MX Simulator

JLV,

Any chance we could somehow get the tileinfo/tilemap to be hosted server side for quicker/easier erode testing? It would be nice to run a moto, update the settings, and run a second moto right after without everyone having to leave, get the update, come back.

Erode testing is a very long process to do correctly right now.

I'm fine with this just being a specific server and not something used for racing/official times or anything.

It could do that, but I'm not sure it's a good idea. What problems are you trying to fix? I'm really thinking I should project the force vectors to the terrain. It could make a big difference in some situations.

jlv wrote:It could do that, but I'm not sure it's a good idea. What problems are you trying to fix? I'm really thinking I should project the force vectors to the terrain. It could make a big difference in some situations.

For example, after a little over a week of testing, I was able to get rid of a lot of the jagged stuff by gradually making the erode radius larger, but each time I had to have people get a new track and then run a new 30 minute moto with a reasonable number of guys in the server (20+) to run a realistic number to get any useful data. The changes were minimal each time, but I wanted to find the smallest I could go without it getting jagged and choppy.

Next up, I need to figure out how to get more braking bumps earlier to help slow down the main lines and open up new ones. This includes still needing to sort out the downhill settings. Hopefully with the current setup I can get this done before the first outdoor round. Testing takes a ton of time and is not efficient at all. If I could test multiple settings in a single night versus one a night it would be a massive improvement.

I also want to try and get ruts in jump faces and on straights.

I still am playing with the erode limit settings to learn those.

And the slip scaling.

I wouldn't say most of the stuff is "broken", just simply can be improved and needs tested and developed (on my side with learning/tuning the settings)

I think it'd be best to use a lot of different tiles that are all slightly different. Ideally have a track with lots of perfectly identical sections so the only difference is the tilemap. Then you can check them in parallel to see what works best. While it would be possible to send out tileinfo changes from the server it'd be pretty hackish so I'd prefer to avoid that. I could also write a script for you that just makes a bunch of track folders with slightly different settings if you want.

I'll go ahead and do the force vector projection thing. While I don't know if it'll help with everything you mention, it will make the force smaller when it's pushing directly into a slope. So berms won't form so quickly and any force into jump faces won't push ahead which contributes a lot to the kicker problem.

jlv wrote:I think it'd be best to use a lot of different tiles that are all slightly different. Ideally have a track with lots of perfectly identical sections so the only difference is the tilemap. Then you can check them in parallel to see what works best. While it would be possible to send out tileinfo changes from the server it'd be pretty hackish so I'd prefer to avoid that. I could also write a script for you that just makes a bunch of track folders with slightly different settings if you want.

Creating the different versions of the track isn't an issue. The challenging part is I need to see how something reacts before knowing the next update and needing 25+ riders (even then...barely enough) to get a realistic enough test to actually learn. And keeping numbers up above that through track updates and server restarts and such is a challenge. While it may be "hackish", I only need it on a single test server.

jlv wrote: I'll go ahead and do the force vector projection thing. While I don't know if it'll help with everything you mention, it will make the force smaller when it's pushing directly into a slope. So berms won't form so quickly and any force into jump faces won't push ahead which contributes a lot to the kicker problem.

Do you have a timeframe on this and how substancial of a change are we looking at? I'm all for evolution and what you're describing sounds awesome, just scared about the limited testing time we may have when it is ready versus the testing time that has already been devoted.

checkerz wrote:Creating the different versions of the track isn't an issue. The challenging part is I need to see how something reacts before knowing the next update and needing 25+ riders (even then...barely enough) to get a realistic enough test to actually learn. And keeping numbers up above that through track updates and server restarts and such is a challenge. While it may be "hackish", I only need it on a single test server.

How many different parameters will you be changing? It it's just a couple I can just pre-generate a couple hundred different versions of the track that have all possible combinations. It'd be way easier for me to generate that for you than to hack up the client to modify the tile settings mid-race.

checkerz wrote:Do you have a timeframe on this and how substancial of a change are we looking at? I'm all for evolution and what you're describing sounds awesome, just scared about the limited testing time we may have when it is ready versus the testing time that has already been devoted.

It'll be an easy change. The only reason I haven't done it yet is I'm in the middle of adding javascript scripting to the server.

jlv wrote: It'll be an easy change. The only reason I haven't done it yet is I'm in the middle of adding javascript scripting to the server.

Serverside javascripting? Sounds freaking amazing! You sir are a legend.
I can't wait to see this. Hopefully no more dll injections

SethEnslowJr wrote:Serverside javascripting? Sounds freaking amazing! :D You sir are a legend.
I can't wait to see this. Hopefully no more dll injections :D

Technically you can do most of the same stuff with external scripts but the integrated javascript is way easier and cleaner. If you do server scripting I think you're really going to like it.