verbal gazelle
Hey guys, can anyone tell me what method is used by vedal to make neuro sama play among us, I mean those lines on the floor and also those little dots on the floor? and something like raycasting?
https://www.youtube.com/watch?v=HxjsPAsg32k
Note: I need to watch the video
hoary ember
Idk, Vedal hasn't shown any updates in months. He could've completely changed the way it works.
vestal yew
i am the one working on the among us mod for neuro
we use floodfill to generate the grid
and then a* pathfinding to move
verbal gazelle
i am the one working on the among us mod for neuro
Congratulations on the work, I hope it goes well 
Thank you very much for the information, I wanted to understand more about this.
@vestal yew One question is that the neuroscience related to the paths already has a predefined walking location, right? It's kind of like she won't learn to find the door just by knocking until she gets through, what reinforcement learning would be like.
I'm very new to this subject, be patient lol
vestal yew
we are not using neural networks for movement
the movement is handled using traditional pathfinding
verbal gazelle
So the AI part would be to choose existing paths?
vestal yew
yes
verbal gazelle
Or maybe just do the quests
That's quite nice
Do you know anything about the mine-playing A.I. that neuro sama used? @vestal yew
vestal yew
no
verbal gazelle
I understand, thanks for the help
narrow skiff
we are not using neural networks for movement
this is not correct
she uses a neural network for moving currently, the potential paths she could go are given as input - along with other data
ionic lantern
Missinformation ain't no way
wraith mortar
comfy corner
verbal gazelle
If you used pathfinding to create the paths because there is this grid on the ground @narrow skiff ?
verbal gazelle
she uses a neural network for moving currently, the potential paths she could go...
This way, she has much more freedom
vestal yew
this is not correct
bruh i wrote the code
the neural network just chooses the destination
but the pathfinding is done without a neural network
south gulch
mom dad quit fighting
narrow skiff
bruh i wrote the code
I wrote that code
the neural network outputs a desired move direction which is received by the plugin and used to override the movement direction
pathfinding is used as input to the neural network NOT output
finite abyss
vestal yew
I wrote that code
the neural network outputs a desired move direction which is r...
yes you are right
i forgot
but that's to be expected given the last time i worked on it
it feels like it was years ago 
vestal yew
That project is most likely dead
ionic lantern
Still dead
vestal yew
hey, i am allowed to make jokes about the state of the project because i worked on it
unlike yourself
ionic lantern
Hmmm
south gulch
so it's in the archives?
icy jasper
I wrote that code
the neural network outputs a desired move direction which is r...
Holy code.
south gulch
are those archives being kept?
verbal gazelle
Do you talk about that? https://github.com/VedalAI/neuro-amongus @south gulch
Not sure if it is updated
south gulch
huh?
oh!
oh, I was just making an "abandoned archive" joke. Sorry for the confusion!
verbal gazelle
So I'm Brazilian and I need to translate some things, maybe the joke wasn't clear because of the translator
lol hahaha
south gulch
That was my fault! Don't worry about it, lol. Thank you for asking your insightful question
verbal gazelle
verbal gazelle
If you used pathfinding to create the paths because there is this grid on the gr...
I think now maybe I understand about the grid(loodfill), basically it was used to map the terrain, thus providing data for pathfind(a*) to choose the best route between the tasks, is that correct? @vestal yew
However, this would only be an initial mechanic because, as Vedal said, the neurosama would be able to do it without necessarily using a route drawn by pathfind, perhaps walking across the entire mapped grid(loodfill)?
