aiset / aiset_all / spec / join / ai
EDIT:
Now I found a whole lot more commands that are excluded from training lessons. These are the commands that can be used on the server, not only affecting the local computer.
axlist axload axclear axlaps track weather qual laps hours wind clear ck cv restart qualify end names
And all the host functions like maxguests, adminslots etc. I can't list them all now but could note ones that stick out:
pit_all zero_all setlap
I guess all these can reasonably be excluded from replays too.
(KiDCoDEa)
DELETED
by kristofferandersen : off-topic
Please keep SHIFT-U edit.
When I'm making layouts, lots of times I'll drive track I've made, save the replay, then watch it to see if objects should be added/deleted/moved - especially marker boards, or edit a barrier type that should be different, do something behind the barrier, etc. Things that don't change track boundaries, so my vehicle can still follow same route without going OOS. Then I can make changes at a specific place, then replay, make further changes if required, and move on to another part of the layout ... wash, rinse, repeat.
When I'm making layouts, lots of times I'll drive track I've made, save the replay, then watch it to see if objects should be added/deleted/moved - especially marker boards, or edit a barrier type that should be different, do something behind the barrier, etc. Things that don't change track boundaries, so my vehicle can still follow same route without going OOS. Then I can make changes at a specific place, then replay, make further changes if required, and move on to another part of the layout ... wash, rinse, repeat.
I think that is safe as I only want to disable various text commands (partially listed above) during replays. I'm not trying to remove all ways that you can make a replay go OOS.
Cross post to a detailed guide on how to set up your LFS to drive motorcycles with mouse X and Y combined and bonus video of a lap around Aston.
https://www.lfs.net/forum/post/2063883#post2063883
Great feeling to drive around the track with this new update. In my opinion, the biggest difficulty is uneven asphalt that create otherwise hard corner just due to a small bump or off camber.
https://www.lfs.net/forum/post/2063883#post2063883
Great feeling to drive around the track with this new update. In my opinion, the biggest difficulty is uneven asphalt that create otherwise hard corner just due to a small bump or off camber.
Can I report what I find?
Okay, first of all, I haven't tried D48 yet, I'm still on D46, but I can feel a whole lot improvement there ! now we could do offroading with bike, quite tricky but a whole lot improvement, where previously it's difficult to keep the bike didnt fall even on straight line
https://cdn.discordapp.com/attachments/296706819425042442/1170413807277973525/1104.mp4
but secondly, I think there's something strange, when I tried every fast bike mod, I notice there's something that wobbly when turning... I dont know if this caused by TC ignition cut that active as assist, or something else, this wobbly steering didnt happen before on old bike control
https://cdn.discordapp.com/attachments/296706819425042442/1170650144472039434/Live_for_Speed_2023-11-05_16-04-45.mp4
I've uploaded it on LFS discord, but they suggest me to report it directly on the test patch, so here it is
Okay, first of all, I haven't tried D48 yet, I'm still on D46, but I can feel a whole lot improvement there ! now we could do offroading with bike, quite tricky but a whole lot improvement, where previously it's difficult to keep the bike didnt fall even on straight line
https://cdn.discordapp.com/attachments/296706819425042442/1170413807277973525/1104.mp4
but secondly, I think there's something strange, when I tried every fast bike mod, I notice there's something that wobbly when turning... I dont know if this caused by TC ignition cut that active as assist, or something else, this wobbly steering didnt happen before on old bike control
https://cdn.discordapp.com/attachments/296706819425042442/1170650144472039434/Live_for_Speed_2023-11-05_16-04-45.mp4
I've uploaded it on LFS discord, but they suggest me to report it directly on the test patch, so here it is
that is a issue with that bike only fyi
[quote="Drifteris;2063719"]I've tested your bike on Westhill and they can't even make past the second turn without sliding off the road (out of 10, 2 crashed on the first turn). Must be issue with insane downforce on that bike.
https://i.imgur.com/8Kvg7fx.gif
There was no logic behind the dowforce decision besides trying to make the bikes feel good. Simply put I don't think even the current physics feel correct under braking. The massive downforce made the bike fun to ride. I'm not really sure if it made the bike faster cornering but it was far easier to run consecutive laps without crashing. I've played with the setup and every number in the editor for a year and have never liked the way anything worked besides that downforce slider.
https://i.imgur.com/8Kvg7fx.gif
There was no logic behind the dowforce decision besides trying to make the bikes feel good. Simply put I don't think even the current physics feel correct under braking. The massive downforce made the bike fun to ride. I'm not really sure if it made the bike faster cornering but it was far easier to run consecutive laps without crashing. I've played with the setup and every number in the editor for a year and have never liked the way anything worked besides that downforce slider.
I understand why you chose to use unrealistic settings before, even if I don't agree with that approach.
To make a good mod in the proper way, the thing is to get as accurate values and settings as possible. If there are physical problems it's really my job to sort that out by improving the physics model. I do understand slight tweaks to mods to deal with imperfect physics.
But the mod still has massive downforce now, even after the steering model changes. Speaking as a mod user (not as a developer) I'd like to explain that makes me simply not want to use the mod at all. No bike in the world has F1-style downforce, as it's impossible. In my opinion the undertray should be deleted, the downforce set to zero and then work on the bike's geometry, mass, weight distribution, suspension, drag, etc. to make it as near to a real 600 as possible, then it will be a fun mod to use.
To make a good mod in the proper way, the thing is to get as accurate values and settings as possible. If there are physical problems it's really my job to sort that out by improving the physics model. I do understand slight tweaks to mods to deal with imperfect physics.
But the mod still has massive downforce now, even after the steering model changes. Speaking as a mod user (not as a developer) I'd like to explain that makes me simply not want to use the mod at all. No bike in the world has F1-style downforce, as it's impossible. In my opinion the undertray should be deleted, the downforce set to zero and then work on the bike's geometry, mass, weight distribution, suspension, drag, etc. to make it as near to a real 600 as possible, then it will be a fun mod to use.
This thread is closed
Test Patch D4 (now D48)
(704 posts, closed, started )
FGED GREDG RDFGDR GSFDG