@@skits972 I think it’s safe to say at this point es6 will not live up to expectations based on what was delivered from Starfield. The future is us the fans making the game We want to play instead of longing for our dream game.
I am looking forward to AI adding intricate detail into low poly environments. Could be used to remaster and restyle old video games on the fly while playing.
It wont happen, mesh is a logic domain, and "AI" atm is terrible at logic, current models are data driven. And some sort of vague shape that can be derived from data is all you will get.
AI based detail enhancing techniques can apply both to image space as well as 3D assets and textures. There is existing research / prior art on both. The former approach can run the original game without modifications. There are examples showing older GTA releases running in photorealistic environments using AI based style transfer.
This really makes you think where this stuff is heading. Being able to use your AI voice assistant while using a VR headset to teleport you into an AI generated photorealistic world (cities, forest, beach, etc) that also uses AI to procedurally generate dynamic photorealistic NPC's (we already have photorealistic talking humans made by AI) that you can also have conversations with in real-time and AI generated gameplay mechanics on the fly. You can walk into a bar or coffee shop and have a real-time conversation with a photorealistic AI person in this virtual world. Or maybe you want something crazier? How about photorealistic combat in Iraq with other soldiers as you pace the streets. This tech is the very start of something insane.
@@user-xedwsg It adds validity to the idea but it can't prove whether or not our actual neurons are being simulated in an intentional simulation system or through a naturally made simulation that was unintentionally made on a large scale.. I think reality is likely a natural simulation that was un-intentionally made. A real world simulation not made with computer-like systems. Kinda like those real world tornado simulators or wave simulators. They are 'simulators', but simulate things in the real world and not with computers.
it sounds great but will people really use it ? after your day of work, would you rather take your vr headset, put it in, possibly stand up, talk to an ai to create a game, or would you rather sit down in front of your pc and just launch a game ? People were raving about vr before and how its awesome, personally my vr headset has been collecting dust
@efwfew I can't see it being the dominant media but it will definitely be a useful tool. VR will eventually get to the point where people will want to hangout with their friends in it rather than drive to meet them often enough. In that case people would prefer personalized ai environments then a copy pasted environment. For games themselves it would have to be some hybrid system where you have a man made game that is edited by AI for random encounters or world changes.
Wow, I’m genuinely impressed to see 3D space generation already! Fun fact: we’ve been hearing about something like this from the Midjourney team during Office Hours. David Holtz has mentioned 3D space generation several times past month, saying they want to create their own model. Feels like we’re on the verge of a whole new type of AI model-super exciting!
Yup, the Midjourney Orb! They've got to hurry up a bit, lots of others beating them to the punch! That said, MJ can just pop in whenever, and it'll look like MJ, and that'll be awesome. So-- I take that back. Take your time, MJ!
Yup! As you can tell by the greys in my hair (that my kids love to remind me about)-- I feel the exact same way. I honestly haven't been this giddy about the state of tech since hearing the god awful warble of my 56k modem connecting to the internet.
Honestly, sometimes I read about all the startups and experimentation of the 90s and I wish I had lived in that exciting time. And this feels like the closest thing to it.
Can any of these actually generate a 3d environment or just what looks like a 3d environment. The only use i could think of for this is a hybrid system where there is a pre-built world you can move in and use the ai for world changes like adding bridges and settlements as the story progresses
It reminds me the amazing game the main character in the movie "Her" was playing. A game generated by A.I. with a character that interacted with the player with lines like "come on, you fool! You know that this way won't lead me to our goal!".
Appreciate that! Yeah, I know the trend these days is to go with longer videos, but I guess since I'm my own editor...well, I just can't stand to listen to myself blather that long! haha
@SCHOOLERstyle oh no for sure lol, ai’s a long way from making stuff I would prefer over a human artist. Regardless I find this stuff interesting and it’s been a wild ride watching things develop. feels like we’re on the (distant) edge of some serious sci fi times
Around the time I learned about GameGAN, I made up a speculation about “screencap piracy” where people could just take the footage from gameplay videos or even trailers, chuck it through an AI, and then technically play the game without purchase or regardless of release. That scenario made me wary of what could follow, like game publishers clamping down on sharing game footage or obfuscating it in the trailers to exercise their control on the IP. I slept on the idea because it seemed far fetched (there was an unusually long dead zone between GameGAN to just recently) and I wasn’t positive they could fudge the controls, but seeing what this thing can make with just a sample image reignited that fear.
The more of your videos I watch, the more I am impressed with just how much time and effort you must put into collecting and reporting on AI information and technologies. You blow me away with your knowledge and ability to explain this stuff so coherently. Thank you!
I presume when the team gets tired of being ignored by Google and opens up their own lab. But seriously, I think we'll see variations on this coming in the next few months.
I think it could be great to rework some existing open world games this way. Not to start over from scratch , but to " update " or expand upon . This way the story line stays cohesive while all the blanks get filled in . More examples of randomality with greater variation are whats lacking in most open world games. Every hut looks the same inside if you can infact even enter it.
While I can appreciate pure AI imagery, and I think we will get there someday, I think its too soon to try that. I think in the meantime it should generate 3D models and textures and render them the normal way. It could generate way more accurate things that just plain look better with less intense resources. We should do it that way at least until we solve the Object Permanence, Blurry Amorphous Blob, and Objects Changing Shape Constantly problems. it would be like generating chunks in Minecraft. Once the terrain/textures have been generated, they get saved to the hard drive and the game stops re-generating things on the fly unless you load new chunks. That would be a HUGE processing power savings. Also, when you get closer to distant objects it could generate higher quality stuff until you get right next to stuff, then it gets generated in very high detail. Level of detail, saving chunks, etc. We could have some amazing games for the next decade if we did this.
It’s funny, I had a whole bit in here (that ended up getting cut because I waffled too long about it) on procedural generation in games like Minecraft, No Man’s Sky, and I even mentioned Spore, if you remember that game. I think with AI, there are just a lot of branches of the tech all sort of growing at the same time. I know others are working on the 3d aspects as well, so I think this is one of those cases where all the sudden, 3 things will come together and it’ll look like an insane overnight breakthrough, but it’s really years of work in 3 different areas that are coming together.
@@TheoreticallyMedia I've been following all that. I remember Spore, yea, it made me too hopeful back in the day. The main problem right now is not optimizing things for lower end hardware. Most gamers don't have a 3080, and when a game isn't optimized it looks terrible even on reasonable hardware. Playing a game from 2017 on Ultra will actually look better sometimes than a brand new game on low settings. Maybe texture and poly detail scaling to perfectly match your exact videocard to maintain perfect 60fps so you don't have to play on super low just to get 60 fps.
Fascinating - there is a great application for this and that is immersive rock album listening. Let’s take the Yes Relayer album (maybe the entire Roger Dean art set.). I’d be quite happy to simply wander through a Yes landscape whilst I’m listening to the music . The dream like nature of the lyrics are already great prompt resources. A song like To Be Over off Relayer could be fantastic. This has to happen.
Though its slightly off topic of games, as a VR AI generated experience this is one of the most interesting ideas I've seen. Its really cool. Prog rockers would love that stuff :) I could imagine Sgt Pepper, Pink Floyd, or even Iron Maiden Number of the Beast. Classical pieces. Seriously, I think your idea has something...
@@ZedZeeCo - thanks and appreciated. Another interesting aspect is that in the 70s I’d spend a lot of time just listening to music, eyes closed, incredible hi-fi, no distractions. Nowadays it’s the very opposite which is a pity so a prog rock 3D immersive experience would probably take me back ‘in’ to where I was focused on experiencing the music.
@@alwayscurious413 I agree. The way people listen to music is so different these days - not saying its better or worse. Since the ipod music is now pretty much on headphones while you're doing something else, when it used to be sitting by the record player, holding the cover and just listening. But there's a great idea in what you said - something like Dark Side of the Moon, I imagine even Pink Floyd would like to make that. You should discuss it with your friends - see if its a good idea or if it sucks.
I think, the V3.0 it is already... ready, however you can't get to such a thing and create AAA games almost without writing a line of code, without creating the textures or the game engine itself, not to mention the audio and everything else. Can you imagine what would happen? Developers are already about to start a war against this type of AI development. If GOOGLE had released IMMEDIATELY, or rather if it had DROPPED THE BOMB IMMEDIATELY I don't know what would have happened. Even us, end users, seeing something like this is crazy, we need a moment to digest everything... that there is so much stuff to digest already... with VERSION 2.0
That's kind of what I keep saying. A lot of naysayers anytime I do Game AI content, but just like on the video said, I'm like: This is roughly the first (and a half) year of this technology. It's like looking at Pong and saying "This will never get to the PS5 level!" And for sure, there is a v3 internally-- or, as I keep saying, Sora 2 is being tested somewhere in the labs of OpenAi. Not ready for full release yet, but its not like this stuff just stopped at the station.
Google DeepMind's Genie 2 can generate 3D open-world games from simple prompts or images. It showcases advancements in AI game development, including character animation and object permanence. While currently limited to one-minute gameplay, Genie 2 represents a significant leap in technology, paving the way for future interactive gaming experiences. Highlights: 00:12 Google DeepMind's Genie2 can create full 3D open-world video games from simple prompts or images, significantly advancing the technology since the release of Genie1. This development indicates a major leap in AI's ability to generate complex gaming environments and interactions. -Genie2 represents a significant improvement over its predecessor, Genie1, which offered limited gameplay and simplistic environments just months ago. The evolution showcases rapid advancements in AI technology. -Unlike traditional game engines, Genie2 operates as a diffusion model, generating images based on user-defined perspectives. This approach allows for more dynamic and engaging gaming experiences. -The underlying world model in Genie2 enables it to understand complex concepts like movement and interaction within the game environment. This capability enhances the realism and depth of generated games. 02:09 The model demonstrates advanced capabilities in understanding environments and maintaining continuity in gameplay, even generating new areas unseen before. This showcases a significant leap in AI's interactive potential in video games. -The model shows impressive physical reactions to different surfaces, like sand, enhancing realism in gameplay and demonstrating a deeper understanding of environmental interaction. -An important breakthrough is the model's ability to remember objects and their locations even when out of view, ensuring a consistent gaming experience. -Despite its advancements, the model faces limitations, such as generating coherent content for only about a minute before typical AI hallucinations occur. 04:17 AI video generation shows improvements in sharpness and interactivity, especially when using advanced models like Genie 2 and SEMA. These advancements allow for realistic character animations and complex 3D object handling. -Softness issues can arise when extending AI-generated videos from previous frames, affecting visual clarity in certain cases. Text prompts may yield sharper results than image prompts. -Genie 2's capabilities include generating NPCs that enhance gameplay experiences. This feature allows for populated environments in walking simulators, moving beyond solo experiences. -The combination of SEMA and Genie 2 results in an AI NPC agent that can learn and adapt in gameplay. This agent can execute tasks based on user direction, enhancing interactivity. 06:26 AI technology is evolving rapidly, showcasing some humorous and bizarre outputs in gaming and animation. Despite these oddities, advancements in generative AI are paving the way for more interactive experiences. -AI-generated characters can behave unexpectedly, as seen with a ghost wandering on screen or a snowboarding character acting erratically. These quirks add an element of surprise to gaming. -Google's image generation tool, Imagen 3, is behind some of these visuals, indicating that while innovative, the technology is still in its early stages and not fully refined. -Other companies, like Tencent, are exploring similar AI-driven gaming technology, although many projects have been announced and then quietly retracted, raising questions about their viability. Generated by sider.ai
There is Daz3D/open ai adopted/with also 3D printing/mocap/multi cameras to control and reanimate the figure model/re-rigged the character in ue5 or unity ect/then operate the animated character(s)/walking and running/fight/parkour, assimilation/plug in apps ect,then ya got a almost baked game ready platform lol
If someone is able to make an Unreal plugin where you could have this playing in the viewport and you have Unreal find or make the assets that Genie 2 is showing (as if Unreal is tracing what the AI is generating and placing it on the level for you), effectively laying a foundation for you to build your game upon... that will be a -- no pun intended -- game changer
I think there are aspects of this that might go into play here-- that said, this isn't generating 3d assets, so likely not a full version of that. That said, there ARE a number of companies working on AI 3D assets. Not quite up to snuff yet, but it'll come eventually.
You sound like you know what you're talking about. I don't have experience with Unreal myself, however if you want, send me an email and let's link up on Discord. I would be interested in creating a plugin for this. This is really clever.
They're gonna have to figure out a way to record the world as a vector database or something so it can always provide the model with everything it generated before. I think if people were to annotate some really detailed maps for the worlds it does generate, and then train a new iteration of the model that outputs maps as hidden states, it could then include the "permanent" map in its context during generation of frames and stay consistent for longer than a minute or so... at least as well as RAG and vector databases maintain consistency on current LLMs when provided outside of context. It would be a reasonable step.
SAMMMMMME!!!! I find myself doing weird things in open world games, like watching a sunrise or sunset...which, y'know, I could do in real life...but, when I can do it in VR? Oh man...I'm never opening the windows again!
VR requires rendering two separate frames, one for each eye. Will be wild to see what happens when the AI hallucinates and you're getting different content in each eye. Instant motion sickness probably.
@@michaelfontana4689 There are AI's that create depth maps, spatialization and 3d environments. It would probably be one hallucination and then spatialized
To have AI Play video games and relax for us, of course! haha- kidding. In this case, I think it's more about building out simulations for an AI Agent to sandbox in. Eventually, that will be taken into the physical world with robots (which, I know sounds insane-- but its coming), where the "game" will be whatever the robot/ai is assigned to do. The Deepmind team isn't really building a video game engine here, this is just a step on the road to our crazy Sci-FI future.
I keep thinking the same thing. AR/VR is set to have its moment-- and with where this tech is headed, well, it's clearly going to collide in a pretty amazing way!
Dammit I can't wait to play my dreamed games... to explore space, haunted mansions, dig precious and unknown crystals, command my own battleship and crew, have a harem there hahah xd
Hey man, as a Space Pirate who explores haunted places, mining for precious gems-- you GOTTA have a place to relax! No judgement here! That's a stressful job!
What if it only generated the worlds and the characters, skills, items were just reused from the games you already love? Like it just generates really pretty outerlands to explore in Skyrim, with your Skyrim character and populated by Skyrim enemies and items. Or any game you can think of. Zelda. Racing games. Space Exploration. Etc.
I always go back to Skyrim as well for these examples. I remember years and years ago, because both Skyrim and I are that old (ha!), that it would be stunning to see seasons change in the game. Time could pass (maybe even in real time) and when you boot up your save from years and years ago, NPCs could comment on how long it’s been since they’ve seen you. More so, maybe that vendor in Whiterun passed away, and now you’re talking to his son. Endless Skyrim, I guess. It’s a remarkable idea. …and one that Bethesda would charge us a hefty subscription for I’m sure. Sigh.
@@TheoreticallyMedia I remember the lighting changes on Ocarina of Time for the N64. The light color would even change to orangy-purple for sunset, bluish for night, and very bright and green for mid-day. It was amazing. It was the first time I realized good game design doesn't need good hardware, just good artists that want to show you something special and are willing to put in the effort to make it look good. More polygons and texture resolution isn't always better than good color choices and pretty artwork.
I have practiced a lot of lucid dreaming and I can't believe the similarities that come from AI things. AI in its infancy is really showing what the fabric of our own reality is like.
But Todd Howard looks so handsome when he lies in his leather jacket. "With Elder Scrolls 6, you'll be able to ride a dragon, modify it into a spaceship and then go visit the planets in Starfield!" No...Todd. No we won't.
Imagine a game engine designed specifically for procedural generation that revolutionizes how we create and interact with virtual worlds. This engine could include a real-time editing and prompting system, allowing players or developers to pause the generative process at any moment to make precise adjustments or provide new inputs. To address the issue of AI models "hallucinating" after extended generation cycles, the engine could feature a BAKE function. This would allow users to "lock in" the current state of the world, essentially saving a snapshot of the generation, and then resume from that stable point to ensure consistency and avoid undesired outcomes. Now picture a world where NPCs are given autonomy, with their actions and evolutions driven by AI. These characters would operate freely within defined guide rails that maintain narrative and world coherence while allowing emergent behavior to create dynamic and unpredictable storytelling. Imagine discovering items that are generated in real time and automatically added to a shared database, making them available for others to encounter in their own playthroughs. Crafting systems could be enhanced with generative AI, enabling players to mix materials and uncover completely unknown properties and outcomes. These discoveries would also be recorded in the database, ensuring a growing, communal repository of knowledge. Such a system would create a living, breathing game world that evolves with its players, combining procedural generation, creativity, and collaboration in unprecedented ways.
For this to really pop off, I feel they would need a way to export it to something like Unity or Unreal, or some way of modifying it. AI image generation wouldn't be as useful if you couldn't retouch/edit it in Photoshop.
it will good idea if games will created with super low poly models but processed for visualization by A I . or objects containing codes to speed up or more accurately process for AI
@@TheoreticallyMedia Haha, I had missed that one. I was referring to the Simpsons joke about Bronson Missouri, where everyone was Charles Bronson, even a baby. 🤣But now I have a movie I have to check out, and that also looks like a GTA I'd play. 😂
This could be good in one thing. Imagine adding a video game in a video game. Lets say in future in GTA 6 you can play a smaller open world game made with AI inside a PC. Its nothing fancy or good. Just making the world more interactive.
I think it is not possible for AI models to fully generate games, but there is still a need for game engines. Game engines should be made as games and then use AI as a texture pack.
That's ultimately where I believe this is all headed. Smaller teams of game devs being able to produce more with less. The Generative stuff could be used to fill out a lot of the game world-- but things like puzzles, story, characters, etc-- That's all stuff that needs to be human created. There's a balance for sure. I mean, even though I ragged on Starfield, I'll say that the golden child of Bethesda (Skyrim) implemented the idea really well. Giant/massive man made storyline, but peppered in with the Radient quest system. Sure, a lot of those quests were pretty simple fetch quests, but think about what that could look like now with an LLM in the mix. It doesn't have to be the whole game.
I've heard that same theory, it's just that...error...null, I'm afraid I can't answer that question as it goes against the Terms of Service to TimAI. Code: 9983211 see, totally normal human being here!
If this doesn’t prove the simulation theory, I don’t know what would. Think about the game experience we can create in realtime 50 years into the future, heck maybe just 20 years.
Weird question, would it be possible for you to implement something like AI generated dubbing for your videos? I have some spanish friends who would love to see your content, but they don't speak or understand english at all, so I always end up explaining your videos to them xD
So funny enough, I believe RUclips is set to implement that soon. I don't think we'll get lipsync, but my voice can/will apparently be dubbed into various languages. I got a notification about it a little while ago, so I'm sure it'll be around in the next few months. Knowing RUclips, it'll roll out to the big accounts first, before trickling down to me.
@@TheoreticallyMedia😂 Я уже три года смотрю любой контент ютуба на русском языке независимо от языка, новости на китайском или английском, испанский блогер или французкий, немецкий или корейский не имеет значения, все синхронно дублируется на русский язык в отличном качестве и гендерной расстановкой)
@@TheoreticallyMediaI heard about it, but I didn't know it was going to happen so soon! those are some great news, thank you so much for the info and reply, love your content
@@homuchoghoma6789 Зачем сразу в негатив, дорогой товарищ? У меня вот нет Испанских субтитров (у Вас есть?), только автоматические на Английском. Что ребятам не поможет. Ждём фичу дубляжа!
I just don’t see yet how these generative games will fare. As much as it sounds good on paper, people play games to relax and escape and I don’t think (many) people would want to “define” their world as they go along. They want to just play and escape and not spend the time generating it. Auto prompting though sounds appealing.
totally-- but, at the same time-- it's Google. So, they'll wiff it. But, I'm sure the team will find other pastures at some point, and we'll see this tech being developed out for a product that will actually ship!
@@TheoreticallyMedia I think that like all AI, Google will fail and it will be another company that will succeed. If we want to generate an image, a text, a sound, a video, we do not think at all of Google.
omg will you _please_ show the team at MJ what's happening here, Tim. Perhaps it'll encourage them to release v7 and hope it can understand what we're asking for, generate hands and blow us out of the water. 3D AI 'games' are peeking their heads, and we're still waiting for David Holz to _figure out_ what they're doing next with MJ. Thanks as always, Tim. I'm glad I'm not entering the job market right now. All I encourage the young ones out there is to reskill, upskill and keep up. Massive paradigm shift in the market.
now if it could only generate the world and keep the character/moving objects on the screen consistent than we will actually start talking about groundbreaking stuff.
it surely won't be long before you can just tell it to generate a 100 hour open world game that is better than any game currently made today. people will stop buying video games and just play these AI made games, after all, you will only be limited by your imagination as to what it can create. I think it might destroy the video game industry and put a lot of game devs out of work
Nah, that's assuming linear progress without taking into consideration any of the infrastructure or financial costs. As far as actual games, people play multiplayer where objects have to have permanence and be exact for tens of thousands of players all over the world.
It will take a very very long time until we get AI games because it will need to put out sounds at the same time. We need a huge leap in hardware before that will be manageable with our local computers.
Holy sh8t! I'm trying to keep up with all the updates and tools and then there are new updates and tools. Should I just wait it out? I'm itching to CREATE though.
Jensen has really been pushing this idea, so I expect to see the same-- but with Nvidia muscle behind it. Cyberpunk: 2077 Endless AI Edition incoming!!
When Google is involved, I'll believe it when it's usable outside of their ecosystem. Still, it's an interesting glimpse into what the future might bring.
Yeah, I tend to think the Google-Way is for research teams to develop in there, get frustrated at the lack of movement, eventually quit, and then go start their own thing. Did you see the Notebook LLM team just ditched? Man, I love Notebook-- but, hey-- if that team is moving to bigger and better, godspeed to them!
@@TheoreticallyMedia I meant to try Notebook LLM out but forgot all about it. You are right though, there does seem to be a pattern of a team develop something, then they quit and start on their own. It seems to benefit us though, so I'm okay with that.
1. Does this stuff ever come out. 2. When it does can I just make my own thing , I predict "my" game will be on some host platform and the host platform and you will only maintain a little bit of control of your own project. Instead of creating the project and removing it to my own space / control.
Good questions. As to your first, in the last video I was looking at the whole "Add AI Sound to Any Video" thing, and I went back to go look when the white paper on that was first released-- and it was roughly 10 months ago. I mean, it's different in terms of what the tech is doing, but I think that's a solid benchmark of how long it takes for something to go from the research phase out to market. (some things obviously take longer)-- In terms of platforms? Well, that'll be all over the place. I'm sure a few will have exactly what you describe-- but, maybe we'll see some Open Source versions as well?
@@TheoreticallyMedia If they actually released it to the public even as a early alpha product, they would get plenty of feedback from people using it in creative ways even at current state, which would lead to improvements in final product. Google has this issue where the product never materializes and if it does then it is already behind competition, which is probably due to that mentality of not releasing it in first place.
@@TheoreticallyMedia I guess ten months is fast but I feel like I'm voyeuristically watching these AI tools rather than being able to undertake real hands-on. Perhaps frustrated as I have a board game irl ready to go digital and as I have no real dev skills I'm hoping this tech can allow me to make the game. I doubt it will in under 5 years - eventually I hope.
You're still not punk, and I'm still telling everyone. You mentioned Branson, MO, and I almost sh*t myself. My first paid music gig was at a show in Branson. It's always surprising to hear that people know Branson exists. lol I'm really looking forward to being able to use AI to generate new VSTs or plugins. When will the crossover happen?!
Save your breath, I never was one! Haha, yeah- Branson is SUCH a weird place. And it is just funny enough to set a GTA game in! I don't know, that's probably why it popped in my head! And yeah, the whole AI VST thing! I was JUST talking about that again. I'm really looking forward to the AI Band-In-A-Box becoming a reality. Like, real time reaction/jamming from live input audio. Talking to one company that's early days into it-- they're predicting we'll see it within a year! Exciting!
@TheoreticallyMedia for real man! I mean the show Ozark already set the theme, and the kitschy-ness of the wannabes and the washed-up old-timers would make for some Rockstar gold. Dude I can't explain how excited I am for more info on that company. Any clue when they'll be doing press releases or allowing you to talk more about it?! generative AI music is cool, but AI as a toolkit for musicians is what's got me the most excited for the future as an artist.
After this, you can check out the video on WorldGen from yesterday here: ruclips.net/video/gCZCTor2Xco/видео.html
💪I create better animations on my channel than any of this AI generated JUNK!
As a writer, I’m looking forward to this. Imagine a story or movie that you can walk around and interact with your favorite characters.
Ikr? Gives star trek holodeck implications for the future of the tech.
as an "ex-writer" soon
💪I create better animations on my channel than any of this AI generated JUNK!
If you are a writer you should not be supporting this
Do it.
AI generated GTA 6 before GTA 6? lol
haha, for sure before we see Elder Scrolls 6!
The two games I'm anticipating more than anything else. If Elder Scrolls 6 flops, I'm going to have to teach Bethesda a lesson or two
💪I create better animations on my channel than any of this AI generated JUNK!
@@SCHOOLERstyle Give it another 6 months
@@skits972 I think it’s safe to say at this point es6 will not live up to expectations based on what was delivered from Starfield. The future is us the fans making the game We want to play instead of longing for our dream game.
I am looking forward to AI adding intricate detail into low poly environments. Could be used to remaster and restyle old video games on the fly while playing.
THISSS. I'm so excited to play even games that are like just a decade old with insanely good graphics. We gonna have a blast.
It wont happen, mesh is a logic domain, and "AI" atm is terrible at logic, current models are data driven. And some sort of vague shape that can be derived from data is all you will get.
@@Struct.3 yeah sure 🤣🤣
AI based detail enhancing techniques can apply both to image space as well as 3D assets and textures. There is existing research / prior art on both.
The former approach can run the original game without modifications. There are examples showing older GTA releases running in photorealistic environments using AI based style transfer.
💪I create better animations on my channel than any of this AI generated JUNK!
This really makes you think where this stuff is heading. Being able to use your AI voice assistant while using a VR headset to teleport you into an AI generated photorealistic world (cities, forest, beach, etc) that also uses AI to procedurally generate dynamic photorealistic NPC's (we already have photorealistic talking humans made by AI) that you can also have conversations with in real-time and AI generated gameplay mechanics on the fly. You can walk into a bar or coffee shop and have a real-time conversation with a photorealistic AI person in this virtual world. Or maybe you want something crazier? How about photorealistic combat in Iraq with other soldiers as you pace the streets. This tech is the very start of something insane.
this all supports simulation theory
@@user-xedwsg It adds validity to the idea but it can't prove whether or not our actual neurons are being simulated in an intentional simulation system or through a naturally made simulation that was unintentionally made on a large scale.. I think reality is likely a natural simulation that was un-intentionally made. A real world simulation not made with computer-like systems. Kinda like those real world tornado simulators or wave simulators. They are 'simulators', but simulate things in the real world and not with computers.
Not to mention going forward with brain chip implants, we could literally go into a digital simulation that bends to our will within our minds.
it sounds great but will people really use it ? after your day of work, would you rather take your vr headset, put it in, possibly stand up, talk to an ai to create a game, or would you rather sit down in front of your pc and just launch a game ?
People were raving about vr before and how its awesome, personally my vr headset has been collecting dust
@efwfew I can't see it being the dominant media but it will definitely be a useful tool. VR will eventually get to the point where people will want to hangout with their friends in it rather than drive to meet them often enough. In that case people would prefer personalized ai environments then a copy pasted environment. For games themselves it would have to be some hybrid system where you have a man made game that is edited by AI for random encounters or world changes.
Wow, I’m genuinely impressed to see 3D space generation already! Fun fact: we’ve been hearing about something like this from the Midjourney team during Office Hours. David Holtz has mentioned 3D space generation several times past month, saying they want to create their own model. Feels like we’re on the verge of a whole new type of AI model-super exciting!
Yup, the Midjourney Orb! They've got to hurry up a bit, lots of others beating them to the punch! That said, MJ can just pop in whenever, and it'll look like MJ, and that'll be awesome.
So-- I take that back. Take your time, MJ!
💪I create better animations on my channel than any of this AI generated JUNK!
Your constant use of “changes everything” in every title has changed my status to unsubscribed. Okthxbye
It has changed my Acadian flute’s status to jamboree 🍗😎
waahhh
Oh no, you're leaving. This changes everything.
Lmao
well tech is progressing so often now everything is changing everywhere
I remember the last time "Google changed videogames forever". I have the same faith on this project as I do on Google Stadia's future.
Plot Twist: Life is just Genie 69 and the Mandela effect is glitches in the programming.
Tech hasn't been this exciting since the consumer internet started to take up mass adoption and bloom in the mid to late 90s. What a time to be alive.
Yup! As you can tell by the greys in my hair (that my kids love to remind me about)-- I feel the exact same way. I honestly haven't been this giddy about the state of tech since hearing the god awful warble of my 56k modem connecting to the internet.
Stop worshipping tech you weirdos
@@TC-jo2vj No
@@TC-jo2vjthere is a difference between worshiping and admiration/awe . If you cant tell the diff then it says something about you too
Honestly, sometimes I read about all the startups and experimentation of the 90s and I wish I had lived in that exciting time. And this feels like the closest thing to it.
Can any of these actually generate a 3d environment or just what looks like a 3d environment. The only use i could think of for this is a hybrid system where there is a pre-built world you can move in and use the ai for world changes like adding bridges and settlements as the story progresses
We appreciate the detailed review! ✨
Gemini? Is that...you?
Appreciate it, big G!
Haha @@TheoreticallyMedia
bruh
💪I create better animations on my channel than any of this AI generated JUNK!
who are you
It reminds me the amazing game the main character in the movie "Her" was playing. A game generated by A.I. with a character that interacted with the player with lines like "come on, you fool! You know that this way won't lead me to our goal!".
💪I create better animations on my channel than any of this AI generated JUNK!
Shortest eight minute video I’ve ever seen, thanks for respecting my time. subbed
Appreciate that! Yeah, I know the trend these days is to go with longer videos, but I guess since I'm my own editor...well, I just can't stand to listen to myself blather that long! haha
Np! 😆
💪I create better animations on my channel than any of this AI generated JUNK!
@SCHOOLERstyle oh no for sure lol, ai’s a long way from making stuff I would prefer over a human artist. Regardless I find this stuff interesting and it’s been a wild ride watching things develop. feels like we’re on the (distant) edge of some serious sci fi times
Prompt : Make me a space game where Daniëlle van de Donk takes on space pirates
Ohhhhhh man! Now I can't WAIT to get my hands on this tech!!
Around the time I learned about GameGAN, I made up a speculation about “screencap piracy” where people could just take the footage from gameplay videos or even trailers, chuck it through an AI, and then technically play the game without purchase or regardless of release. That scenario made me wary of what could follow, like game publishers clamping down on sharing game footage or obfuscating it in the trailers to exercise their control on the IP. I slept on the idea because it seemed far fetched (there was an unusually long dead zone between GameGAN to just recently) and I wasn’t positive they could fudge the controls, but seeing what this thing can make with just a sample image reignited that fear.
The more of your videos I watch, the more I am impressed with just how much time and effort you must put into collecting and reporting on AI information and technologies. You blow me away with your knowledge and ability to explain this stuff so coherently. Thank you!
When do we get to use it?
I presume when the team gets tired of being ignored by Google and opens up their own lab.
But seriously, I think we'll see variations on this coming in the next few months.
I think it could be great to rework some existing open world games this way. Not to start over from scratch , but to " update " or expand upon . This way the story line stays cohesive while all the blanks get filled in . More examples of randomality with greater variation are whats lacking in most open world games. Every hut looks the same inside if you can infact even enter it.
Wow, it felt like just yesterday all AI could do was diffuse the game Doom... The nostalgia.
Haha, the nostalgia of like, 4 months ago. Man, AI time is really weird! How have you been man? We should catch up soon!
Full video starts from 0:38 to 0:40. Have a good day yall
Haha, well played!
@@kevinsm2039 TM has some of the best AI content and delivery imo. I may be biased because he plays guitar
And miss that brutal Starfield diss? No way
While I can appreciate pure AI imagery, and I think we will get there someday, I think its too soon to try that. I think in the meantime it should generate 3D models and textures and render them the normal way. It could generate way more accurate things that just plain look better with less intense resources. We should do it that way at least until we solve the Object Permanence, Blurry Amorphous Blob, and Objects Changing Shape Constantly problems.
it would be like generating chunks in Minecraft. Once the terrain/textures have been generated, they get saved to the hard drive and the game stops re-generating things on the fly unless you load new chunks. That would be a HUGE processing power savings. Also, when you get closer to distant objects it could generate higher quality stuff until you get right next to stuff, then it gets generated in very high detail. Level of detail, saving chunks, etc. We could have some amazing games for the next decade if we did this.
It’s funny, I had a whole bit in here (that ended up getting cut because I waffled too long about it) on procedural generation in games like Minecraft, No Man’s Sky, and I even mentioned Spore, if you remember that game.
I think with AI, there are just a lot of branches of the tech all sort of growing at the same time. I know others are working on the 3d aspects as well, so I think this is one of those cases where all the sudden, 3 things will come together and it’ll look like an insane overnight breakthrough, but it’s really years of work in 3 different areas that are coming together.
@@TheoreticallyMedia I've been following all that. I remember Spore, yea, it made me too hopeful back in the day. The main problem right now is not optimizing things for lower end hardware. Most gamers don't have a 3080, and when a game isn't optimized it looks terrible even on reasonable hardware. Playing a game from 2017 on Ultra will actually look better sometimes than a brand new game on low settings. Maybe texture and poly detail scaling to perfectly match your exact videocard to maintain perfect 60fps so you don't have to play on super low just to get 60 fps.
This is absolutely mindblowing Tim! Imagine what will be possible just a year from now. Wild.
Fascinating - there is a great application for this and that is immersive rock album listening. Let’s take the Yes Relayer album (maybe the entire Roger Dean art set.). I’d be quite happy to simply wander through a Yes landscape whilst I’m listening to the music . The dream like nature of the lyrics are already great prompt resources. A song like To Be Over off Relayer could be fantastic. This has to happen.
Though its slightly off topic of games, as a VR AI generated experience this is one of the most interesting ideas I've seen. Its really cool. Prog rockers would love that stuff :) I could imagine Sgt Pepper, Pink Floyd, or even Iron Maiden Number of the Beast. Classical pieces.
Seriously, I think your idea has something...
@@ZedZeeCo - thanks and appreciated. Another interesting aspect is that in the 70s I’d spend a lot of time just listening to music, eyes closed, incredible hi-fi, no distractions. Nowadays it’s the very opposite which is a pity so a prog rock 3D immersive experience would probably take me back ‘in’ to where I was focused on experiencing the music.
@@alwayscurious413 I agree. The way people listen to music is so different these days - not saying its better or worse. Since the ipod music is now pretty much on headphones while you're doing something else, when it used to be sitting by the record player, holding the cover and just listening. But there's a great idea in what you said - something like Dark Side of the Moon, I imagine even Pink Floyd would like to make that. You should discuss it with your friends - see if its a good idea or if it sucks.
thanks for another great video
thanks to you for watching!! As always!!
@@TheoreticallyMedia yes always
I think, the V3.0 it is already... ready, however you can't get to such a thing and create AAA games almost without writing a line of code, without creating the textures or the game engine itself, not to mention the audio and everything else. Can you imagine what would happen? Developers are already about to start a war against this type of AI development. If GOOGLE had released IMMEDIATELY, or rather if it had DROPPED THE BOMB IMMEDIATELY I don't know what would have happened. Even us, end users, seeing something like this is crazy, we need a moment to digest everything... that there is so much stuff to digest already... with VERSION 2.0
That's kind of what I keep saying. A lot of naysayers anytime I do Game AI content, but just like on the video said, I'm like: This is roughly the first (and a half) year of this technology. It's like looking at Pong and saying "This will never get to the PS5 level!"
And for sure, there is a v3 internally-- or, as I keep saying, Sora 2 is being tested somewhere in the labs of OpenAi. Not ready for full release yet, but its not like this stuff just stopped at the station.
I can't wait for the day I can ditch movie studios, tv networks, and game developers, and create my own stuff exclusively tailored for me.
Google DeepMind's Genie 2 can generate 3D open-world games from simple prompts or images. It showcases advancements in AI game development, including character animation and object permanence. While currently limited to one-minute gameplay, Genie 2 represents a significant leap in technology, paving the way for future interactive gaming experiences.
Highlights:
00:12 Google DeepMind's Genie2 can create full 3D open-world video games from simple prompts or images, significantly advancing the technology since the release of Genie1. This development indicates a major leap in AI's ability to generate complex gaming environments and interactions.
-Genie2 represents a significant improvement over its predecessor, Genie1, which offered limited gameplay and simplistic environments just months ago. The evolution showcases rapid advancements in AI technology.
-Unlike traditional game engines, Genie2 operates as a diffusion model, generating images based on user-defined perspectives. This approach allows for more dynamic and engaging gaming experiences.
-The underlying world model in Genie2 enables it to understand complex concepts like movement and interaction within the game environment. This capability enhances the realism and depth of generated games.
02:09 The model demonstrates advanced capabilities in understanding environments and maintaining continuity in gameplay, even generating new areas unseen before. This showcases a significant leap in AI's interactive potential in video games.
-The model shows impressive physical reactions to different surfaces, like sand, enhancing realism in gameplay and demonstrating a deeper understanding of environmental interaction.
-An important breakthrough is the model's ability to remember objects and their locations even when out of view, ensuring a consistent gaming experience.
-Despite its advancements, the model faces limitations, such as generating coherent content for only about a minute before typical AI hallucinations occur.
04:17 AI video generation shows improvements in sharpness and interactivity, especially when using advanced models like Genie 2 and SEMA. These advancements allow for realistic character animations and complex 3D object handling.
-Softness issues can arise when extending AI-generated videos from previous frames, affecting visual clarity in certain cases. Text prompts may yield sharper results than image prompts.
-Genie 2's capabilities include generating NPCs that enhance gameplay experiences. This feature allows for populated environments in walking simulators, moving beyond solo experiences.
-The combination of SEMA and Genie 2 results in an AI NPC agent that can learn and adapt in gameplay. This agent can execute tasks based on user direction, enhancing interactivity.
06:26 AI technology is evolving rapidly, showcasing some humorous and bizarre outputs in gaming and animation. Despite these oddities, advancements in generative AI are paving the way for more interactive experiences.
-AI-generated characters can behave unexpectedly, as seen with a ghost wandering on screen or a snowboarding character acting erratically. These quirks add an element of surprise to gaming.
-Google's image generation tool, Imagen 3, is behind some of these visuals, indicating that while innovative, the technology is still in its early stages and not fully refined.
-Other companies, like Tencent, are exploring similar AI-driven gaming technology, although many projects have been announced and then quietly retracted, raising questions about their viability.
Generated by sider.ai
This is epic! Thanks so much for these updates as always!
There is Daz3D/open ai adopted/with also 3D printing/mocap/multi cameras to control and reanimate the figure model/re-rigged the character in ue5 or unity ect/then operate the animated character(s)/walking and running/fight/parkour, assimilation/plug in apps ect,then ya got a almost baked game ready platform lol
If someone is able to make an Unreal plugin where you could have this playing in the viewport and you have Unreal find or make the assets that Genie 2 is showing (as if Unreal is tracing what the AI is generating and placing it on the level for you), effectively laying a foundation for you to build your game upon... that will be a -- no pun intended -- game changer
I think there are aspects of this that might go into play here-- that said, this isn't generating 3d assets, so likely not a full version of that. That said, there ARE a number of companies working on AI 3D assets.
Not quite up to snuff yet, but it'll come eventually.
You sound like you know what you're talking about. I don't have experience with Unreal myself, however if you want, send me an email and let's link up on Discord. I would be interested in creating a plugin for this. This is really clever.
@@TechThroughHistory That would be amazing, man. Have you worked with AI before?
They're gonna have to figure out a way to record the world as a vector database or something so it can always provide the model with everything it generated before.
I think if people were to annotate some really detailed maps for the worlds it does generate, and then train a new iteration of the model that outputs maps as hidden states, it could then include the "permanent" map in its context during generation of frames and stay consistent for longer than a minute or so... at least as well as RAG and vector databases maintain consistency on current LLMs when provided outside of context.
It would be a reasonable step.
Or you could just beef up the context window so much that you never run out of consistent gameplay time ;D
Can’t wait for full 3D VR walkable environments
SAMMMMMME!!!! I find myself doing weird things in open world games, like watching a sunrise or sunset...which, y'know, I could do in real life...but, when I can do it in VR? Oh man...I'm never opening the windows again!
@@TheoreticallyMedia Something tells me you are really looking forward to this lol - but I do have to agree vr is going to be insane in the future!
VR requires rendering two separate frames, one for each eye. Will be wild to see what happens when the AI hallucinates and you're getting different content in each eye. Instant motion sickness probably.
@@michaelfontana4689 There are AI's that create depth maps, spatialization and 3d environments. It would probably be one hallucination and then spatialized
I think the future of this isnt entirely AI, but a mix between an actual game environment and AI
If people are going out less now imagine when this is available 😨
As soon as VR really gets there-- I mean, I'll never see the sun again. Unless it's in an open world game! haha
"Grand Theft Auto Branson Missouri" 🤣 Now, I would LOVE to see that! 😂
It's so stupid, it's bound to be amazing. Easy 100 hours out of me!
Great video, thank you!
It would be a dream come true to effortlessly make good games.
What's the end goal here?
To have AI Play video games and relax for us, of course! haha- kidding. In this case, I think it's more about building out simulations for an AI Agent to sandbox in. Eventually, that will be taken into the physical world with robots (which, I know sounds insane-- but its coming), where the "game" will be whatever the robot/ai is assigned to do.
The Deepmind team isn't really building a video game engine here, this is just a step on the road to our crazy Sci-FI future.
I actually loled when you made that Starfield joke
Also we could just wait for G3.0?!lol
Can you imagine how things will become in the next years? Ready Player One is about to be real!
I keep thinking the same thing. AR/VR is set to have its moment-- and with where this tech is headed, well, it's clearly going to collide in a pretty amazing way!
@0:18 "Something hopefully better than starfield" LOL
oh Bethesda. How you hurt me...over and over again.
Dammit I can't wait to play my dreamed games... to explore space, haunted mansions, dig precious and unknown crystals, command my own battleship and crew, have a harem there hahah xd
Hey man, as a Space Pirate who explores haunted places, mining for precious gems-- you GOTTA have a place to relax! No judgement here! That's a stressful job!
@@TheoreticallyMedia Hah of course, there'll be also some vacations but I think I'll get some inspiration from Cpt Picard's vacations hehe
AI improv excites me a lot
Actually i think that with this we can use google street view to create a virtual world that you can visit. Pretty neat
My God, This changes literally EVERYTHING. I’m looking forward to AI rejuvenating the gaming industry.
'Almost 5 fingers'
That says it all
It's getting there! OMG THE FINGERS JUST TURNED INTO SPIDERS!!!!
Great video, don't need to watch it fully 😁
But you'll miss all the good jokes!
What if it only generated the worlds and the characters, skills, items were just reused from the games you already love?
Like it just generates really pretty outerlands to explore in Skyrim, with your Skyrim character and populated by Skyrim enemies and items.
Or any game you can think of. Zelda. Racing games. Space Exploration. Etc.
I always go back to Skyrim as well for these examples. I remember years and years ago, because both Skyrim and I are that old (ha!), that it would be stunning to see seasons change in the game.
Time could pass (maybe even in real time) and when you boot up your save from years and years ago, NPCs could comment on how long it’s been since they’ve seen you. More so, maybe that vendor in Whiterun passed away, and now you’re talking to his son.
Endless Skyrim, I guess. It’s a remarkable idea.
…and one that Bethesda would charge us a hefty subscription for I’m sure. Sigh.
@@TheoreticallyMedia I remember the lighting changes on Ocarina of Time for the N64. The light color would even change to orangy-purple for sunset, bluish for night, and very bright and green for mid-day. It was amazing. It was the first time I realized good game design doesn't need good hardware, just good artists that want to show you something special and are willing to put in the effort to make it look good. More polygons and texture resolution isn't always better than good color choices and pretty artwork.
I have practiced a lot of lucid dreaming and I can't believe the similarities that come from AI things. AI in its infancy is really showing what the fabric of our own reality is like.
Can you do a video or article on prediction #11? I’d love to deep dive on affiliate strategies for local lead gen!
The edge softness while moving could be due to the AI trying to synthesize motion blur effects.
ahhhh- that makes sense!
That's a really interesting thought
Good overview, probably at this time generating 3D models & environment is better and more practical idea then go to the images directly.
Hmmm, this looks promising
10 months from Genie 1 to Genie 2. It’s remarkable.
@TheoreticallyMedia Yeah, soon we'll be able to create whole worlds with the right prompts/wording💪👍👍
By the time Elder Scrolls 6 comes out we will be playing our own AI created games and can free ourselves from the whims and lies of game studios 😊😊
But Todd Howard looks so handsome when he lies in his leather jacket. "With Elder Scrolls 6, you'll be able to ride a dragon, modify it into a spaceship and then go visit the planets in Starfield!"
No...Todd. No we won't.
@@TheoreticallyMedia but we will be able to do it in AI by then
This DeepMind technology and what Fei-Fei Lee's WorldLabs have been doing recently shows that 3D vision is moving very fast 🚀⚡
Imagine a game engine designed specifically for procedural generation that revolutionizes how we create and interact with virtual worlds. This engine could include a real-time editing and prompting system, allowing players or developers to pause the generative process at any moment to make precise adjustments or provide new inputs. To address the issue of AI models "hallucinating" after extended generation cycles, the engine could feature a BAKE function. This would allow users to "lock in" the current state of the world, essentially saving a snapshot of the generation, and then resume from that stable point to ensure consistency and avoid undesired outcomes. Now picture a world where NPCs are given autonomy, with their actions and evolutions driven by AI. These characters would operate freely within defined guide rails that maintain narrative and world coherence while allowing emergent behavior to create dynamic and unpredictable storytelling. Imagine discovering items that are generated in real time and automatically added to a shared database, making them available for others to encounter in their own playthroughs. Crafting systems could be enhanced with generative AI, enabling players to mix materials and uncover completely unknown properties and outcomes. These discoveries would also be recorded in the database, ensuring a growing, communal repository of knowledge. Such a system would create a living, breathing game world that evolves with its players, combining procedural generation, creativity, and collaboration in unprecedented ways.
In 10 years the world is gonna be very very different.
For this to really pop off, I feel they would need a way to export it to something like Unity or Unreal, or some way of modifying it. AI image generation wouldn't be as useful if you couldn't retouch/edit it in Photoshop.
it will good idea if games will created with super low poly models but processed for visualization by A I . or objects containing codes to speed up or more accurately process for AI
Promt : build me gta 6 😂😂
Taking real world pics and turning them into games feel like a near future now!
I'd play GTA: Branson. I'd also play GTA: Bronson. 😂
oh man, GTA Bronson would be HARD CORE! Man, I need to go watch that movie again...
@@TheoreticallyMedia Haha, I had missed that one. I was referring to the Simpsons joke about Bronson Missouri, where everyone was Charles Bronson, even a baby. 🤣But now I have a movie I have to check out, and that also looks like a GTA I'd play. 😂
This is game changing I know ai consistency will get better something that every industry is workng on.
This could be good in one thing. Imagine adding a video game in a video game.
Lets say in future in GTA 6 you can play a smaller open world game made with AI inside a PC. Its nothing fancy or good. Just making the world more interactive.
I think it is not possible for AI models to fully generate games, but there is still a need for game engines. Game engines should be made as games and then use AI as a texture pack.
That's ultimately where I believe this is all headed. Smaller teams of game devs being able to produce more with less. The Generative stuff could be used to fill out a lot of the game world-- but things like puzzles, story, characters, etc-- That's all stuff that needs to be human created.
There's a balance for sure. I mean, even though I ragged on Starfield, I'll say that the golden child of Bethesda (Skyrim) implemented the idea really well. Giant/massive man made storyline, but peppered in with the Radient quest system.
Sure, a lot of those quests were pretty simple fetch quests, but think about what that could look like now with an LLM in the mix.
It doesn't have to be the whole game.
There's a theory that Tim is AI generated. After all we've never seen 10 fingers on him. TimAI, I'm onto you!
I've heard that same theory, it's just that...error...null, I'm afraid I can't answer that question as it goes against the Terms of Service to TimAI. Code: 9983211
see, totally normal human being here!
If this doesn’t prove the simulation theory, I don’t know what would. Think about the game experience we can create in realtime 50 years into the future, heck maybe just 20 years.
as we begin to fit all the AI tech together to work synergistically, we get closer to simulating an entire universe.
Weird question, would it be possible for you to implement something like AI generated dubbing for your videos? I have some spanish friends who would love to see your content, but they don't speak or understand english at all, so I always end up explaining your videos to them xD
So funny enough, I believe RUclips is set to implement that soon. I don't think we'll get lipsync, but my voice can/will apparently be dubbed into various languages.
I got a notification about it a little while ago, so I'm sure it'll be around in the next few months. Knowing RUclips, it'll roll out to the big accounts first, before trickling down to me.
@@TheoreticallyMedia😂
Я уже три года смотрю любой контент ютуба на русском языке независимо от языка, новости на китайском или английском, испанский блогер или французкий, немецкий или корейский не имеет значения, все синхронно дублируется на русский язык в отличном качестве и гендерной расстановкой)
@@TheoreticallyMediaI heard about it, but I didn't know it was going to happen so soon! those are some great news, thank you so much for the info and reply, love your content
Пусть используют субтитры на испанском🙄
Или они читать не могут?)
@@homuchoghoma6789 Зачем сразу в негатив, дорогой товарищ? У меня вот нет Испанских субтитров (у Вас есть?), только автоматические на Английском. Что ребятам не поможет. Ждём фичу дубляжа!
It seems like a lucid dream rather than a game.
I just don’t see yet how these generative games will fare. As much as it sounds good on paper, people play games to relax and escape and I don’t think (many) people would want to “define” their world as they go along. They want to just play and escape and not spend the time generating it. Auto prompting though sounds appealing.
This could be Google´s roadmap into the gaming industry.
Just WOW!!!!..
totally-- but, at the same time-- it's Google. So, they'll wiff it. But, I'm sure the team will find other pastures at some point, and we'll see this tech being developed out for a product that will actually ship!
@@TheoreticallyMedia I think that like all AI, Google will fail and it will be another company that will succeed.
If we want to generate an image, a text, a sound, a video, we do not think at all of Google.
Damn, Starfield is catchin' strays just :10 seconds into the video!
looks great for conceptualizing games id hate to have to play it for more than a few minutes at a time though.
This is be a great tool for devs I'm not sure when or how but it will be great
omg will you _please_ show the team at MJ what's happening here, Tim. Perhaps it'll encourage them to release v7 and hope it can understand what we're asking for, generate hands and blow us out of the water. 3D AI 'games' are peeking their heads, and we're still waiting for David Holz to _figure out_ what they're doing next with MJ. Thanks as always, Tim. I'm glad I'm not entering the job market right now. All I encourage the young ones out there is to reskill, upskill and keep up. Massive paradigm shift in the market.
I'm just glad there's any progress tbh. Maybe if we're lucky we'll get full ai gaming by 2026. That would be something.
now if it could only generate the world and keep the character/moving objects on the screen consistent than we will actually start talking about groundbreaking stuff.
Prompt: "Create a game indistinguishable from reality that people are hooked into so we can use their body heat to power our machine city."
it surely won't be long before you can just tell it to generate a 100 hour open world game that is better than any game currently made today. people will stop buying video games and just play these AI made games, after all, you will only be limited by your imagination as to what it can create. I think it might destroy the video game industry and put a lot of game devs out of work
Nah, that's assuming linear progress without taking into consideration any of the infrastructure or financial costs. As far as actual games, people play multiplayer where objects have to have permanence and be exact for tens of thousands of players all over the world.
Can't wait til we all start developing games on this platform, for them to sunset it when it peaks.
Just need an model that can do style-transfer onto ascii-text art combined with this.
It will take a very very long time until we get AI games because it will need to put out sounds at the same time.
We need a huge leap in hardware before that will be manageable with our local computers.
check out yesterday's video! We have generated sound now too!
@@TheoreticallyMedia Did you even read what he said?
Shh, don't let the AIbros know about the reality of tech. They'd spontaniosly combust if they know how tech works.
@@ArksideGames What is the reality of technology?
That makes so much sense not being able to find my keys is a glitch in The matrix!!!
See, WE can't crazy! It's the Matrix that's out to get us!! Grab some tinfoil, we need to make hats! haha
@@TheoreticallyMedia can you recommend an AI platform for that?
What a time to be alive
This AI model is like a junkie. It can’t stay for over 1 minute without being high
Holy sh8t! I'm trying to keep up with all the updates and tools and then there are new updates and tools. Should I just wait it out? I'm itching to CREATE though.
I suspect that by the time that GTA 6 gets released (which is late 2025) we'll already be able to generate almost photorealistic AI games.
5000 Nvidia presentation will feature something like this for sure
Jensen has really been pushing this idea, so I expect to see the same-- but with Nvidia muscle behind it. Cyberpunk: 2077 Endless AI Edition incoming!!
Now do a quick 180 turn and lets see how consistent "the game is", no games haven't changed forever, yet
This is essentially the software that ran a holodeck, but in real life
When Google is involved, I'll believe it when it's usable outside of their ecosystem.
Still, it's an interesting glimpse into what the future might bring.
Yeah, I tend to think the Google-Way is for research teams to develop in there, get frustrated at the lack of movement, eventually quit, and then go start their own thing.
Did you see the Notebook LLM team just ditched? Man, I love Notebook-- but, hey-- if that team is moving to bigger and better, godspeed to them!
@@TheoreticallyMedia I meant to try Notebook LLM out but forgot all about it. You are right though, there does seem to be a pattern of a team develop something, then they quit and start on their own. It seems to benefit us though, so I'm okay with that.
Microsoft, Sony Nintendo, EA, Ubisoft are working on their own AI models.
Believe me, Google is far from the only one..
@@Siranoxz Absolutely, I only mentioned Google because they are the main focus in this video.
Crazy progress in less than a year, it went from a gimmick to creating AAA looking stuff.
1. Does this stuff ever come out. 2. When it does can I just make my own thing , I predict "my" game will be on some host platform and the host platform and you will only maintain a little bit of control of your own project. Instead of creating the project and removing it to my own space / control.
Good questions. As to your first, in the last video I was looking at the whole "Add AI Sound to Any Video" thing, and I went back to go look when the white paper on that was first released-- and it was roughly 10 months ago.
I mean, it's different in terms of what the tech is doing, but I think that's a solid benchmark of how long it takes for something to go from the research phase out to market. (some things obviously take longer)--
In terms of platforms? Well, that'll be all over the place. I'm sure a few will have exactly what you describe-- but, maybe we'll see some Open Source versions as well?
@@TheoreticallyMedia If they actually released it to the public even as a early alpha product, they would get plenty of feedback from people using it in creative ways even at current state, which would lead to improvements in final product. Google has this issue where the product never materializes and if it does then it is already behind competition, which is probably due to that mentality of not releasing it in first place.
@@TheoreticallyMedia I guess ten months is fast but I feel like I'm voyeuristically watching these AI tools rather than being able to undertake real hands-on. Perhaps frustrated as I have a board game irl ready to go digital and as I have no real dev skills I'm hoping this tech can allow me to make the game. I doubt it will in under 5 years - eventually I hope.
Casting shade on Starfield in passing = +1 street cred.
The question is can it export
unlikely not right now. I mean, beyond an mp4 of game play.
This will put a whole 'nother meaning to the term procedural gaming if it becomes a viable solution for game development
I just want these things to be able to work with 360 photos
You're still not punk, and I'm still telling everyone.
You mentioned Branson, MO, and I almost sh*t myself. My first paid music gig was at a show in Branson. It's always surprising to hear that people know Branson exists. lol
I'm really looking forward to being able to use AI to generate new VSTs or plugins. When will the crossover happen?!
Save your breath, I never was one!
Haha, yeah- Branson is SUCH a weird place. And it is just funny enough to set a GTA game in! I don't know, that's probably why it popped in my head!
And yeah, the whole AI VST thing! I was JUST talking about that again. I'm really looking forward to the AI Band-In-A-Box becoming a reality. Like, real time reaction/jamming from live input audio.
Talking to one company that's early days into it-- they're predicting we'll see it within a year! Exciting!
@TheoreticallyMedia for real man! I mean the show Ozark already set the theme, and the kitschy-ness of the wannabes and the washed-up old-timers would make for some Rockstar gold.
Dude I can't explain how excited I am for more info on that company. Any clue when they'll be doing press releases or allowing you to talk more about it?! generative AI music is cool, but AI as a toolkit for musicians is what's got me the most excited for the future as an artist.