These AI-Driven Characters Dribble Like Mad! 🏀 

You are very kind. Reading these papers gives me a lot of joy and I am delighted to see if it goes through for you Fellow Scholars too!

You can press shift+enter to jump to a new line

cat I typed it on my phone and the ui just doesn’t work. (Sorry everyone)

cat just press enter. Why the shift? I always just press enter on desktop and mobile

@@bluebaconjake405 Shift + enter is a soft return as compared to enter on its own which is a hard return. So, pressing shift + enter is guaranteed in most cases to give you a new line in applications where pressing enter on its own may do the same thing as pressing the "ok" or "send" button.

I think you mean EA

@@cmdrblaze6487 LMAO you don't know 2K Basketball games then

@@remyshea954 "LMAO" what is this response...

@@sceplecture2382 dude are you like 40 years old holy shit

@@ThugHunterfromIsrael LAUGHING MY ASS OFF

even if it is the 10th, there will certainly be two more papers for movement based machine learning in the future.

@Solve Everything its now 5 minute papers haha

Wtf is 2nd paper

@@ctrey8395 this ^

@@ctrey8395 I hope, in this case, second paper is someone using this to enable easy generation of generalized game animations, especially for close melee combat. This seemed to have dynamic interaction components at multiple levels... we can definitely expand on that element and make this into a full general kinematics AI or something... I'm not sure I imagine that this will eventually lead to something where you can customize the behavior of animations in a very generalized manner, especially if it's possible to somehow implement something similar to the technology that allows painting landscapes by simply defining regions. I wonder what we would get if we trained it on stylized animations as well.. such as in smash.

Haha nice

more like badass geeks instead of nerds

Dude these nerds gonna sell these tech to 2K21 and get front roll seats of NBA for life.

@@zhehaoshen3635 but then they gotta sit through basketball, better to use the money on ttrpg source books and figures

@@tuseroni6085 A level 35 Micheal Jordan enters the dungeon, with his max out dunk skill, he can one hit kill any Laker type monster.

Unfortunately. "This project is only for research or education purposes, and not freely available for commercial use or redistribution."

Looks like a hybrid of machine learning and motion matching

@@FlynTie everything starts as "for research", we'll get practical applications of such methods eventually.

The algorithm is accessible. The project is for research, but it's just a matter of time until commercial aplications start showing up. Imagine indie companies being able to create the same ammount of animations that a game like Red Dead Redemption has with just a couple of bucks.

@@FlynTie Not sure how feasible motion capture data and techniques even are for an indy developer. Seems out of reach anyway.

I wish but responsiveness would make it unplayable

@@onbggt4556 true, it would just be multiple modes of "let the human win" at varying degrees of difficulty. That's probably the case now though.

Just a shame that they've moved on to simulating gambling experiences instead of basketball just as all this cool research is coming out.

@@shallowandpedantic2320 same as fifa they're just going where the money is 😔

@@shallowandpedantic2320 NBA2k19 did great for implementing mini GTA V in game it's just sad that there's no competition since NBA2K is far ahead than other basketball game

NBA3K1

that would be sick

I would troll the hell out of the AI.

That is a cool idea, but you also have to make sure that visual flair like that both cooperates with the gameplay and doesn't negatively affect the player experience. Let's say you implemented the sand-boot thing in a platformer game; when the player triggers it, you either have to let the player keep their normal speed and range of movement, leading to very awkward on-the-fly animations, or you have to impede the player's movement to make it more realistic, in which case the player's enjoyment is compromised.

@Honudes Gai That is such a bizarre opinion to have.

@@imveryangryitsnotbutter I don't think enjoyment has to be compromised. I could see a system where the player has to be mindful of controlling the character in such a way that they don't fall over. For example, running too fast down a steep slope, the character's feet will eventually get away from them and they'll fall on their face. Or running full pelt in one direction and sharply turning could cause the character to lose balance. I imagine it to be something like that realistic skateboarding game "Skate" where all this stuff is possible except there's no skateboard involved

Actually now I think of it, the GTA games have been doing this with the Euphoria motion engine thing and the results can be hilarious (jumping down stair cases, hanging onto car doors while they drive away etc)

You are very kind. Thank you so much! 🙏

Not only that but increasing the quality at the same time this looks way better than most mocap animations I've seen cause of the stitching

You are very kind, thank you so much! Happy to hear you've been enjoying the series!

Professional computer basketball science.

Where can I sign up? 😁

@@AppReviewGerman reach out to 2Immersive4u on Facebook ✌️

Movement is 'obvious'; symbolic language is not as the meaning is arbitrary, working off at least two layers

@@seamusoblainn4603 That is true! I really hope AI could be used to communicate with certain animals in the future. But it is certainly huge and difficult task.

My god, imagine actually interactive 1-on-1 fights that does not have to be prescripted

YES! Please make it comes true! That would be fkg awesome.

Could theoretically do this already in EA Sports UFC 3

If the AIs are as advanced as this, it'd probably be an extremely boring fight to most human observers. In human MMA, most people like it when one guy makes a mistake, then gets punished for it. Equal AI fights would likely lead to ridiculously tactical battles, with very little "action."

@@Fermion. I'd like to see it used to make every fighter slightly different but realistically

That kind of defeats the purpose,I do agree in the pipeline having this would speed things up especially for tricky scenes,but then it becomes "if we could have mimicked the human body this precisely why did we bother animating" .One of animations benefits is its limitations and watching those limitations be broken,but another un part of animation is actually making it.It remind me when directors are asked why they still use 35mm cameras and not just put a filter/overlay to create the effect and the answer is often "the limitaions force focus".Their has been a temphasisi in the years past that animation needs to et more realistic and so does its movement and all of that often excludes that animation can be more than just realsitic.so I just want to say we can have really detailed animation,but it does not need to be in every pipeline.

@@koketsok1513 I think he is referring to animation in the context of video games where realism is absolutely the prime objective (excluding non-realistic art styles of course). I don't think you need to worry about the artistry of animation going anywhere but in the use cases where you want realism above all else it unfortunately isn't possible to accomplish without motion capture and learning algorithm tools to stitch the mocap together.

@@TPF00T I think you are right.In terms of sport/physics based gaming realistic movement would e whats wanted and It could be a benefit to the piplene,but with how many companies like EA are nickle and dimming their player base I doubt they would introduce this.Perhaps as tech like this becomes cheaper to get companies will be held to a higher standard.

Now implement that into Boston Robotics and you will have an basketball playing robot. :D

Where?

They would have to integrate it into the games angine or they migjt need to create an entire new engine to support it. I dont think it would be easy but Im no programmer.

@@EweChewBrrr01 If it saves money developers will probably use it.

and you'd still see reused "sportsball 2k19" assets in a 2k21 game even if 2k developers had this AI, and still charge $70 for the game.

@@SqueezedKittens people who actually buy those yearly sports games are brainless, just the same package in a different box.

@rASP I can think of a few reasons: 1) They first need to know about this even exists 2) Their site says "This project is only for research or education purposes, and not freely available for commercial use or redistribution" 3) There's not much information on it. Is there an easy to use API for it? How do integrate it with your app? How much does it lower performance? What inputs are required and what's the output? Will the outputs even work with my current workflow? Will I have to write even more code to integrate it with my application? 4) Little documentation and adoption. This means if you have any problems you'll have to fix them yourself without being able to easily get the help from those before you. 5) Not every game dev needs 3d animated character. It only makes sense to use this if it will improve the workflow of your project. What technologies you use depends on what type of experience you want. Its because of reasons like those above it makes sense for a lot of developers to not want to invest time into learning to use this technology. However, if someone does manage to make a version of this and wrap it in a nice API, I'm pretty use sure a number of people would use it. (Like an AI denosier feature was added to blender like 1 or so years after 2 minute papers did a video on it and that was a pretty useful for speeding up renders)

lmao imagine your character unconsciously taunts the enemy when you're playing good

Here they are not looking for realism, because render time is important. Simulations have already tackled the materials interactions, and realistic environment could have been easily implemented - but the render time would take a lot of time

@@odyseuszkoskiniotis6266 We're seeing some of these becoming faster and faster

some games already do quite a decent job at facial expressions on personalized character faces, such as black destert online, which isnt the newest game anymore either. But to be honest i hardly see a real use for it, as when do you actually see it? most games show a view of the back of the head 99% of the time, or in first person games that becomes entirely useless. might still be a thing for custom generated cutscenes though. I always appreaciate them more if the use your personal character instead of a plain video sequence. So you could see a real custom reaction of your characters face when he dunks the ball in or something