Game sounds from the PS will not be affected by this.Than
Thanks for clarifying this, and will I keep in parallel game sounds sent to my headphones through the PS5?
I would give it a go.
Therefore no low pass filter or cross over is needed?Game sounds from the PS will not be affected by this.
SIMHUB is "reading" some kind of vehicle telemetry data from your GT7 game on the PS.
Based on this data you can configure how SIMHUB activates/controls the vibration in your bass shakers.
Thanks for you efforts! It seems that everyone wishes to have track name, a real need.Hello !
I'm ready to make the effort
I have started to learn AI/ML/DL. I see here a great opportunity to put in practice what I'm learning.
For that, some data is needed as you mention to train the model. I have made a small tool to capture the needed data for training: it generates a csv file with x/y/z/speed/rpm of the track. The tool is there. A binary release also available here
It will display the track in real-time with color based on speed. Full track dump should not be needed, as I hope the algorithm will figure out quickly the track name.
View attachment 1220154
Once data is collected for all tracks, the magic of learning should help to classify the tracks quickly as you start driving on it. But this will come later.
Feel free to send me your track dumps, so the learning set can be optimized !
Is it not simply easier to guess the track by simply checking the position when crossing the start line (which will start the timer)? With a few tweaks here and there to account for different layouts. It's a <one lap delay (less for rolling starts) but reliable nonethelessHello everyone !
Are you still interested in "guessing" the track name ? I finally got something more or less usable. It is still a POC, as I wanted to play with AI/ML in python.
I have learnt few tracks, and now the discovering of the track works pretty well, it is converging very quickly.
I can recognize the following tracks: dict_values(['Brands Hatch Grand Prix Circuit', 'Brands Hatch Indy Circuit', 'Alsace - Village', 'Dragon Trail - Seaside', 'Daytona Tri-Oval', 'Suzuka Circuit East Course', 'Fuji International Speedway (Short)'])
Track candidate : Brands Hatch Grand Prix Circuit with confidence 60.7%
Track candidate : Brands Hatch Grand Prix Circuit with higher confidence 60.7%
Track candidate : Brands Hatch Indy Circuit with confidence 51.6%
Track candidate : Brands Hatch Indy Circuit with higher confidence 62.0%
Track candidate : Brands Hatch Indy Circuit with higher confidence 68.8%
Track candidate : Brands Hatch Indy Circuit with higher confidence 69.1%
Track candidate : Brands Hatch Indy Circuit with higher confidence 69.9%
Track candidate : Brands Hatch Indy Circuit with higher confidence 71.0%
If you are still interested, I would need track capture in csv. I have made the tool to create them, available on github : https://github.com/vthinsel/GT7Tracks/blob/master/GT7Map2CSV.py
It requires 2 arguments: the PS5 IP, and the track ID to record as per https://github.com/ddm999/gt7info/blob/web-new/_data/db/course.csv
View attachment 1222248
The live display creates some lag in the CSV capture, so one you have done a full lap, stop and wait for speed to indicate 0 in the console.
Hello everyone, I would need your help: I followed SimHub instructions with ports 33739 and 33740. But the message doesn't disappear and I get no connection with my PS5, which is wired on the same router than my PC. What could be wrong? I have no VPN, I'm on a private network.
EDIT: And now it works I ordered a Nobsound, 2 BST-1 and 2 TT25, for my Playseat Challenge: I guess I will spend hours finding the best setup!
Overkill ? Yes and noIs it not simply easier to guess the track by simply checking the position when crossing the start line (which will start the timer)? With a few tweaks here and there to account for different layouts. It's a <one lap delay (less for rolling starts) but reliable nonetheless
ML is probably overkill for this. 😅
I should be able to even provide exact coordinates for the start lines and grid positions.
I have to mention that current model can't distinguish reverse layouts. XYZ are not enoughOverkill ? Yes and no
First, I did this for fun, to experiment AI/ML, which is absolutely great !
Second, it works pretty well. It can provide a perfect track guess in less than 1 sec in many cases. Nurburgring is identified in 1 request
Making distinction beetwen tracks with different sections like Brands Hatch GP and Brands Hatch Indy requires you to go to a unique portion though.
Thanks to Evisolpxe for the new tracks to learn from !
View attachment 1222883
The more track we add, the more complex it can become for the model to work properly, but there are ways to improve the model too, I just began learning ML theory few weeks ago.
Great app, worked straight away for me 😎Since I have now released EzioDash for iOS, I have for the time being gone back to experimenting with other things based on the telemetry data. But what I found now is beyond my math knowledge, and I basically have no clue about matrix transformations and all that crap
I’ve been using Race Dash, works very well.Great app, worked straight away for me 😎
I’m an Apple household, so I’ve got 2 Macs, 2 iPads & an iPhone available to me, so zero way to run Simhub (unless I buy another Fusion license for my Mac mini to see if it works there)
I’ve been searching for an iOS or OSX app I can use with GT7, so far EzioDash is the only one that works!
I’ll let you know. Yesterday i spent most of the day playing in sim hub changing the settings on my Buttkicker and hp8.What works well and what doesn’t?