60 Fps Patch
Posted Feb 09 2018 - 03:21 AM
I had this in table form but the post unraveled the table. I guess I need to learn how to take a picture of the table?
Posted Feb 10 2018 - 10:22 AM
I updated my post #30 to include the target pole position times for the 60fps AI. These data will be useful in trying to recreate a 60fps version of the 1967 Historical Mod.
This is truly a wonderful Mod ... thanks to all who participated in its creation. It looks even better at 60fps and higher levels of screen resolution.
Posted Feb 13 2018 - 09:10 AM
As with the other "non F1 Mods" this mod likes the 60fps AI v1 patch WITHOUT the AI patch that reduces cornering speeds by 10%. If you have been following this thread ... that is the same result that was experienced in the 66 and 71 CAN AM Mods. The GEM 2 AI setting centered on 1.07 (i.e. the mod is a little too fast even with the 60fps ai V1 patch. Here are the tracks / pole positions / and delong_speed_adj_coef data:
1967 World Sport Car Championship/
(GEM 2 AI set at 1.07)/
Track/ Actual Pole Position (in sec)/ Target Pole= +10% (in sec)/ delong_speed_adj_coef/
1. Daytona/ 115/ 127/ 0.9776/
2. Sebring 67/ 168/ 185/ 0.9021/
3. Monza 10K/ 174/ 191/ 1.0163/
4. Spa 67/ 216/ 238/ 0.8860/
5. Targa Florio/ 2232/ 2455/ 0.9050/
6. Nurburgring/ 512/ 563/ 0.9550/
7. Bsarthe/ 204/ 224/ 0.9143/
8. Brads Hatch 67/ 96/ 106/ 0.9750/
Since the AI are slowed down slightly at a GEM 2 setting of 1.07 (1.00 is normal; higher is slower), the need for the adjustments to driver aggression and alertness did not seem necessary. If you do decide to adjust these driver attributes bear in mind that each of these 8 tracks has its own separate driverGT.ini (ii.e. you will have to adjust each track specific driverGT.ini separately; there are 8 of them).
If you run additional tracks, there is a general overall driverGT,ini for all tracks other than these 8. This overall driverGT.ini already has the aggression indices set at 0.800 (as opposed to the general 1.000 range normally used). If you feel the AI are still too aggressive you can set the alertness indices to 1.200 across all 19 AI drivers for additional prevention against AI crashes.
These 8 tracks had one of the larger spreads in delong_speed_adj_coef adjustment that I have experienced; namely 10% from low to high. 3 of the 10 tracks needed significant adjustments from the norm of 1.07. I'm sure the developers of this excellent MOD adjusted these tracks at 36 fps, so the 60fps AI v1 Patch must be having a divergent effect on AI performance? As you add additional tracks, keep this in mind .... you may have to reset the delong_speed_adj_coef (in the trackGT,ini file) on the tracks you add. The other settings suggested here should work for all new tracks.
If this post does not make sense to you, I suggest you reread my prior posts that describe the reasoning and processes behind these suggested adjustments ... still unsure, send me a question and I will attempt to explain further.
Hope this information helps ... more to come,
Edited by Roger_F, Feb 13 2018 - 09:28 AM.
Posted Feb 21 2018 - 09:53 AM
After completing the above, open up the attached chart and apply the revised delong_speed_adj_coef data to each track66.ini file. While you are at it, you should also set the max coef to 2.5000 and the statistics reference value to a few seconds over the target pole position. This needs to be done for each track referenced in the attached chart.
You are almost done ... run a simulated qualifying session for each track to make sure your target AI pole positions are close to target values and enjoy racing at 60fps. I have found that I can run at max screen resolution for my screen, highest detail levels for most tracks and 60fps with only a slight drop in frame rates at the start and a full 60fps for the whole race.
If any of this seems confusing, I suggest you reread my prior posts in this thread and if still concerned drop me a line and I will try to help you individually. I am basically applying the same methodology to each of the 12 or so Mods I have in my GPL installation. This methodology is detailed in prior posts.
Edited by Roger_F, Feb 22 2018 - 12:59 AM.
Posted Feb 25 2018 - 11:37 AM
Once you get much below 1.0 for GEM 2 AI you also need to reset the driver aggression data to 80% of their current values (standard is around 1.020 new values will be around 0.820) for each of the 19 AI drivers. These data are in the driv65.ini file ... in the 65 and 66 Mods :one files controls all tracks. JonnyA also suggests that you can reset the driver alertness settings to the 1.2000 range for additional crash avoidance; this also helps substantially! Try one or both adjustments and see which one you like best.
After completing the above, open up the attached chart and apply the revised delong_speed_adj_coef data to each track65.ini file. While you are at it, you should also set the max coef to 2.5000 and the statistics reference value to a few seconds over the target pole position. This needs to be done for each track referenced in the attached chart.
You are almost done ... run a simulated qualifying session for each track to make sure your target AI pole positions are close to target values and enjoy racing at 60fps. I have found that I can run at max screen resolution for my screen, highest detail levels for most tracks and 60fps with only a slight drop in frame rates at the start and a full 60fps for the whole race. The 65 cars were about 1% to 2% slower than the 1966 F1 cars and that was reflected in my data for "target pole positions."
In my installation of the 1965 F1 Mod I also have the 1964 through 1968 Tasman Series Cars. Physics for the 64 through 66 Tasman cars are from the 65 F1 Mod and the 67 and 68 Tasman Cars use the 67 FD physics (F1 chassis with the 2.5 litre race engines). Interestingly, both Tasman groups are a little faster that they should be on multiple 65 tracks ( I tested at East London, Silverstone, and Watkins Glen). In order to get reasonably correct target pole positions I had to use Gem 2 AI values of .97 for the 64 through 66 group and 1.07 for the 67 and 68 group.
I did not adjust aggression, alertness or track track pole times for the Tasman Cars at this point in time. There are too many tracks that these cars can run on to make this task reasonable. Trying to find target pole times for the Tasman Series also seems to be extremely difficult ... if not impossible. I adjust the tracks by feel as I race on them. If I consistently qualify at the front of the pack I speed up the AI; If I consistently qualify at the back of the pack, I slow them down: using the delong_speed_adj_coef in the respective track65.ini file. Most Tasman tracks have less abrupt transitions than the typical F1 track ... so AI crashes seem to be less of a problem.
If you need more of an explanation I suggest you reread my prior posts in this thread and if still concerned drop me a line and I will try to help you individually. As I've said before, I am basically applying the same methodology to each of the 12 or so Mods I have in my GPL installation. This methodology is detailed in prior posts.
Edited by Roger_F, Feb 26 2018 - 07:30 AM.
Posted Mar 08 2018 - 10:37 AM
Since the 1967 F2 mod is similar, physics wise, to the 1965 F1 mod, I have reused some of the same phraseology to explain the conversion to 60fps. As in the previous post (on the 1965 F1 Mod), I used the the 60fps ai1v patch and the 78l Ai Speed Control V2 patch. This seems to be a good starting point for almost all F1 / F2 Mods.
After some testing I determined that the GEM 2 AI needs to be set at .97 (this is less of an adjustment than was required for the 1965 F1 Mod). 1.00 is standard for GEM 2 AI and generally needs no adjustment if the delong_speed_adj_coef data is also close to 1.00000. Once you get much below 1.0 for GEM 2 AI you need to reset the driver aggression data to 80% of their current values (standard is around 1.020; new values will be around 0.820) for each of the 19 AI drivers in each F2 race year. If the delong_speed_adj_coef data for each trackF2.ini file goes much over 1.00000 you also need to adjust the Aggression and / or the Alertness indices in the drivF2.ini (1967) and dr68F2.ini (1968) files. Try 1.2 for the driver Alertness indices if the aggression adjustment is still too crash prone for your tastes. This interesting bit of data on Alertness came from JonnyA on a previous post in this thread.
If you study the attached tables for 67 F2 and 68 F2 you will see about a third of the tracks fall into this adjustment criteria range? I'm not sure why this is occurring but it is clear that this group of tracks were not collectively adjusted for the F2 mod. Since one general driver.ini file controls the whole collection of tracks for each race year, the previously described adjustment is necessary for both the drivF2.ini (1967) and dr68F2.ini (1968) files (this is 19 AI drivers to adjust for each race year ... 1967 and 1968).
This is probably a good time to introduce the excellent 68 F2 addition to this mod by Jonathan Abbott (username JonnyA). This addition uses the physics of the 67 F2 Mod but includes the cars that actually competed in the 1968 F2 Season to correct detail for 1968. The result is a separate 68 season that can be matched against the actual tracks used in the 1968 F2 championship and non championship venues.There are 3 unique tracks in the official 1968 Championship races (that are different from 1967) and about 4 unique tracks in the non Championship races. JonnyA's new 1968 carset is outstanding visually and (in my opinion) noticeably improves the mod. New drivers are defined for 1968 and an excellent new car set, that is historically accurate, give us a complete new season to enjoy. Though the 68 addition uses the same physics it only includes the cars that were carried over to 1968; the net result is a somewhat faster AI field for 1968. Many many thanks, JonnyA!!!
After completing the above suggested changes, open up the two attached charts and apply the revised delong_speed_adj_coef data to each trackF2.ini file. While you are at it, you should also set the max coef to 2.5000 and the statistics reference value to a few seconds over the target pole position. This needs to be done for each track referenced in the attached two charts. There are a few non Championship tracks that I chose not to include in my two, race year set up. If you want to include them follow the procedures outlined in previous threads in this topic. If you still need help drop me a line and I will walk you through the process. You are almost done ... run a simulated qualifying session for each track to make sure your target AI pole positions are close to target values in the two charts and enjoy racing at 60fps.
Additional tracks will need new target pole position data to follow the process. A google search will usually yield either non championship "best lap" or "pole position" F2 data; convert to seconds and add 10% and your ready to roll. There were 16 non championship races in 1967 and 7 races in 1968. Not all of these tracks are available in GPL and some are better tracks to race on than others. I roughly split the non championship races between 67 and 68 with no repeats in my final set up. I left out a few tracks for various reasons.
If you have been following this thread you should now have quite a good collection of 60fps capable GPL Mods ... more to come.
Edited by Roger_F, Mar 08 2018 - 10:50 AM.
Posted Mar 12 2018 - 03:21 AM
Posted Mar 12 2018 - 09:20 AM
Lets first try and make this instructional so we can help others who may be having similar problems. If I were to download every file I have altered in this entire 60fps AI Tuning thread, the downloaded files would number several hundred and explaining where to put all those files would be a very interesting task ... to say the least.
For instructional purposes I have selected the Mexico track from the 67 Historical Mod. The attached screenshot is from the file track67.ini located inside the track file Mexico. (1967 HISTORICAL / Tracks / Mexico / track67.ini). I opened up this file with notebook and scrolled down to the sub category [ GP ]. Under this sub category [ GP ] are two values and attached explanations. The current value of delong_speed_adj_coef = 1.05000 and this value needs to be changed to 1.01000. the information following the " ; " is read only and does not affect the operation of GPL.
There is also a value for delong_speed_adj_maximum which is currently 2.38000 and this should perform slightly better if you reset this to 2.50000. All GPL Mods need this set to 2.5000 for every track for the AI to shift at the proper times. Since 2.38000 is so close to 2.50000 you could probably leave this unchanged but for future reference in other Mods this number needs to always be set at 2.50000!
Now each of the remaining eighteen (18) 1967 Historical file tracks have this same track67.ini file included therein. Even though the "track67.ini" file name is exactly the same in each track file the data included is unique to each individual track file. The data located in the chart attached to my original post is relevant to the same place in each track67.ini file and you need to replace the original value with the new value provided in the chart included in the original post. After replacing the value you need to "save" the file using the dropdown menu in the upper left hand corner of the header of the track file. Do this 18 times and you are ready to roll. You also need to follow the instructions for the GEM 2 AI file value plus any other instructions unique to the Mod post you are trying to replicate.
If I sent eighteen (18) track67 ini files (each with exactly the same name) some of you would likely struggle to match the right file with the right track ... changing two pieces of data in the same existing track.ini file (for each track) seems safer from an accuracy standpoint and easier from an execution standpoint. if you don't delete / replace the file it will always be in the right place and the two numbers you change will allow the 60fps AI adjustments to work as planned.
I hope this helps ... if you still have problems send me another reply.
Best of luck,
Edited by Roger_F, Mar 12 2018 - 09:42 AM.
Posted Mar 12 2018 - 01:42 PM
really good experiments. I just have begun to read your first post here . i remember I was tweaking the AI's files , but didn't play with the track.ini folders
You spent many hours of tweaks, I see, and going back and forth process can be a pain . Appreciated a lot !
Edited by M Needforspeed, Mar 12 2018 - 01:53 PM.
Posted Mar 12 2018 - 11:51 PM
Posted Mar 13 2018 - 02:51 AM
Reducing the GEM 2 AI value to less than 1.00 speeds up the AI (down the straights and in the corners); For example: .85 is more than 15% faster than 1.00. The addition of the 78l Ai Speed Control V2 patch to Gem 2, slows down the AI in the corners by about 10%. These two in combination put the 60fps patch in relative balance for F1 and F2 cars and tracks. Interestingly, the 78l Ai Speed Control V2 patch does not seem to be a good addition to the CANAM and SCGT mods?
The adjustments to "delong_speed_adj_coef" balances the AI between tracks so they act correctly as a group to historical norms. Settings above 1.0000 speed up the AI everywhere; higher values are faster. Lower values slow down the AI everywhere as well. Most Mods do not do this balancing because they do not know what tracks you are going to choose to apply to their Mod. So the track group you choose for a particular Mod is probably out of balance by definition at both 36fps and 60fps. For some reason the 60fps ai1v patch seems to be more sensitive to this balance than the standard 36fps in the GPL simulation.
The third set of possible adjustments to driver.ini aggression and alertness are done to reduce the crash tendencies caused by the combination of lower GEM 2 AI settings and higher delong_speed_adj_coef settings. Even at 36fps, low GEM 2 AI settings and high delong_speed_adj_coef settings will cause the AI to go crazy and crash into each other and you.
Putting all this together allows one to set a particular "Mod" and its chosen tracks in balance to the history of the series being represented. As you can probably see, all "Mods" are not created equal; hence the varying adjustments are necessary between Mods. A lot of these suggested adjustments would also help the Mods at 36fps.
Adding to all this confusion are new AI packages for the many existing tracks in the GPL Track Database. These improvements also effect AI lap times as well as on the track AI performance. Trying to build an historically correct GPL simulation for any given time period or series is quite a difficult task. Hats off to those individuals that give us all the great Mods that endear us to GPL!!!!
Thanks for listening,
Posted Mar 13 2018 - 07:17 AM
Posted Mar 13 2018 - 08:02 AM
On the 71 CANAM and SCGT Mods, the AI seem to run better without the Speed Control patch on the Championship Tracks specifically designed for those mods.
1 user(s) are reading this topic
0 members, 1 guests, 0 anonymous users