The new vehicle performance statistics in version 0.6 ("Performance" Tab in the vehicle selector) are not manually created but by an automated test. It needs a dedicated level and quite some time to do various tests of offroad behavior, acceleration, braking and so on. Please note that this is an optional step for the creators of mods, regular users won't need to do anything. How to generate performance statistics for your own mods: 1) Make sure to have at least BeamNG 0.15.0.0 2) Create a file called work.json in the beamng main directory with these contents: Code: [{"type": "testVehiclesPerformances", "vehicle":"citybus", "pcFile":"hero"}] You can specify a specific vehicle with changing the "vehicle" attribute and additionally a specific configuration of your test vehicle via the "pcFile" attribute. If the pcFile attribute is left empty, it tests all configurations of that car. Note: every configuration needs to have a valid info_*.json file already, if not it's skipped 3) Run the batch file called BeamNG.drive - Auto Tests.bat in the main beamng directory and wait for it to do its thing until it closed itself again Contents of that file: Code: Bin64\BeamNG.drive.x64.exe -batch -nocrashreport -quitonfatalerror -lua extensions.load('util_worker') 4) Copy the generated info*.json files from your /Documents/BeamNG.Drive/vehicles folder back into your vehicle source files 5) Clean up the *.json and *.touched files from the documents path if you want to rerun the test Here's a little video of how such a test looks like (please excuse the horrible bitrate, trying to keep the file somewhat small ) Video - Click to Play - Direct Link Currently we gather the following data points (listing metric only, imperial data is also available) Power (including band) Torque (including band) Weight Weight/Power ratio Top speed 0-100 km/h 0-200 km/h 0-300 km/h 100-200 km/h 100-0 km/h Artificial "Offroad Score" that mostly works on ground clearance and overhang lengths (see first part of the video) Braking performance Propulsion type Powertrain layout Fuel type used Notes: Acceleration tests might not work perfectly for very powerful cars with a generally bad setup Race brakes will need to have the preheat option enabled to give somewhat valid results For the future we'd also like to look into crash data of some sort, but it's not yet decided how exactly that will work. What do you think about it? Is something missing, something that you would love to see? Any problems / errors? Disclaimer: Please note that this is still a work in progress, trying to perfectly launch any car on the first try is quite challenging, so expect some changes in future updates.
Okay, for people who dont get this at first like me. First: This about the performance statistics of cars, which you see in vehicle selecting menu. It wont evaluate the perfomance of your pc. To get this gibberish working you need to create a start.bat in your gamefolder. eg: D:\Steam\steamapps\common\BeamNG.drive Open the texteditor of windows, paste the gibberish into it. Select "save as", select there "all files" instead of ".txt files", name it "start.bat" and save it in your gamefolder. Now you can double click the .bat and it will do its thing. Dont do any input, it will mess with your results.
@B.Tanner: thanks for the additional explanation Also, one more thing: to test all installed configurations, just remove the "-testvehicle something" bit
Where exactly are the json's located? nothing in my documents folder. I pasted the code into the start options in steam. EDIT : i also tried the batch method. when i open it , cmd opens and closes after like 0,1sec again. nothing happens after that Edit 2 : working now. forgot to remove 'BeamNG.exe' from the startoptions in steam. Thanks for all the information though!
I play the game without preheated engines and brakes. I recognized the perfomance test is running with the options you've chosen. idk if the difference is significant enough but it might be a good idea to tick those option back in for better comparability.
with the config testing , this is how my command looks like. but it's only spawning random configs. Code: -nocrashreport -console -level autotest.mis -lua extensions.loadModule('util/saveDynamicData') -testvehicle sunburst -testconfiguration vmcup where exactly do i have to place the config part to test my selected configurations only?
one last question , (i know i'm asking a lot , but i might help others too) is it possible to test multiple configurations? or select the folder of the cars (so that only the mod configurations will be tested for example)? because it is a bit time-expensive to wait till every car finishes and type in the new config. with selecting only the cars that need to be tested , (in this case mod folder) , the game could do everything in the background without breaks.
I'd add this info, Pre-heated brakes have to be applied in performance cars, otherwise test of acceleration will be invalid, because car will launch with brakes applied due to their low thermal efficiency. Tested on Rally and Hill Climb configs.
I'd love to see a max lateral Gs test. Have the vehicle drive in a circle until it loses grip, then log the maximum G forces it encountered during the test. Also, downforce (front and rear) is really important to vehicle performance. Use the metric as X kg at Y km/h
Lateral Gs are surprisingly complicated since you ideally need to drive in a perfectly steady circle with increasing speed. Keeping that circle with *any* car right off the bat is a very complex control logic problem. But yea, we are looking into that (especially if fixed speed + increasing wheel angle will give us results close enough) Downforce could be a nice one and probably not too hard to test either
Unfortunately this is not possible at the moment. Feel free to edit the extension lua file to match you needs though.
Thanks for the reply! What if you drove to 40mph (or full throttle for 20 seconds if the vehicle is slow, like a fully loaded T-Series or mod vehicle), held the speed, then slowly turned the wheel to full lock, until lateral grip is lost?
Yea, that's what i am describing above. It's not the same as actually keeping a steady circle physics wise, but we'll have to see if this gets us close enough to the real values to be usable. After all this whole performance thing is mostly for comparision between the cars, which makes issues in the measuring logic a bit less important.
I agree. as long as the tests are consistent and inclusive (e.g. you get a lot of stats from it), then comparison between cars becomes a numbers game, however false those numbers are when pertaining to real-world performance
@Diamondback what do you think of adding a CoG (center of gravity / mass) stat to the vehicle? It would be a 3 component vector with a % from center on x and z axis, and a height from axle on the y axis. Also higher speed brake tests would be cool too. 100-0 and 200-0 where applicable.