I think graphics in video games has reached a pretty good threshold for realism. What they need to focus on now is realistic AI, animations & physics like breakability
the hands down most disappointing lack of development in video games imo is physics. lighting, texture, and polycount has advances 100 fold in the past 15 years, and games LOOK like Hollywood movies almost. Pretty impressive, until you start playing them. When you shoot a rocket launcher at a brick wall and the most damage it does to the environment is a black circle texture where the explosion hit. I completely understand that physics is very demanding, but come on. It's time for someone to start pushing at least the attempt to create something equating an actual world simulation and not just a bunch of empty polygon shells with preanimated triggers for interaction.
To do a fully physics based environment would require voxels instead of meshes and not only are voxels way more demanding than meshes even without them being touched or manipulated, there's (currently, at least) no voxel acceleration hardware in GPU's to improve voxel density. For now, teardown is pretty much the resolution limit for a fully physics based voxel game.
@@thelelanatorlol3978 I don't think a voxel based system would necessarily be required to start making some pretty big advancements in physics in games. you could use advanced physics simulations using a volumetric system, and incorperate pre-bakes stress and fracture points, based on the physical material the thing is supposed to be made from. from there is just a matter of categorizing every model with a mass variable, and you can apply the interaction of the kenetic energy of the material, to those "pre baked" fracture points in said material. lets take a glass window for example. you don't need to build that volumetrically to simulate it realistically. You simply have a force value that indicated weather the glass breaks, and a kenetic energy value of any object hitting the glass, including material deformation etc. if it passes the fracture threashold, it breaks off along pre bakes fracture lines and the glass shatters and falls to the ground etc. you could take that kind of approach with a LOT of assets.
@@homewardboundphotos This is all basic stuff in the physx engine. Pretty much every destructible you see in games using the physx engine uses this feature. The reason you dont see it applied everywhere is that its just too computationally demanding. Even if there was spare computation power, i at lesst would rather see more accurate physocs calculations than just more physics calculations. Physics jank is fun but only when messing around.
While many AAA games have lackluster storytelling there's a lot of interesting ideas in the indie side of things. I liked how Scorn managed to tell a story about a strange planet without any dialogue whatsoever. The refreshed Stanley Parable is another example of how to take storytelling and put a twist on it. My perhaps biggest disappointment this year is Far Cry 6 that had a much downgraded story compared to its predecessors. GPUs are starting to reach really solid levels of performance where games can throw ray tracing, HDR and much more at 4k 100Hz. The power consumtion per frame is dropping but the MSRP per frame is not moving in the right direction. Nvidia has 67% profit margin so they could easily cut prices by half and still be an insanely profitable company. Hopefully Intel get beefier chips with Battlemage so they can put pressure on the pricing. Their ray tracing is really impressive but they need better shaders to match.
AI accelerated asset creation will bring the cost of making games down by a large margin, small indie studios can compete with large, well established ones in a meaningful way, Developers are much more likely to take risks and innovate thus leveling the playing field.
Your right It SHOULD bring it down by a large margin but I'd guess they will claim the use of high tech equipment has raised production costs. Making GPU's never has cost less but we are all paying more than ever.
I wish.... have you seen the stories Ai tell right now? At best they are slightly amusing for their absurdity..... but then again give it 2 months and it will impress the heck out of me with a new model. hard to grasp the 1000x improvement cycle Ai is undergoing
@@diwajerebation4077 chatgpt is scary shit lol. Ask it about stuff that you know about and most of his answer are complete BS. No doubt future AIs will get better though.
I'm still waiting for proper cloth simulation, fluids and deformation in games that were introduced by Ageia's Physx in 2005. Still nothing. Only Killing Floor 2 managed to utilize PhysX a little to simulates blood/acid and intestines
Add particle dynamics in general, especially airborne. Atmosphere is one of the core aspects of high fidelity graphics in gaming that still lacks far behind everything else.
Just a warning about your sponsor, Urcdkeys is a scam company. They don't sell life-time licenses, they sell temporary keys that can be changed to anyone else at any point as they're business keys. I also don't know if your payment information is safe with them to begin with.
*NVidia in another two years...* "Buy our new RTX6900, it gets PHOTOREALISM at 120FPS!" *NVidia a year after that...* "Buy our new RTX7900 it... uh oh... ahem, Your RTX6900 sucks. 8K at 240FPS is the future!"
mostly has to do with games being built on end-of-life game engines up until recently. UE5 and the engines that follow in the coming years will change that
"Maybe AI can fix it." Small signs that we will slowly but surely become dependent on AI and will not be able to function without it in the future. Brings back memories of the animation WALL-E, where people are purposeless, their roles replaced by technology. Good luck to humanity.
@@aexetanius Why is that not a good thing? With the introduction of tech and the industrial revolution we have practically lost the meaning of being an human, having AI replacing jobs might mean that we are finally able to practice being an human instead of being forced to work for the higher.
Ironically, our brains are more forgiving when it comes to texture resolutions and shading, compared to animations and physics, the lack of which would break immersion like there's no tomorrow.
8:26 I noticed that for some videos, the most viewed part is often right after a sponsored message since people try to skip that part and land too far and need to go back and replay.
Animation and motion capture remain a problem. Most AAA games are over acted because facial nuances need to be exaggerated and movement is still limited to situational pre recorded animations. The last step is for everything to look natural and Nvidia has been working hard on AI animation tools that they use on Jensen for presentations now. Hopefully in some years it will pay dividends over present methods
Looking forward to seeing video and 3D worlds blend, so that VR movies would have depth. The mentioned Diehard, allowing a VR user to move around the scene. Imagine being able to go to any scene in any movie, removing the actors or elements in a scene, and being able to be in the environment by itself, changing the actor, their clothes, etc.
Love how meshletsseem to strategically choose which pieces to kick up to a higher LOD beyond just the Z-depth. Same goes for the massive improvements in photogrammetry to scan objects and extrapolate the corresponding maps, to the point where they'll probably soon even generate node-based shader setups for advanced surface types like color-changing car paint.
I have been a professional 3D artist and i have used world creator, unreal engine and everything, from making game env to photorealistic scenes even working as environment artist for games and TV shows, believe me guys, its not as easy and simple as it is showed, a lot of hard work and time is spend to make it realistic and let alone believable, also the video he is using in the very beginning is from artist named aron westwood(search for " I AM world creator software" on youtube), and it took months and probably years of research and learning to make something like this, i dont want to spread negativity, its better to know the entire mountain while desperation might eat your time.
These technologies are definitely going to be cool, but a good game lives, or underperforms, with compelling stories and characters. Let's hope they have more time for good writing in the future. lol
AI objects are very interesting. Imagine an open world game where you can simply outsource interior layout, furnishing, and decoration to an AI such that every single building interior in the city is unique and is rendered by an AI, freeing up time for programmers.
After watching your video about plague tale requiem, i played it and finished it. I'm just blown away by the graphics. It just looks much better than anything out there to me. Are there any games in your opinion that can match it graphically?
the usual ones. Red dead redemption 2 , last of us 2 , god of war ragnarok, Horizon Forbidden west. They might not match Requiem but are the closest ones available out there
I see you beat me to talking about Neural rendering, although you haven't really taken it to the limit of one of the articles that was on your website.
Has anyone thought of trying high resolution etc with "Pulsing", changing the amplitude, angle, polarisation, tone etc in a subliminal way, just as with Showscan where the number of frames per second, was such that it induced intense emotions in the viewers, also the ability to change reflections, refractions, etc could be used for 3D effects.
i never seen Nanite being called "meshlets", specially as meshlets is kinda used to refer to something rather different. with Nanite does is commonly referenced as as Virtualized Geometry, you have the geometry in this massive hierarchical data structure ,rather actual meshes in memory, and it is organized in such way as you can you get exactly only the geometry you want and with how much detail you want for each part of the scene. every single frame. that why it is called virtualized, you still rendering triangles in the end, but what is actually being rendered on each frame is generated every time it is used. not something that is already in memory the secret sauce of nanite is not even that the meshes are broken in tiny parts. quake 1 did that, ps1 games did that. but in how the heck you organize your geometry in such way that you can pick a lot of it or little of it and still end with a list of triangles that can be rendered on screen that still looks like the thing you trying to render. and all that fast enough to be done every single frame. heck fact they couldn't even get current gpus rendering pipeline to play nice with the geometry generated by nanite. and it is(was? things might have changed) actually a software renderer implemented in gpu believe or not. which would be bad, if was not for the fact that the advantages outshine the disadvantages 1000 to 1 (if you try the exact same scene with traditional rendering and with nanite, nanite is actually faster and uses less vram and won't have LOD poping, is almost like every GPU from the last ~7 years in the planet got a ~30% boost out of no where) while meshlets are usually the things mesh shaders work with, its a gpu thing. it is related to some of the operations that nanite does. but mesh shaders is more about: "i have a mesh and i want to do things with it efficiently directly in the gpu", i never ever seen it being used tbh
He's not totally right, but he's not totally wrong. "Meshlets" refers to the practice of breaking meshes up into hierarchical LoDs of fixed-size "patches" of mesh, and moving the LoD, culling, and dispatch of those patches all to the GPU instead of the CPU. While Nanite doesn't use the specific hardware features inteded to facilitate meshlets, it's still fundamentally using meshlet rendering. It's not true that the triangles are generated each frame, but it may be true that meshlets (patches of triangles) are streamed on-demand from the disk more or less straight to the GPU. What's generated each frame is the list of things to draw, what we call dispatch. What made "dispatch on GPU" possible was compute shaders (and meshlet shaders) and we just hadn't realized it until now because it's a very unusual and inspired bit of lateral thinking about rendering. Quake did not break meshes into tiny parts. It used "brushes" to represent 3d objects, which is the intersections of planes to describe convex solids. These are easy to do certain math on like CSG and splitting into BSP-trees. "Software renderer implemented in gpu" - partial marks for this. Nanite has a software *triangle rasterizer*. The hardware rasterizer quickly loses performance when triangle sizes approach one triangle per pixel. The assumption made in hardware rasterizers (that triangles will be large on average) is very good for traditional render pipelines, and poor for Nanite. You don't really get a VRAM boost out of Nanite unless you really embrace the technology and get rid of your normal maps and displacement maps. The savings is texture memory and not everyone is using it wisely like that.
I worked for Metaplex this year, it’s looking really rough for them in the near future. All the OGs are being laid off and my project was affected heavily. I still believe in the technology 100%, it’s just that they trusted the wrong people to work with them. SOL is still the best tech around but I won’t lie, it’s been eye opening
Any kind of ship or building would get an AI assets set would get many sims a variety they lack right now (or was a hard work adding it like cities skylines). Imagine if dev would make core of the building and AI would move windows and balconies around, throw cracks and damage, add rooftop chimneys and stuff like that. you could easily get a set of many non-repeating homes to use in cities, with variety, adding paints as well, we get a real lib. using this to apply damage to things could be interesting as well. Cracked scopes on guns and different cracks on every gun you pick? I feel like simulating damage on your equipment can get much more interesting with this.
It's definitely the case that realistic graphics wins over any other kind, at least in VR. Stylized graphics in VR are both harder to look at and generally look worse than realistic graphics.
Honestly, the fact that industry is acknowledging that users attention span is becoming shorter, and therefore making a tool that encourages such lacking, is not a good news...
I've never heard of AI objects as a concept (I'm a professional 3D artist and game developer) I can't find anything online, can you link some resources?
Search for "neural rendering" - essentially in a game, a neural rendered object is something that's inserted into your scene at the end. IE your object might be a "panda" and if you put it in water it becomes a "swimming panda" - so it's not something you process into a mesh and render classically - you lose control of how it renders after you state where it appears - hence its not described using vertices/voxels etc etc
The world creator tech really interests me. I play around with a universe simulator that works on the same procedural generation but is still a ways from how good those examples are but over the years is continuously improving towards photo realism.
3:55 the narrative that new technologies will allow games with "better ai" has been around forever, and while there *have* been some milestones the big leap never happened; because especially AAA companies are still mostly focusing on eye candy.
Have you ever wanted to watch a movie, but saw that there are too many things to watch, and watch nothing in the end? 11:27 will make this stronger and it is very scary. I can't imagine the amount of analysis paralysis we will have when deciding not only what to watch, but which edition to watch.
forget GPU rendering on servers and streaming it back to the client. that would be like going back to the beginning of computers where you had client stations that alway connected to servers. no one wants that, not even for rendering. also there is the issue with speed, stability and stuff. and i know because as a software developer i did an application where a client send midi to a server to get audio data as a response. this must be very fast to have almost no latency. this works pretty nice for a few seconds, sometimes a minute until there is the first package thats lost over the internet. requesting that package again breaks latency. and using UDP instead of TCP will solve latency but that means a lost package will stay lost. you would have dropouts. but then again you might send data twice if the speed is very high. in the end its a very hard thing to do. don't expect this to be solved within 10 years. just my opinion since i worked in that area
90s gaming was filled with such a vast range of creativity even with their limitations. I love the amount of risk they took on odd concepts in This era- as well as the PS2/360 era. It should show us that GRAPHICS/Reflections are not the most important part. No. If you can create satisfying movement, gameplay mechanics, physics effects, responsive parts of a environment, a environment that makes you want to explore it, engage it, etc. That stuff is so important. Not just "who can make the largest game world, or a game with the best Graphics." I think Art style, design, vibe, aesthetic is much more important. You don't need top notch graphics to greatly appreciate and enjoy playing that game. That's why we are seeing such a resurgence of players who have gone back to playing older games because they realize the aspects of gaming that they care about and are unhappy with in most modern games. So hopefully gaming will enter a new era. That mixes in aspects of the older era's. With our modern capabilities that will hopefully help make it easier to create a satisfying game. As well as a vast range of game's Instead of everything trying to be Triple AAA or free to play online battle Royale games..
currently youtube shorts - and other shorts - is just a social media crap - it has nothing to do with most watched part of video or the most condensed information.
You totally missed ai rendering...its alreay been shown in the famous GTA5 ai rendering demonstration where they even go so far as present it as having a "interactive framerate". This will basically blow everything out of the water you presented here, still thanks for the video.
And I still... can't..... save game. So I have to replay levels over and over to make any progress, and don't experiment at all. Which I can take for about an hour. Everything on the PC had "save game" until consoles took over. Will these new developments help with that? Been waiting since Crysis.
Well... that's a choice made by developers. Saving is as easy as saving all the states of all the objects - it didn't become any harder that 20 years ago.
@@igorthelight I know. But it's a main reason I no longer buy action games (which I otherwise love). And all the graphic goodness in the world won't fix it. I still read about the tech but that's all.
@@anslicht4487 Most modern games use frequent autosaves. Some let you save in any time (Cyberpunk 2077). Replaying the whole level is quite a rare thing, no?
@@igorthelight By whole level I mean since the last autosave, of course. Which might be ten minutes of intense action. I absolutely *don't* want to keep repeating that at one time. And I can't even break from the game when I need to! AND a quick-save as in Crysis let me experiment endlessly with combat approaches and techniques, exploration etc. So it's still fun 15 years later. *Deleting that feature is the single most stupid "development" in games **_ever_* . Don't get me started 🙂
Wow, that is some food for thought. As as relatively 3D artist specialized in asset creation, who has just recently starting working, what would advice be to somewhat prepare for the coming shift in graphics? I am confident in the quality of my assets, yet that may not matter that much soon. I have no programming knowledge aside from web-stuff, but I have a solid technical understand of graphics for an artist. One idea would be to lean heavier into ‘technical art’ and maybe procedural generation generation, thinking that it will converge with AI generation, right?
I'd guess supervision of procedural ai generation will be in high demand in near future. And single-click solutions still wouldn't be enough. I can see a lot of fine tuning of parameters, combining several ai tools in unexpected ways, teaching it particular art styles and leading it to desired objectives - things like that. And for now it seems basic scripting in python is enough, and there sure will be plenty of no-code solutions in future. Hope some of then will help me with my web app developement. But the hardest problem in develpment is, that there always grows some special kind of mess which actually nobody understands. Something with friction between wetware and hardware abstractions :-)
My recommendation would be to understand the principles of operation and construction of the real world objects that you model. As an example, I've worked as a contractor doing building renovation and maintenance. I've also done level design for FPS games. I find it a little strange sometimes when I see maps built by someone that doesn't understand of real world architecture or building design. There's numerous ways of how to convincingly place assets, to make it look like they really belong there. As an example, lets say you want to place a series of lights down a hallway. As you progress, the bundle of wires that feed each lamp should get smaller, until there's just a single wire at the last lamp. I've seen too maps where people just stick a mess of pipes and conduits in the hopes that it looks "sci-fi". For me, it breaks the realism. I've actually installed these things in real life. When I walk into a building I can visually survey the place, and know what the infrastructure is supposed to do. The more you understand the fundamentals of what you're doing, the more you'll have an edge on others in the field.
@@lihtan Wow, thank you, that is some really solid advice. This type of knowledge is something I knew was helpful, but I did not think about it enough to see it as something that I could 'make my own'. I have no construction knowledge, so I do not often look for these things in games, but I know many 3D-artists don't give these things too much thought, and it is almost never mention in all these breakdown articles, videos and blog posts, although it is very important for realism and modeling efficiency. The hard part is of course acquiring that knowledge as an outsider, but if it was easy, everyone would do it, I guess.^^
Fundamentals. AI won't for a very long time have a good understanding of things like edge flow, T-pose versus A-pose (protip: use A-pose to preserve range of motion when you rig it), compact/efficient UV layout, and how many fingers and toes a human has. Don't auto-retopo if you can avoid it, that does things AIs are good at and often doesn't make things pleasant for an animator down the line. As long as you keep thinking about the value chain connecting you and the other actual humans in the pipeline you can 100% beat an AI in quality.
Instead of being crushed by this juggernaut of disruption for commercial artists, I am interested in riding the change by adapting to requirements. Will commercial artists of the future be the ones who are expert at stating the best prompts? What is your vision of the future for graphic arts?
For VR and AR to go mainstream, we need to find a way to make our bodies trackable for computers without having to wear cumbersome motion tracking suits.
Unless your monitor is at least 40" big and you sit within 50 cm from it, 8K monitor still doesn't make any sense because the resolution of the Mark 1 Eyeballs in your head is not good enough to see the difference between 4K and 8K image, never mind the computing requirements to actually render that 8K image. Spend your money in more accurate color reproduction, stable gamma ramp over all the image (even IPS panels have different/incorrect gamma ramp for the edges of the screen when you sit close to it) and dynamic range, because those things you can actually see.
an artist who uploads really well made videos to youtube called Jazza, brother to the shadiversity guy who talks about armor/weapons in real life, is making a video game at the minute, and he was soo impressed by A.I generated art for things like jewelry, that he said on multiple occasions in different videos that he's using a.i generated art in the game. I'm assuming for U.I elements, also things like inventory. its actually worth a watch, he's really impressed and hyped by what things like midjourney and dall-e produce, and he mentioned how everybody seems to hate A.I these days. but he even reiterated in the video talking about the hate that he's going to continue using a.i generated artwork for the game.
Nanite is not a stable technology yet, there is still a lot of work that needs to be done and also anyone who've tried writing shaders for UE5 will know that, custom pixel/fragment shaders are extremely heavy(slow to run) in current version of UE 5..
No matter how realistic games look or how realistic ai voices sound, humans are always somehow able to detect them...I think we vastly underestimate ourselves
You could dynamically generate high quality textures using AI, so a game could just have a snippet of code that precisely describes the desired texture for a deterministic AI image generator to create the textures at startup, or even at runtime, and effectively compress hundreds of GB of textures into text files that weight just a few MB, though of course AI objects would already not need textures saved on disk. Heck, you could even code an engine to dynamically modify or completely recreate a texture on demand, so for example if a gun got damaged you could directly modify the texture, or the mesh, to add damage based on what happened and store it permanently by using a single AI modell
I assume by 2050, good chunk of generated media is created by handful of people using AI toolsets, today we can generate very decent AI voices without voice actors involved, we can create models, images with AI even basic videos can be made by AI. Whats scary about this is the implications, if every man is a game studio or movie studio, you oversaturate the market so much, you gonna destroy entire industries, actors will find themselves increasingly out of jobs so will movie and video studios. Why hire 500 people to make a 50 million dollar movie when you can hire 5 people who can make the same movie for 5 million using AI toolsets and all within 6 months. In a way this also liberates entertainment but at the same time its a curse.
I play mostly VR games these days and i can't wait to see the graphics we will have soon! I just hope my new 3090 will be able to run the best for a while.
@@Freshbott2 my top five PCVR games are half life alyx, lone echo 1 and 2, wanderer, beatsaber and blade and sorcery. There's so many good games for all tastes, you just have to give them a chance. A lot of people love sim racing / flying in VR. Maybe look at games you like and see what VR games in those styles exist?
Damn that's one nice gaming set up. Here we have massive taxes, I'm trying to build a 12100f 3060ti but it seems I'm looking at something around 2400usd taking into account monitor and maybe tge desk too. Fuck taxes.
What I worry about is that AI will lead to lazy artistry. Imagine that there is an open world game where there are various settlements scattered throughout a variety of biomes. Would the buildings and architecture of a small town on a tropical beach be the same as those found in say the biggest city where the King lives in his castle? Probably not. The beach buildings would be small, likely constructed out of wood and straw, materials found on or near the beach. But the castle town would be brick, would have a medieval feel to them. AI would be able to construct both, but only a good designer and an artist could place both in their appropriate spaces and create the game's lore that supports both settlements. If a good human designer wasn't there the AI could just place whatever building style wherever and shatter the immersion of the player. I'm worried that money driven publishers will see AI as a quick and easy way to cut dev costs so they can maximize profits, at the expense of the quality of the end product. "No need to higher those expensive game designers, we have the latest in AI tech that can do the same thing," say the CEOs who aren't gamers and never play the games their company produces.
Most people don't care about lore that much though, if the world design isn't too jarring most people don't care. And also like you said: money talks, bullshit walks.
We've been pushing for photo-realism for decades, but what good is photo-realism if we can't dynamically alter scenes at will? We should be able to dynamically tear down buildings, having them collapse and leave a pile of realistic and movable rubble on the ground. We should be able to fire a mortar in some random direction in a forest and have that open a crater surrounded with broken trees and debris. We should be able to move mountains. That, to me, is the future. Voxels (the real thing, not Minecraft) are the future. Not necessarily photo-realism.
How close the details it will be?? Can see detail at um, nm, and beyond? So, when look far so you can see the blurry, reducing light and so on, well, the more high detail at low um, high nm, low nm, ya will see the different when look far away, so what is the highest details before you can't see low detail at far away since looking the same unless huge since it is about scale, so 4K and 8K look same if at certain size, so need to be huge. So, unless you produce a giant, the bigger the giant will be equal to the detail at small level since the scale is same. The ratio of details to certain size so not see reducing something so look all real as real life is photorealism, higher than hyperrealism, so for hyperrealism you can see it is not photorealism.
They are mEshlets with an E. While it mashes meshes together a little bit of a mesh is a mesh-let... I hear this term a lot so the beginning of the video was really painful to listen too.
Still, we don't have a decent image to 3D models (maybe something like Monster Mash...on steroids, and for free). Can you imagine a full (good) 3D model generated with a single image? A free AI voice generator without robotic sound would be awesome as well. And some free mocap generated by a pre-recorded video with a single camera, with foot locking and hands detection (PlaskAI on steroids, for free), maybe a face tracking from pre-recorded video data from a single camera. For free... And... I ask for too much, I know haha
![Fully Enterable Buildings In Gaming [Future Of Games 1/6]](http://i.ytimg.com/vi/Xdm98g-xUbc/mqdefault.jpg)
![Fully Enterable Buildings In Gaming [Future Of Games 1/6]](/img/tr.png)
I think graphics in video games has reached a pretty good threshold for realism. What they need to focus on now is realistic AI, animations & physics like breakability
And affordability...
What we need is neural driven AI actors rather than some retarded AI with predefined dialouge etc that some stupid programmer put together.
yeah its so wierd you see games like left 4 dead then compare it to back 4 blood and all the little details were better
Narrator: They won’t.
no. they will make another remakes and remasters of recently remastered games)
the hands down most disappointing lack of development in video games imo is physics. lighting, texture, and polycount has advances 100 fold in the past 15 years, and games LOOK like Hollywood movies almost. Pretty impressive, until you start playing them. When you shoot a rocket launcher at a brick wall and the most damage it does to the environment is a black circle texture where the explosion hit. I completely understand that physics is very demanding, but come on. It's time for someone to start pushing at least the attempt to create something equating an actual world simulation and not just a bunch of empty polygon shells with preanimated triggers for interaction.
100% agree. It's one of the reasons i enjoyed Control so much. Best phyics I've encountered so far.
To do a fully physics based environment would require voxels instead of meshes and not only are voxels way more demanding than meshes even without them being touched or manipulated, there's (currently, at least) no voxel acceleration hardware in GPU's to improve voxel density. For now, teardown is pretty much the resolution limit for a fully physics based voxel game.
id take realistic physics over raytracing and 8k polycount anyday of the week
@@thelelanatorlol3978 I don't think a voxel based system would necessarily be required to start making some pretty big advancements in physics in games. you could use advanced physics simulations using a volumetric system, and incorperate pre-bakes stress and fracture points, based on the physical material the thing is supposed to be made from. from there is just a matter of categorizing every model with a mass variable, and you can apply the interaction of the kenetic energy of the material, to those "pre baked" fracture points in said material. lets take a glass window for example. you don't need to build that volumetrically to simulate it realistically. You simply have a force value that indicated weather the glass breaks, and a kenetic energy value of any object hitting the glass, including material deformation etc. if it passes the fracture threashold, it breaks off along pre bakes fracture lines and the glass shatters and falls to the ground etc. you could take that kind of approach with a LOT of assets.
@@homewardboundphotos This is all basic stuff in the physx engine. Pretty much every destructible you see in games using the physx engine uses this feature. The reason you dont see it applied everywhere is that its just too computationally demanding. Even if there was spare computation power, i at lesst would rather see more accurate physocs calculations than just more physics calculations. Physics jank is fun but only when messing around.
hearing you acknowledge that storytelling in games has been lacking was really validating
Ur not valid
@@clearlisted fine by me
While many AAA games have lackluster storytelling there's a lot of interesting ideas in the indie side of things. I liked how Scorn managed to tell a story about a strange planet without any dialogue whatsoever. The refreshed Stanley Parable is another example of how to take storytelling and put a twist on it. My perhaps biggest disappointment this year is Far Cry 6 that had a much downgraded story compared to its predecessors.
GPUs are starting to reach really solid levels of performance where games can throw ray tracing, HDR and much more at 4k 100Hz. The power consumtion per frame is dropping but the MSRP per frame is not moving in the right direction. Nvidia has 67% profit margin so they could easily cut prices by half and still be an insanely profitable company. Hopefully Intel get beefier chips with Battlemage so they can put pressure on the pricing. Their ray tracing is really impressive but they need better shaders to match.
theres plenty of great storytelling out there, just not in the AAA market
@@LeonardTavast i just wish someone made hardware that is as fast as 4 years ago but consuming 10x less power, but i guess that doesnt make money
that 128K texture from OTOY presentation is probably more to do with virtual texturing and mip streaming, than meshlets
AI accelerated asset creation will bring the cost of making games down by a large margin, small indie studios can compete with large, well established ones in a meaningful way, Developers are much more likely to take risks and innovate thus leveling the playing field.
Your right It SHOULD bring it down by a large margin but I'd guess they will claim the use of high tech equipment has raised production costs.
Making GPU's never has cost less but we are all paying more than ever.
I really hope this is how it turns out. Some greedy fuckers out there in the gaming world
Then there will be no one to compete at all because any random person will be able to use the AI to make them an AAA game.
It will just raise the bar on what is AAA.
@@flamart9703 i.e , "make me a 4 player game with dragons, robots and morality tests."
Now game devs need to use AI to write a good story for them.
I wish.... have you seen the stories Ai tell right now? At best they are slightly amusing for their absurdity..... but then again give it 2 months and it will impress the heck out of me with a new model. hard to grasp the 1000x improvement cycle Ai is undergoing
bro chatgpt is getting scary good
@@diwajerebation4077 chatgpt is scary shit lol. Ask it about stuff that you know about and most of his answer are complete BS. No doubt future AIs will get better though.
No, AI-generated dialogue
Games don't really need stories.
I'm still waiting for proper cloth simulation, fluids and deformation in games that were introduced by Ageia's Physx in 2005. Still nothing. Only Killing Floor 2 managed to utilize PhysX a little to simulates blood/acid and intestines
Add particle dynamics in general, especially airborne. Atmosphere is one of the core aspects of high fidelity graphics in gaming that still lacks far behind everything else.
I'm still waiting for GeoMod from 2001 to make a comeback into games.
Just a warning about your sponsor, Urcdkeys is a scam company. They don't sell life-time licenses, they sell temporary keys that can be changed to anyone else at any point as they're business keys. I also don't know if your payment information is safe with them to begin with.
*NVidia in another two years...*
"Buy our new RTX6900, it gets PHOTOREALISM at 120FPS!"
*NVidia a year after that...*
"Buy our new RTX7900 it... uh oh... ahem, Your RTX6900 sucks. 8K at 240FPS is the future!"
Can't wait for the GTX 50 series where I have to take out an entire 10k loan just to be able to glimpse at the future of graphics
I want to revisit this video after 10 years.
I hope there will be a generation graphics uplift because this last decade I barely see any uplift in games graphics to be honest
mostly has to do with games being built on end-of-life game engines up until recently. UE5 and the engines that follow in the coming years will change that
Storytelling in video games has been in decline for many years sadly. Maybe AI can fix it too
"Maybe AI can fix it."
Small signs that we will slowly but surely become dependent on AI and will not be able to function without it in the future. Brings back memories of the animation WALL-E, where people are purposeless, their roles replaced by technology. Good luck to humanity.
@@aexetanius If we are bad at what we were meant to do (use the meat in our head) then I think it's normal people are leaning toward AI.
@@aexetanius Why is that not a good thing? With the introduction of tech and the industrial revolution we have practically lost the meaning of being an human, having AI replacing jobs might mean that we are finally able to practice being an human instead of being forced to work for the higher.
@@flintfrommother3gaming it is not that we are bad, but creating a game that is not scripted is just very difficult to get natural.
Ironically, our brains are more forgiving when it comes to texture resolutions and shading, compared to animations and physics, the lack of which would break immersion like there's no tomorrow.
From what I have observed the most replayed sections of videos that follow any pattern is right at the end of the ads, this is hilarious to me.
I don't want expensive and realistic graphics, I want good games.
Same i want games with good graphics but not realistic graphics
8:26 I noticed that for some videos, the most viewed part is often right after a sponsored message since people try to skip that part and land too far and need to go back and replay.
Animation and motion capture remain a problem. Most AAA games are over acted because facial nuances need to be exaggerated and movement is still limited to situational pre recorded animations. The last step is for everything to look natural and Nvidia has been working hard on AI animation tools that they use on Jensen for presentations now. Hopefully in some years it will pay dividends over present methods
Looking forward to seeing video and 3D worlds blend, so that VR movies would have depth. The mentioned Diehard, allowing a VR user to move around the scene. Imagine being able to go to any scene in any movie, removing the actors or elements in a scene, and being able to be in the environment by itself, changing the actor, their clothes, etc.
Love how meshletsseem to strategically choose which pieces to kick up to a higher LOD beyond just the Z-depth.
Same goes for the massive improvements in photogrammetry to scan objects and extrapolate the corresponding maps, to the point where they'll probably soon even generate node-based shader setups for advanced surface types like color-changing car paint.
Does anyone know what creator he showcases at 10:15?
You create graphs and charts in Photoshop? Mate, that's what PowerPoint is for!
That is what this is on RUclips, I somehow thought it was displaying the audio volume at that point.
I have been a professional 3D artist and i have used world creator, unreal engine and everything, from making game env to photorealistic scenes even working as environment artist for games and TV shows, believe me guys, its not as easy and simple as it is showed, a lot of hard work and time is spend to make it realistic and let alone believable, also the video he is using in the very beginning is from artist named aron westwood(search for " I AM world creator software" on youtube), and it took months and probably years of research and learning to make something like this, i dont want to spread negativity, its better to know the entire mountain while desperation might eat your time.
These technologies are definitely going to be cool, but a good game lives, or underperforms, with compelling stories and characters. Let's hope they have more time for good writing in the future. lol
Where is the video snippet from 6:44 from? It looks interesting
Currently playing AC: valhalla.. The opening area looks a lot like the first opening shots of this video. The jump between them is insane really.
Looking forward to the 4096 core CPU, RTX 40090Ti, 500 zettabyte SSD's, 256 Petabyte of ram AND a nuclear power plant to run it all.
AI objects are very interesting. Imagine an open world game where you can simply outsource interior layout, furnishing, and decoration to an AI such that every single building interior in the city is unique and is rendered by an AI, freeing up time for programmers.
10 years if progress is linear.
Right now AI is being described as super parabolic. A hockey stick on a hockey stick.
Hyped for CES ! Hopefully great new laptops annnounced
I am not hyped for anything anymore. Wherever you look,its absolute robbery and trash products melting left and right.
Yeah, remember to sell your kidney.
imagine using laptops and not interfacing your pc from your cellphone
Dystopia is now old man!
So what was the footage of the forest and beach from?
After watching your video about plague tale requiem, i played it and finished it. I'm just blown away by the graphics. It just looks much better than anything out there to me. Are there any games in your opinion that can match it graphically?
the usual ones. Red dead redemption 2 , last of us 2 , god of war ragnarok, Horizon Forbidden west. They might not match Requiem but are the closest ones available out there
The answer is probably no, for now
Joke post?
@@Instant_Nerf you're the only one who thinks it's a joke.
@@TropicChristmas @Norma thanks
I see you beat me to talking about Neural rendering, although you haven't really taken it to the limit of one of the articles that was on your website.
Has anyone thought of trying high resolution etc with "Pulsing", changing the amplitude, angle, polarisation, tone etc in a subliminal way, just as with Showscan where the number of frames per second, was such that it induced intense emotions in the viewers, also the ability to change reflections, refractions, etc could be used for 3D effects.
Can you upload in 4k?
i never seen Nanite being called "meshlets", specially as meshlets is kinda used to refer to something rather different.
with Nanite does is commonly referenced as as Virtualized Geometry,
you have the geometry in this massive hierarchical data structure ,rather actual meshes in memory, and it is organized in such way as you can you get exactly only the geometry you want and with how much detail you want for each part of the scene. every single frame.
that why it is called virtualized, you still rendering triangles in the end, but what is actually being rendered on each frame is generated every time it is used. not something that is already in memory
the secret sauce of nanite is not even that the meshes are broken in tiny parts. quake 1 did that, ps1 games did that.
but in how the heck you organize your geometry in such way that you can pick a lot of it or little of it and still end with a list of triangles that can be rendered on screen that still looks like the thing you trying to render. and all that fast enough to be done every single frame.
heck fact they couldn't even get current gpus rendering pipeline to play nice with the geometry generated by nanite. and it is(was? things might have changed) actually a software renderer implemented in gpu believe or not. which would be bad, if was not for the fact that the advantages outshine the disadvantages 1000 to 1
(if you try the exact same scene with traditional rendering and with nanite, nanite is actually faster and uses less vram and won't have LOD poping, is almost like every GPU from the last ~7 years in the planet got a ~30% boost out of no where)
while meshlets are usually the things mesh shaders work with, its a gpu thing. it is related to some of the operations that nanite does. but mesh shaders is more about: "i have a mesh and i want to do things with it efficiently directly in the gpu", i never ever seen it being used tbh
He's not totally right, but he's not totally wrong. "Meshlets" refers to the practice of breaking meshes up into hierarchical LoDs of fixed-size "patches" of mesh, and moving the LoD, culling, and dispatch of those patches all to the GPU instead of the CPU. While Nanite doesn't use the specific hardware features inteded to facilitate meshlets, it's still fundamentally using meshlet rendering.
It's not true that the triangles are generated each frame, but it may be true that meshlets (patches of triangles) are streamed on-demand from the disk more or less straight to the GPU. What's generated each frame is the list of things to draw, what we call dispatch. What made "dispatch on GPU" possible was compute shaders (and meshlet shaders) and we just hadn't realized it until now because it's a very unusual and inspired bit of lateral thinking about rendering.
Quake did not break meshes into tiny parts. It used "brushes" to represent 3d objects, which is the intersections of planes to describe convex solids. These are easy to do certain math on like CSG and splitting into BSP-trees.
"Software renderer implemented in gpu" - partial marks for this. Nanite has a software *triangle rasterizer*. The hardware rasterizer quickly loses performance when triangle sizes approach one triangle per pixel. The assumption made in hardware rasterizers (that triangles will be large on average) is very good for traditional render pipelines, and poor for Nanite.
You don't really get a VRAM boost out of Nanite unless you really embrace the technology and get rid of your normal maps and displacement maps. The savings is texture memory and not everyone is using it wisely like that.
I worked for Metaplex this year, it’s looking really rough for them in the near future. All the OGs are being laid off and my project was affected heavily.
I still believe in the technology 100%, it’s just that they trusted the wrong people to work with them. SOL is still the best tech around but I won’t lie, it’s been eye opening
Sry but i don't think you should make a "graphics video" in 1080p resolution, no matter if the most ppl use 1080p
Any kind of ship or building would get an AI assets set would get many sims a variety they lack right now (or was a hard work adding it like cities skylines). Imagine if dev would make core of the building and AI would move windows and balconies around, throw cracks and damage, add rooftop chimneys and stuff like that.
you could easily get a set of many non-repeating homes to use in cities, with variety, adding paints as well, we get a real lib.
using this to apply damage to things could be interesting as well.
Cracked scopes on guns and different cracks on every gun you pick? I feel like simulating damage on your equipment can get much more interesting with this.
This is going to fuck up artistic views in favour of "more realistic" graphics in games. And it is just the beginning.
No? none of these technologies prevent artists from making stylized games.
It's definitely the case that realistic graphics wins over any other kind, at least in VR. Stylized graphics in VR are both harder to look at and generally look worse than realistic graphics.
@@thelelanatorlol3978 you have no idea what youre talking about
@@diwajerebation4077 I literally own and daily use a vr headset, im pretty fucking qualified.
Oh, wrong comment wtf
Honestly, the fact that industry is acknowledging that users attention span is becoming shorter, and therefore making a tool that encourages such lacking, is not a good news...
8:32 can't see any such viewer information anywhere
I've never heard of AI objects as a concept (I'm a professional 3D artist and game developer) I can't find anything online, can you link some resources?
Search for "neural rendering" - essentially in a game, a neural rendered object is something that's inserted into your scene at the end. IE your object might be a "panda" and if you put it in water it becomes a "swimming panda" - so it's not something you process into a mesh and render classically - you lose control of how it renders after you state where it appears - hence its not described using vertices/voxels etc etc
Think of it as an AI image generator that uses the frame you just rendered as the input, then inpaints all the panada references in the final scene.
@@JohnnyWednesday wow, cool!
Enlightening! "All you will need is the script." Bring it on!
The world creator tech really interests me. I play around with a universe simulator that works on the same procedural generation but is still a ways from how good those examples are but over the years is continuously improving towards photo realism.
@J.C. Denton Space Engine.
Why does it say deloitte at 6:53??
3:55 the narrative that new technologies will allow games with "better ai" has been around forever, and while there *have* been some milestones the big leap never happened; because especially AAA companies are still mostly focusing on eye candy.
It would be pretty cool if these models could be turned into 3D STLs for resin printing.
Have you ever wanted to watch a movie, but saw that there are too many things to watch, and watch nothing in the end? 11:27 will make this stronger and it is very scary. I can't imagine the amount of analysis paralysis we will have when deciding not only what to watch, but which edition to watch.
I still feel physics is the next frontier for games.
8:14 Will this finally mean the end of "10 minute and a couple of seconds" videos with only 15 seconds of relevant info?
- Wow, Nanite is so revolutionary! how did you achive that?
- Well, we changed texture streaming directory from one to another...
forget GPU rendering on servers and streaming it back to the client. that would be like going back to the beginning of computers where you had client stations that alway connected to servers. no one wants that, not even for rendering. also there is the issue with speed, stability and stuff. and i know because as a software developer i did an application where a client send midi to a server to get audio data as a response. this must be very fast to have almost no latency. this works pretty nice for a few seconds, sometimes a minute until there is the first package thats lost over the internet. requesting that package again breaks latency. and using UDP instead of TCP will solve latency but that means a lost package will stay lost. you would have dropouts. but then again you might send data twice if the speed is very high. in the end its a very hard thing to do. don't expect this to be solved within 10 years. just my opinion since i worked in that area
90s gaming was filled with such a vast range of creativity even with their limitations. I love the amount of risk they took on odd concepts in This era- as well as the PS2/360 era. It should show us that GRAPHICS/Reflections are not the most important part. No. If you can create satisfying movement, gameplay mechanics, physics effects, responsive parts of a environment, a environment that makes you want to explore it, engage it, etc. That stuff is so important. Not just "who can make the largest game world, or a game with the best Graphics." I think Art style, design, vibe, aesthetic is much more important. You don't need top notch graphics to greatly appreciate and enjoy playing that game. That's why we are seeing such a resurgence of players who have gone back to playing older games because they realize the aspects of gaming that they care about and are unhappy with in most modern games. So hopefully gaming will enter a new era. That mixes in aspects of the older era's. With our modern capabilities that will hopefully help make it easier to create a satisfying game. As well as a vast range of game's Instead of everything trying to be Triple AAA or free to play online battle Royale games..
currently youtube shorts - and other shorts - is just a social media crap - it has nothing to do with most watched part of video or the most condensed information.
Thank you, coreteks, gr8 content!
You totally missed ai rendering...its alreay been shown in the famous GTA5 ai rendering demonstration where they even go so far as present it as having a "interactive framerate". This will basically blow everything out of the water you presented here, still thanks for the video.
And I still... can't..... save game. So I have to replay levels over and over to make any progress, and don't experiment at all. Which I can take for about an hour. Everything on the PC had "save game" until consoles took over. Will these new developments help with that? Been waiting since Crysis.
Well... that's a choice made by developers. Saving is as easy as saving all the states of all the objects - it didn't become any harder that 20 years ago.
@@igorthelight I know. But it's a main reason I no longer buy action games (which I otherwise love). And all the graphic goodness in the world won't fix it. I still read about the tech but that's all.
@@anslicht4487 Most modern games use frequent autosaves. Some let you save in any time (Cyberpunk 2077). Replaying the whole level is quite a rare thing, no?
@@igorthelight By whole level I mean since the last autosave, of course. Which might be ten minutes of intense action. I absolutely *don't* want to keep repeating that at one time. And I can't even break from the game when I need to! AND a quick-save as in Crysis let me experiment endlessly with combat approaches and techniques, exploration etc. So it's still fun 15 years later. *Deleting that feature is the single most stupid "development" in games **_ever_* .
Don't get me started 🙂
@@anslicht4487 Fair point!
Before playing any game just make sure that it has "Manual save" feature.
Wow, that is some food for thought. As as relatively 3D artist specialized in asset creation, who has just recently starting working, what would advice be to somewhat prepare for the coming shift in graphics?
I am confident in the quality of my assets, yet that may not matter that much soon. I have no programming knowledge aside from web-stuff, but I have a solid technical understand of graphics for an artist. One idea would be to lean heavier into ‘technical art’ and maybe procedural generation generation, thinking that it will converge with AI generation, right?
I'd guess supervision of procedural ai generation will be in high demand in near future. And single-click solutions still wouldn't be enough. I can see a lot of fine tuning of parameters, combining several ai tools in unexpected ways, teaching it particular art styles and leading it to desired objectives - things like that. And for now it seems basic scripting in python is enough, and there sure will be plenty of no-code solutions in future.
Hope some of then will help me with my web app developement. But the hardest problem in develpment is, that there always grows some special kind of mess which actually nobody understands. Something with friction between wetware and hardware abstractions :-)
My recommendation would be to understand the principles of operation and construction of the real world objects that you model. As an example, I've worked as a contractor doing building renovation and maintenance. I've also done level design for FPS games. I find it a little strange sometimes when I see maps built by someone that doesn't understand of real world architecture or building design. There's numerous ways of how to convincingly place assets, to make it look like they really belong there. As an example, lets say you want to place a series of lights down a hallway. As you progress, the bundle of wires that feed each lamp should get smaller, until there's just a single wire at the last lamp. I've seen too maps where people just stick a mess of pipes and conduits in the hopes that it looks "sci-fi". For me, it breaks the realism. I've actually installed these things in real life. When I walk into a building I can visually survey the place, and know what the infrastructure is supposed to do. The more you understand the fundamentals of what you're doing, the more you'll have an edge on others in the field.
@@lihtan Wow, thank you, that is some really solid advice. This type of knowledge is something I knew was helpful, but I did not think about it enough to see it as something that I could 'make my own'. I have no construction knowledge, so I do not often look for these things in games, but I know many 3D-artists don't give these things too much thought, and it is almost never mention in all these breakdown articles, videos and blog posts, although it is very important for realism and modeling efficiency.
The hard part is of course acquiring that knowledge as an outsider, but if it was easy, everyone would do it, I guess.^^
Fundamentals. AI won't for a very long time have a good understanding of things like edge flow, T-pose versus A-pose (protip: use A-pose to preserve range of motion when you rig it), compact/efficient UV layout, and how many fingers and toes a human has. Don't auto-retopo if you can avoid it, that does things AIs are good at and often doesn't make things pleasant for an animator down the line. As long as you keep thinking about the value chain connecting you and the other actual humans in the pipeline you can 100% beat an AI in quality.
Instead of being crushed by this juggernaut of disruption for commercial artists, I am interested in riding the change by adapting to requirements. Will commercial artists of the future be the ones who are expert at stating the best prompts? What is your vision of the future for graphic arts?
What we see @ 5:12 looks more like semi procedural generation than AI generation to me, no?
AI will absolutely revolutionize the gaming industry. AI dialogues, terrains, interactions, models, content, etc.
For VR and AR to go mainstream, we need to find a way to make our bodies trackable for computers without having to wear cumbersome motion tracking suits.
Unless your monitor is at least 40" big and you sit within 50 cm from it, 8K monitor still doesn't make any sense because the resolution of the Mark 1 Eyeballs in your head is not good enough to see the difference between 4K and 8K image, never mind the computing requirements to actually render that 8K image.
Spend your money in more accurate color reproduction, stable gamma ramp over all the image (even IPS panels have different/incorrect gamma ramp for the edges of the screen when you sit close to it) and dynamic range, because those things you can actually see.
WOW! What a time to be alive
an artist who uploads really well made videos to youtube called Jazza, brother to the shadiversity guy who talks about armor/weapons in real life, is making a video game at the minute, and he was soo impressed by A.I generated art for things like jewelry, that he said on multiple occasions in different videos that he's using a.i generated art in the game. I'm assuming for U.I elements, also things like inventory. its actually worth a watch, he's really impressed and hyped by what things like midjourney and dall-e produce, and he mentioned how everybody seems to hate A.I these days. but he even reiterated in the video talking about the hate that he's going to continue using a.i generated artwork for the game.
Nanite is not a stable technology yet, there is still a lot of work that needs to be done and also anyone who've tried writing shaders for UE5 will know that, custom pixel/fragment shaders are extremely heavy(slow to run) in current version of UE 5..
what isnt heavy in ue
@@diwajerebation4077 A mirror ball on a checker
WHAT ABOUT 7900 XTX WHICH WAS SOLD IN ALL AROUND OF POLAND ?
your voice will REVOLUTIONIZE narrating
This is more like it, Awesome video. +1
More positive tech news outside the usual stuff we read on tech websites.
Remember to touch grass every time the Wi-Fi drops.
Let's goooo! 🔥🚀
No matter how realistic games look or how realistic ai voices sound, humans are always somehow able to detect them...I think we vastly underestimate ourselves
You could dynamically generate high quality textures using AI, so a game could just have a snippet of code that precisely describes the desired texture for a deterministic AI image generator to create the textures at startup, or even at runtime, and effectively compress hundreds of GB of textures into text files that weight just a few MB, though of course AI objects would already not need textures saved on disk. Heck, you could even code an engine to dynamically modify or completely recreate a texture on demand, so for example if a gun got damaged you could directly modify the texture, or the mesh, to add damage based on what happened and store it permanently by using a single AI modell
I’m working on a technology that watches all this auto created media for me, so I don’t have to. Frees up tons of time to… wait, what!?
Now all we need is an AI to enjoy all the AI generated content so that we don't have to exist ;)
I am sure this is already coming down the pipe soon
Bruh 3:40 is literally just drone footage like that's not a game my dude they just recorded that on their drone it's not an actual game
in the next 10 years, we won't have to use assetes anymore. an AI will recreate an scene, without using tools like blender anymore
Technology like this makes me question reality....
"Today's sponsor is..."
(hits fast forward several times)
I assume by 2050, good chunk of generated media is created by handful of people using AI toolsets, today we can generate very decent AI voices without voice actors involved, we can create models, images with AI even basic videos can be made by AI.
Whats scary about this is the implications, if every man is a game studio or movie studio, you oversaturate the market so much, you gonna destroy entire industries, actors will find themselves increasingly out of jobs so will movie and video studios.
Why hire 500 people to make a 50 million dollar movie when you can hire 5 people who can make the same movie for 5 million using AI toolsets and all within 6 months. In a way this also liberates entertainment but at the same time its a curse.
I play mostly VR games these days and i can't wait to see the graphics we will have soon! I just hope my new 3090 will be able to run the best for a while.
What VR games are there? I’ve held off buying a headset cause there’s only one game I wanna play.
@@Freshbott2 my top five PCVR games are half life alyx, lone echo 1 and 2, wanderer, beatsaber and blade and sorcery. There's so many good games for all tastes, you just have to give them a chance. A lot of people love sim racing / flying in VR. Maybe look at games you like and see what VR games in those styles exist?
Damn that's one nice gaming set up.
Here we have massive taxes, I'm trying to build a 12100f 3060ti but it seems I'm looking at something around 2400usd taking into account monitor and maybe tge desk too.
Fuck taxes.
What I worry about is that AI will lead to lazy artistry.
Imagine that there is an open world game where there are various settlements scattered throughout a variety of biomes. Would the buildings and architecture of a small town on a tropical beach be the same as those found in say the biggest city where the King lives in his castle? Probably not. The beach buildings would be small, likely constructed out of wood and straw, materials found on or near the beach. But the castle town would be brick, would have a medieval feel to them. AI would be able to construct both, but only a good designer and an artist could place both in their appropriate spaces and create the game's lore that supports both settlements. If a good human designer wasn't there the AI could just place whatever building style wherever and shatter the immersion of the player.
I'm worried that money driven publishers will see AI as a quick and easy way to cut dev costs so they can maximize profits, at the expense of the quality of the end product. "No need to higher those expensive game designers, we have the latest in AI tech that can do the same thing," say the CEOs who aren't gamers and never play the games their company produces.
Most people don't care about lore that much though, if the world design isn't too jarring most people don't care.
And also like you said: money talks, bullshit walks.
Oh wow didnt know world creator this is amazingly realistic
Some day, i will ask a computer to remaster "Aliens-colonial marine's " xbox game, and everyone will see what i see today.
We've been pushing for photo-realism for decades, but what good is photo-realism if we can't dynamically alter scenes at will?
We should be able to dynamically tear down buildings, having them collapse and leave a pile of realistic and movable rubble on the ground.
We should be able to fire a mortar in some random direction in a forest and have that open a crater surrounded with broken trees and debris.
We should be able to move mountains.
That, to me, is the future. Voxels (the real thing, not Minecraft) are the future. Not necessarily photo-realism.
Hier von Gronkh, du auch?
How close the details it will be?? Can see detail at um, nm, and beyond? So, when look far so you can see the blurry, reducing light and so on, well, the more high detail at low um, high nm, low nm, ya will see the different when look far away, so what is the highest details before you can't see low detail at far away since looking the same unless huge since it is about scale, so 4K and 8K look same if at certain size, so need to be huge. So, unless you produce a giant, the bigger the giant will be equal to the detail at small level since the scale is same. The ratio of details to certain size so not see reducing something so look all real as real life is photorealism, higher than hyperrealism, so for hyperrealism you can see it is not photorealism.
If I had to rank elements of games. Storytelling is the least important element of games.
the lighting yyour showing irt wont run on game
They are mEshlets with an E. While it mashes meshes together a little bit of a mesh is a mesh-let... I hear this term a lot so the beginning of the video was really painful to listen too.
I am +50 and still trilled with all that is comming on AI graphigcs whatever.
Still, we don't have a decent image to 3D models (maybe something like Monster Mash...on steroids, and for free). Can you imagine a full (good) 3D model generated with a single image?
A free AI voice generator without robotic sound would be awesome as well.
And some free mocap generated by a pre-recorded video with a single camera, with foot locking and hands detection (PlaskAI on steroids, for free), maybe a face tracking from pre-recorded video data from a single camera. For free... And...
I ask for too much, I know haha
What is the advantage of an AI object?
Why does this remind me of The Matrix, Total Recall, Johnny Mnemonic, etc? And you thought Fake News was bad enough now. "Tune in and drop out" :D
Lets go play laser tag in a football field sized holodeck that you can choose your battle field. 😳🔥
My youtube already has Dynamic video 😳