Gtuner IV 1.07 PUBLIC RELEASE: Computer Vision

[hammer]

Here is where you can start:
https://www.consoletuner.com/wiki/index ... _scripting


Well, I just started looking into it and it is more confusing than i thought,
will there be a chance that you are writing a script with antirecoil slots specified to different weapons?

And everybody just needs to adjust some minor tweaks to make it work for his sensitivity and for his likings?

Best regards,
Golf

That will prob happen sooner or later

[residualflash]

Wow this is awesome. Great work. Currently I am using an autohotkey script to detect weapons in Apex Legends using Image Search and sending data to GTuner via KMG capture. Having this baked into GTuner is awesome. Also thank you for providing the anti recoil values for each weapons. J2Kbr would you be willing to share the GCV script you used to find the values? I need to figure out how to convert these values to use with my XIM.

[Gjin Yuko]

Wow this is awesome. Great work. Currently I am using an autohotkey script to detect weapons in Apex Legends using Image Search and sending data to GTuner via KMG capture. Having this baked into GTuner is awesome. Also thank you for providing the anti recoil values for each weapons. J2Kbr would you be willing to share the GCV script you used to find the values? I need to figure out how to convert these values to use with my XIM.

Nice to see a fellow Ximmer! Waiting for them scripts

[hammer]

Just a little info I gathered.

I wrote back and forth with a user on discord and he couldnt get the opencv package installed.

He is from russia so probably his ISP or what ever blocked the connection to pythonhosted.org to get the library from there.(even with a local .whl file pip tries to access the internet for some reason)
So if you get a "Warning retrying ...NewConnectionError..." use a free VPN to get that done.

Secondly he installed python to Program Files/Python but thats not where opencv gets installed.
When you use

Code: Select all

pip install opencv-python

inside the commandline it will be installed inside

Code: Select all

C:\Users\username\AppData\Local\Programs\Python\Python38

Note: You have to enable hidden folders to see AppData.

So when you install python dont take the zip or change the default path from the installer. Otherwise gtuner will never find opencv on the system if its in a different path than the one stated above.
You could specify a custom path with pip install but you probably want to keep this as easy as possible to setup I think?! That's probably common knowledge for most regular windows users but for some its not.

And I tried a blank .gcv script and it always gave me an error. The problem was that I named it test.gcv but test is a folder(?) inside the python path and for some reason it tries to access it somehow. So dont name a test , test ;)

Thx @DontAtMe for pointing that out in discord for me. He had the same issue and investigated.

[USER101]

Great Move!
I am glad to see progress in this direction. Unknown to most, I have been making strides behind the scenes using techniques like this. When I first mentioned my desire to pursue object recognition or some sort of visual feedback system in conjunction with the T2 a few years ago it was heavily criticized, including DMs that suggested I not mention the idea or that I reword my thoughts in a way as to minimize the gameplay assistance potential. Alas, I kept my comments to myself. Now, it seems that we are more progressive and open to the technology and the great thing is that you went in a direction the perfectly aligns with my own research. My biggest issue was input lag as I was forced to use 3rd party software and KMG for feedback. Now that I can directly hook python it opens up a whole new realm of possibilities. I appreciate the hard work and dedication that goes into this device and will assist anyway I can. I am more of a discord person but as we venture into object detection I may be able to better contribute via step by step video tutorials at some point in the near future. Here is a clip from mid 2019 that shows some of the progress made back then and it has only gotten better. So far, I see no reason this and other great training sets can't be directly incorporated into the new computer vision system.

[J2Kbr]

-- UPDATE (February 18th, 2020) --

New revision of the Gtuner IV 1.7-BETA is now available, more details here:

viewtopic.php?f=2&t=14717&p=95655#p95655

[jelly44]

J2Kbr - As this is all foreign to me, I was wondering what I would need to acquire now in terms of hardware to benefit from the next GTuner release, consequently, the new functionalities.

Considering someone who plays on a console (PS4 in my case) and uses KMG capture for m&k via a windows OS laptop.

Thanks

[J2Kbr]

Will there be support for Xim Apex users? Like Anti Recoil etc..

This new feature can also be used with XIM/Apex and other devices, as long as you have the Titan Two in the chain.

[J2Kbr]

el gato hd60 pro doesnt seem to work.. its detected but nothing visual

You need to run a .py script via the Computer Vision pane. Copy the code from the minimal example from the very first post or run the Aped Legends one from the Online Resource.

With the new beta release a GCV script is not needed anymore to test the video input. In the scripts combobox there is a new option "Video Input Check" for this purpose.

[J2Kbr]

on the overlay its saying no weapon detected no barrel etc but i do have a weapon etc? sorry im noob am i not using this right lol

Post a picture of the computer vision container, running the game.Make sure you have a weapon equipped too.

https://i.imgur.com/fjFMf3l.png

The Apex Legeds GCV script I published, at first moment, is to serve as example. It is not working for you because the colors of your capture card seems too much saturated and/or with high contrast. Take a look in the image file downloaded with the script (apex_legends.png), these are the templates used to detect weapon, attachment, etc. If the color does not match, the detection will fail. Down-the-road I will be improving this GCV script to be more robust against differences in lighting and color saturation.