The Evolution of Video Game Graphics Is Complicated

[@Innocenceii]

******_EDIT, PLEASE READ_****** As a lot of computer scientists/people who work in the industry have pointed out, unfortunately I got 1 fact slightly incorrect about the history of compilers and Cartridges. In hindsight, I'm not sure how I managed to let this one slip through the cracks, but I want to make sure as many people leave with the right information as possible. So, compilers were never put onto consumer cartridges. (They were put on developor cartridges though, and that's what Ed was actually echoing in this video). This is because the C language can be compiled then exported to a cartridge separately ( Duh! How'd I forget that!?). However, since compiler technology at the time was so inefficient compared to today, C code compiled into assembly language at the time would still be serveral times the memory size of assembly language coded by hand. (That is true even to this day, compilers/interpreters have gotten ridiculously good at producing optimized code but, in theory, nothing will beat someone who is either crazy enough, stupid enough, or patient enough to do it by hand!) So while compilers themselves weren't put on the cartridges, the cartridge size penalty for using them was still a very much real event that plagued game development for quite some time. Sorry for the miscommunication, hope you enjoyed the video nonetheless! Enjoy your day 🙏 ***********_END EDIT_************ Yo! Been a while, but hope you guys enjoy this one. This was probably the hardest video I ever had to edit, so hopefully it shows and you guys got a solid 30 minutes out of it! It was also super fascinating to work on, since it hits close to home with respect to my studies. If you haven't already, check the description for more info! Until next time ✌✌

[@timsmyth7942]

you came back with this? all is well again 🫡

[@LeloucheViBritainia]

It's seriously been a long time. Glad to see new video

[@blackslashmail]

Big ups for sale his video 💜

[@coolmacatrain9434]

Graphics not as good in 2022? ... Because this generation of consoles has been designed with 4K, HDR, 120hz & RTX in mind ... and so developers have had to go down that road as well. The problem? ... most gamers don't have access to the type of TV's needed to see any of that stuff

[@mrdadecounty3056]

Bro I love all your videos already shared on Twitter bring u more views and subscribers.. you would be someone I can sit down with and have a conversation for hours about tech.

[@NeophyteG]

Those old school dudes were awesome. they had to grind that hard to make video games. If it weren’t for them we wouldn’t have the games we have today. Crazy hats off to them. 😊

[@MicrosoftClubxD]

Yuroichi!

[@princssnasty]

not that it isn’t just as much of a grind to develop games today, but yeah that assembly language shit was wild. loved that guy’s stories

[@how2pick4name]

Thanks. lol

[@darkzeroprojects4245]

I tbh am trying to learn from those old ones.

[@princssnasty]

@@darkzeroprojects4245 why? as jaw-dropping as what they did in their time was, none of that shit’s useful to making games now

[@PowerUpT]

My thoughts on the matter have always been that people will stop caring about game graphics fidelity in the coming years because it is plateauing. People will then start to care about the artistic integrity of how things look instead of them looking "Realistic", which will be a great thing. The funny thing is: there is historical precedence! Back in the olden days of the 80s and early 90s, people always tried to get the best sound possible in games, making voice clips audible past being heavily compressed to fit in games, and trying to get music to sound like actual instruments. Then the day comes where CDs became commonplace, and people were stoked. The SEGA CD led the charge on consoles, CD Audio was on PCs, and eventually the PlayStation cemented CD quality audio as a part of gaming. Then the magical part happened: people stopped caring. The sound advancements became so good that it doesn't matter how much sound replicates reality, but rather how artistic sound is uitilied. This is what I'm excited for in gaming: For artistry to usurp fidelity, and I believe indies are leading the charge already because of limitations.

[@jerrodshack7610]

The same thing has already happened to animated films. The trend of ultrarealism has finally been abandoned in favor of much more stylized animation driven by art direction and the type of exaggerated movement only animation can accomplish.

[@rohitchaoji]

I like to believe it's kinda like photography. A professional mirror based camera that costs a ton of money will probably give you some really high fidelity and high quality images, but you still need artistry for the photograph to be any interesting. If you don't have a sense of composition, lighting and knowledge of colour treatment in post-processing, no amount of technological leaps will make a mediocre photo look great.

[@coder0xff]

When the focus on technology ends, the focus on gaming begins.

[@aryantiwari7105]

What background music are you using ?

[@LurkingLarper]

Well put. Nintendo is a good overall example in this since Switch is woefully inferior in hardware specs compared to its competition, yet can run one of the prettiest games ever in Breath of the Wild. Style, artistry and art direction in general have been way more important to me for half a decade than raw graphics alone. I think it's only a question of time before the rest of the industry catches up to that.

[@imbored1179]

while i always prefer art direction and style over raw graphics, i cant help but fawn over the graphical improvements to games over the years. ghost of tsushima particularly blew my mind back when it came out.

[@Drstrange3000]

The reason that game looks so good is because it IS stylized. It also has some really impressive tricks and technological advances.ents in how it renders it's foilage and details. Still one of the most gorgeous looking games I have played.

[@stratosphere2323]

The PS5 version of Ghost of Tsushima specifically takes the experience to a whole new level. I just recently finished my first playthrough on the PS5 and it was absolutely stunning.

[@crestofhonor2349]

For me personally graphics and art style have always been one and the same. Art style is part of the graphics and is one factor in what the graphics look like. A realistic game isn’t necessarily graphically superior to a more cartoony or more CG movie like look

[@fantasyconnect]

As others have said, you like GoT not because it's graphically impressive, it's not, actually. It has low res low and low poly environments with somewhat more complex character models. That's how it loads so quickly in fast travel. What the game does have, is a solid understanding of world design and an almost painterly touch to the regional colour schemes. Every area is almost monochromatic. Yellow and brown, off-white and green, purple and earthy tones. The wind effects, the light on the grass or on metal etc. The game is masterfully made.

[@kamay7021]

Ghost of Tsushima looks meh to me

[@matthuck378]

Photorealism is just one of many art styles/aesthetics. It's also the one that non-artists tend to like the most, and the one that tends to age the worst.

[@Yuri-xx2gi]

And the one that makes a game take forever to launch

[@DragonOfTheMortalKombat]

Yeah it does age worse because there is so much potential for improvement but photorealism is also the easiest to mod to bring back to life and often the most immersive. Not every game should be photorealistic though.

[@vogonp4287]

​@@Yuri-xx2gi And cost the most to make.

[@somaelselino7627]

Wise words, sir.

[@henryfleischer404]

The advancements in raytracing these last few years have been amazing. In early 2020, A good render in Blender took me hours. Now it takes half an hour.

[@matthewirizarry8467]

When i was doing 3d modeling and animation in the early 2000s, a photo realistic render took me 12 or more hours. The idea of raytracing being done in real time at the consumer level is still mind blowing to me.

[@Jlewismedia]

I don't think raytracing is the future *Except for baked lighting, like in blender and game engines. Whatever we can do with raytracing in realtime on any future hardware, we will be able to do more than twice as fast with other techniques in realtime.

[@matthewirizarry8467]

@@Jlewismedia I dont really get what you're trying to say. Other techniques? Like what? Volumetric lighting? Every game engine i've worked with has been pushing toward something like raytracing, or trying to emulate those results. And they did so by using multiple lighting, texturing, and shadow casting systems and trying to blend them together. Raytracing allows the developer to get rid of all of that in favor of a single all encompassing technology at the hardware level. Any game can have near perfect lighting, on any surface for little to no performance cost and minimal development cost. Raytracing is definitely the future. The only downside it has is market penetration of the chipsets that support it, but that will not be a problem as more manufacturers develop their own raytracing implementations.

[@Jlewismedia]

@@matthewirizarry8467 It's a total opinion, I program 3d graphics for work and use unity/unreal/etc. Raytracing is super expensive, currently lighting in most games uses something like a direction of light/intensity that is applied to a shader based on it's distance from the source. Usually it's repeated for every single light point or it's baked entirely and no light is actually calculated but sampled from a texture instead. It's a VERY fast method of rendering realtime lighting. Raytracing on a programming level is just way more complex, it's 3d mathematics on top of your shader code. UE5's lumen uses signed distance fields (raytacing) but can run on non RTX cards, so that could be the future indeed. There is also volumetric/voxel lighting as you mentioned which is usually a mix of the two or some other equation. For it to become mainstream we will need enough demand from software to develop the hardware further. If that happens and RT keeps being adopted and used then I was wrong and its absolutely possible. But I think something else might pop up along the way that doesn't require dedicated and expansive hardware to render. It's also an accessibility thing, if you built your game engine around only rayracing, everyone without a rt card can't play your game. So if that became the norm people on integrated graphics, kids with cheap laptops, etc. would totally be unable to play games on pc. But again, it's total opinion it is absolutely possible that raytracing will become the default rendering path at some stage, I just think something else is going to pop up.

[@temotskipuri3151]

​​@@Jlewismedia who cares about kids on cheap laptops lol. That's not the developers problem. The kid (or his parents) just has to buy a better laptop, or if they're that broke, just not game at all.

[@RamblinRick_]

I've been a developer for over 50 years; wrote my first program in HP BASIC in January 1971. I've used about a dozen different languages plus about 8 Assembly languages for different processors. Hadd to use them to write devicce drivers for my first home computer, SWTPC 6800 (circa 1976). I really enjoyed Assembly and C++ the most. Wrote some games for Win 3.1 in C++. Good times...

[@hemangchauhan2864]

The thing really like about this is you've actually went out and talked to engineers. You actually went into the technical details to back your arguement. Amazing effort, I really hope people get inspired from it.

[@OdinStarz]

Dude! What a video. From the script, to the history, the editing, and love. As an IT specialist who specialize in software enterprise development , it is crazy to see what the modern languages of code development allows us to do. I reuse code all the time for the software we make and it never dawns on me that that in itself is a privilege to be able to do. Anyways, you are highly underrated, and if this is the quality you are looking to push out this year, surely it will pay dividends. Keep up the great work!

[@Allplussomeminus]

I'll say this: playing 3D platformers in VR is a trip. I played Super Lucky's Tale and couldn't get over how it felt like I was playing in a living toy box.

[@ChrisSowa]

There is absolutely no reason a compiler would have to live on the cartridge of old systems. I think your interviewee lead you down a wrong path. The compiler lives on the host that is building the ROM which will eventually gets loaded onto your cartridge memory. The compiler is just part of the tool chain that builds the ROM. I have a feeling he was trying to say that c-compiled builds were larger compared to something written in ASM completely from scratch. This is because the whole point of the compiler is to interpret your c source code and transform it into machine code, but it takes a lot of liberties getting there.

[@tablettablete186]

I was wondering about the same thing, glad I am not the only one!

[@thethirdtomas3]

So the compiler is stored on the host? How much storage did these old systems have?

[@DivinuZ]

The compiler is a program on the computer where you write the code and it’s job is to translate (compile) C (or other) code to Assembly code so you don’t have to write the Assembly yourself. However, and this was especially true back then, the Assembly a compiler generates typically is more bloated, less efficient and not always aware of CPU specific instructions, compared to what you can produce manually.

[@steverempel8584]

I think what they meant is that all the boilerplate code created by the compiler was too big for the cartridge. A huge chunk of code needed for just a "hello world" program, but doesn't really grow from there. Try making a Microsoft word document, with just a blank page and save it. You'll see just a blank page takes up room. It's sort of the same thing with a compiled C program.

[@user-sl6gn1ss8p]

on top of that, old systems could be very specific in how they worked, due to limitations, and compilers where still much simpler, so being able to directly direct the hardware was arguably much more important for performance as well

[@humanbass]

My favorite graphics are from the Snes: Yoshi Island. It was actually quite a feat by its time, and looks timeless.

[@Syngrafer]

If I showed my kid Super Mario World, she wouldn't like it because of the graphics. However, if I showed her Yoshi's Island, she'd love it. A unique art style goes really far.

[@Anonymous-Wolf]

Toy story 2 looks pretty good , I mean yes of course it is low poly but they have nailed the colors and lightings same goes for wwe here comes the pain

[@promero14]

WindWaker, BOTW. Thoes games are going to be timeless.

[@Syngrafer]

@@promero14 I'd say WW already is. If I was going by graphics alone, I'd much rather re-play WW than TP. I'd say BOTW will be too, but it's still fairly young as a game.

[@traviscunningham7062]

Don’t forget Zelda SS. Underrated masterpiece with a beautiful art style.

[@EmilianoGonzalez11n6]

What a comeback video! I rarely comment on videos but you went above and beyond, from writing to editing. Top tier.

[@jaefrmbk2k]

💯

[@bob_0146]

Great video man. I definitely feel that SSDs in consoles are revolutionary at least in terms of game design, but we have yet to see the results of that as the vast majority of devs are still making cross-gen games.

[@xfunkypicklex]

This is something I would certainly like to see as this generation continues and support for the ps4 and xbone platforms reallly starts to drop. Maybe there's more games that I'm just not aware of like this but dead space remake is the only game I can recall that warns you on the PC platform if you are running on a traditional hard drive.

[@JayDaGod_1]

Welcome back bro! Keep the technological-based videos in the loop! I think they’re you’re bread and butter for sure

[@Kniffel101]

12:30 Uhh... that's BS. The compiler runs on the computer, and outputs the binary code to the computer, not the console. I don't know what Ed was trying to say about a compiler "taking the space of the whole cartridge" a couple minutes before that. Maybe he was referring to the C standard library or the code the compiler generates? You may wanna write him a PM and ask him to clarify what he was trying to say, so you can pin a comment and people don't go away with incorrect knowledge! =) Anyway, cool to see you're back!

[@PatrickJBauer89]

I feel like he tried saying, that bytecode generated from 70s/80s C compilers was a lot less optimized and probably to big to fit on let's say a 2kb ROM chip.

[@Kniffel101]

@@PatrickJBauer89 Yeah, that's likely. Though a C compiler generates machine code, not byte code! =D

[@tablettablete186]

12:05 Wait, I still don't get why you had to package the compiler with the game, you can simply compile your C code in another machine and just put the final results in the cartridge. In fact, this is how we distribute software nowadays, compile the code and just send the binaries. The exception is when you have a JIT compiler/interpreter, because you send source code and compiles it in the client (JS, Python, Lua...). As a developer, I really didn't get this. Maybe the C compiler wasn't mature enough?

[@PatrickJBauer89]

I stumbled over the same part and agree with you, that it's a false statement. I researched it a bit and reasons for using assembly over C were extreme memory limitations, so that the memory has to be managed manually by devs. Compiled code (especially at that time) would produce byte code with a lot less optimizations than newer compilers would generate today. Some games would even hack around using specific CPU instructions on specific cycles to produce specific results, which you couldn't replicate in a higher level code.

[@tablettablete186]

@@PatrickJBauer89 Well, this makes way more sense! So compilers weren't able to produce the code that the devs needed, so they had to use ASM. Thanks for taking you're time to research this and make this answer, really appreciate!

[@MagninGames]

Wow! What an excellent video. No wonder it took you a while to polish it so well. I would recommend it highly to all game development students. It gives them a good idea of what came before them and where the future may take us. Thanks for such an excellent video!

[@Innocenceii]

Ed!! Thanks so much for being a part of it. Your perspective really helped clear some thoughts up for me. Would do it again in a heartbeat!

[@minecraftermad]

You can just compile c to asm and shove that on the cartridge. You just need a better machine for coding than your target is. C compiled code might be larger due to the fact that optimizing for asm size is difficult. Or rather is easier if you just manually optimize for size. So in short. You don't need a compiler on the console.

[@Magnus_Loov]

The problem was probably memory limitations on the machines themselves. I guess they eventually got some special C-compiler cartridges that had some additional RAM memory on them so that they could do the coding directly on the console. I don't think they could make the coding on other machines back then. I don't think there was something like the special development systems that the later Xbox and Playstations got. The budget for that wasn't just there for the old 80:s consoles.

[@user-sl6gn1ss8p]

@@Magnus_Loov I really don't think they actually develop on the console itself. You'd at least have to provide, for example, a keyboard port and software to deal with that (unless you expect people to program pokémon using the gameboy's d-pad : p). And where would that software be, anyway? You'd need a cartridge with double rom banks or something, one of which the system would have to be able to write to. The minimum viable product for that is already a fairly complicated development kit, really. Having a "pc" with assembler software for the target platform connected to some rom writter thingy is actually a much more simple setup. Also, I see no reason why your assembler would have to run on a native platform, assemblers are fairly simple and emit machine code after all. That being said, compiled c code would tend to be larger than direct assembly, especially back then (less optimized compilers and more idiosyncratic platforms), and especially at the scale they would be working.

[@MichaelLandes131]

Yeah this part of the video confused me. There's no way it would ever make sense to put a compiler in the cartridge. Maybe they meant to say that the c runtime would have taken too much space, that would make a little more sense.

[@TorQueMoD]

Great video man, but we're not even close to being finished with breakthroughs with game engines and rendering. The difference is we're getting extra details on a micro level now, rather than the macro that we're used to. So you get things like higher quality animations with features like motion warping, procedural contextual animations, increased fidelity for meshes, better physics simulations, dynamic lighting, shadows and reflections etc. As impressive as an engine like UE5 is, there's still a lot more coming in the next 10 - 15 years. It's just going to be harder to notice in a side-by-side comparison because it's going to be in ways that you don't even see in a static screen shot. These are all things that I know you classify as itteration rather than innovation, but A.I. is where we're going to see a huge innovation in how games are made, and probably how they're played as well :)

[@biggnesss7192]

Holy shit this might be the best video I've seen in 2 years. Please upload more documentary type videos.

[@itirix1]

It's actually terrible as far as information value goes. So many inaccuracies and even just plain false statements. Editing and presentation is top notch, but man you can't present a semi-scientific topic if you know nothing about it, it seems.

[@Celis.C]

The moment you posed the question what would be the next big thing, my gut feeling was "creativity". Right after, you started discussing stylism. The true potential will be unlocked through infinite creativity. Isn't that how we got here in the first place?

[@agssilv5919]

Yes but at the same time some of gamings most treasured mechanics , moments etc etc have happened thx to limitations

[@blackdragonstory1122]

I can actually see a vr world that has ai npcs that are different from our world yet feel like they are real. The biggest thing we are missing rn is an actual sao vr tech cuz the current one is just eyes vr. The vid above did hype me up for the potential fun gaming will bring,although I wonder if the gaming industry has the drive to invest in new tech. Since ai became a big thing rn that allows them to make games easier and cheaper would they use that money to invest in the new stuff or just keep the same. Like even if they do bring new tech that prob costs them a lot,will there be enough people willing to pay for it. The biggest obstacle besides not having vr tech is probably internet speeds that arent the same everywhere and would affect gameplay so maybe that part of the world needs an upgrade too. I guess 5G was atempting to do that but its not the same as adsl and actual wired connection. Funny thing about real vr is if you have to be asleep to play with it we could probably also see someone else thoughts and maybe even see their dreams.

[@PrettyGuardian]

​@@agssilv5919 That was the creativity of that time. How can you render something engaging given the set of limitations? As we remove more and more limitations, creators have to push themselves in a completely new way, it's almost like exploring unknown territory.

[@scott2181]

It's not clear what you're saying about using assembly language over other high-level languages, like C. The C compiler - translating C into machine code - would run on another computer, not the console itself. The reason people tended not to use C or other high level languages, was that compilers of the time were rarely tuned, or had no knowledge of optimisation. Code size was larger, emitting more instructions, or compiler not using architecture tricks to squeeze impressive performance out of these low-power machines.

[@peperamonrodriguez5247]

Yes, i found that comment confusing aswell.

[@Dexter01992]

I'm doing a 3D course for real time gaming purposes, and teachers mentioned us a major graphical breakthrough development-wise has happened around 10 years ago with the wide adoption of PBR materials. Before them, every model you'd make would need to have its specified shaders and custom textures for *every* environment you'd put a prop in, requiring tons and tons of work to ensure this random barrel you decided to introduce in a sci-fi neon city doesn't visually stick out of place simply because its environmental shadows don't match the rest. PBR calculates that for you while both simplifying and streamlining the texturing phase of your modelling, making it less technical and more artistic focused in the process. Was very interesting to learn that.

[@bubblesnopemporium5880]

I know that there is a lot of buzz around the AI hype train, but I really could see that being our next breakthrough. We already see a lot of it in terms of ray tracing and stuff, so it might not be as clear cut as the first three examples. But, I think that asset generation, storyboarding, debugging, basic level layouts, etc could drastically decrease the time developers spend on the fundamental building blocks their game. This would be less noticeable on the player's end, however, since it wouldn't necessarily result in a change of the structure the game is built on. It could give developers much more time to focus on the creative aspect of game-development, while the AI helps with the foundation

[@theelectricprince8231]

Don't know if AI will replace all workers but I am certain AI will make work flows frictionless. Less tedium and more focus on the meat

[@tiagotiagot]

Something must've been lost in translation. Doesn't make sense for games to be sold with C source code and compiler in the cartridge itself; they would've been compiled and then the resulting machine code would be put into a cartridge, and only that would be in what's sold to the consumers.

[@jeremycronic]

Hopefully the future of gaming is anything that makes creating games Drastically easier. That way developers can focus much more on actual Gameplay. Unreal Engine 5 seems like a good start.

[@xomvoid_akaluchiru_987]

The thing I love about graphics is that optimization doesn't always mean making the workflow of the program run better but also making quicker algorithms and thinking up knew techniques to make things look better, or reduce the amount of not needed visuals being calculated. Take Culling and LODs for example, UE5's LODs are leaps and bounds better than the manual LODs that we've been making in terms of accuracy and visuals.

[@Tekkas_GG]

Again, this was such a good video! I don't mind you taking some breaks from uploading every once in a while, because these are such a treat when they drop!

[@myoriginalname]

You might want to correct the compiler in cartridge part. No need for compilers to run compiled code. That’s the point of it.

[@batman_141]

Great video! Very informative. Really liked how you explained the history of coding through assembly languages and eventually a uniform language and an engine. Also, on the point of realism vs art direction, I really like the realistic graphics of games like Rdr2 but upon playing elden ring, I was blown away by the art direction of the game, it was like a painting all across. More games should take that approach.

[@sk8erkenny]

i love the editing and the information. visuals were great, script was informative while also not being too serious and boring. 10/10. learned a lot and was entrained the whole time haven’t had a youtube video do that to me in a while ;)

[@sortacoolmaybe]

What a well-written and incredibly engaging video. You did a great job on the defining graphics portion of the video, at least from the average person’s POV.

[@somekidwithanm4]

Absolutely well done sir! From scripting, research, editing, graphics, audio etc. Exceptionally well done and glad I found you!

[@sachamorgese7280]

What a comeback, indeed! Absolutely love the documentary style, you should definitely do more!

[@nessilver]

Really amazing video man. Don't see many people talk about how deep and complicated was the beginning of all of this, video games and computing as a whole

[@RazorFastClaws]

Not saying you're wrong, but imo even in the old times games used recycled routines and at times most of the code with a few changes and optimisations. Assembly was needed because the machines were slow. C compilers were not good in creating optimum code and it resulted in bigger and slower code thus less performance and more cart/disk space. Nothing to do with a needed c compiler on cartridge. The code used is already compiled so a compiler on the cart or console is useless. The machines had different cpu's and custom hardware (like scrolling, sprites, sound, etc.), graphic modes and limits. Graphics needed to be converted (manually) into a different resolution, number of colours, etc. Same for the sound. So ofcourse most if not all of the code needed a rewrite. Cross- compiling or assembling was not a thing yet. The result was that games looked, sounded and played quite different between conversions and sometimes were a completely different game.

[@gipgap4]

Yeah, I’m no games programmer but I think this guy has a few assumptions that are incorrect, mainly because he focuses on the console markets of the USA and Japan. In Europe and 🇬🇧, the home computer was king where after a short while in the 1980’s, programmers were reusing assets and game engines for subsequent titles. Everything had to be in assembly language if your game was to run anywhere up to speed because CPU’s were slow and many computers had no graphics assistance in regards to specialised chips unless you had a Commodore 64. Ultimate play the Game, more famously known as Rare in the USA, would often recycle their graphics and reuse their game engines on the subsequent games. Their “Filmation” games which included Knight Lore, Alien 8 and Pentagram and Rare tweaked their engine to create “Filmation 2” to scroll their isometric graphics in games such as Nightshade and Gunfright. Games developers needed to create their own assets and games engines that they could use and reuse because deadlines were tight so any way developers could save time during development of a game, they would.

[@RazorFastClaws]

@@gipgap4 the Atari 400/800 already had hardware for scrolling and player missiles, but yeah the C64 was a bit more advanced with HW for sprites. I don't know about msx from the top of my head but I'm sure the msx2 had HW for scrolling at least.

[@DalazG]

I'm not sure if there will ever be a point where graphics cannot improve. Right now 4k is a thing. But imagine decades from now. Imagine playing a vr game, you could pick up a game asset magnifying glass and observe things on a molecular level. I don't even think that would be possible in decades but we will slowly start improving beyond our visual limits too

[@CharlesVanNoland]

The key is machine learning filling in all the details for artists, so they can paint worlds as coarsely and abstractly or as specifically and detailed as they want. Once machine learning becomes a streamlined part of the game design and art asset workflow, it's game over, assuming the graphics hardware has become capable of rendering photoreal simulations by that point. Also, assembly language wasn't a big deal back in the day because games and their assets were very simple compared to games even from the 90s. It took less time to write Pong in assembly than it did to write Doom in C.

[@theslay66]

I put my money on AI too. We are far from realising the full potential that this technology can have for games. Not only from a developement standpoint, where the devs will only need to define rough guides and let the AI hallucinate the scenery and the events, but you could have it react and adapt to the user as well, learning from his preferences and habits to generate an unique experience. Ever seen this AI that generates photorealistic sceneries from simple color blobs ? That's just the technology in its infancy.

[@alessandrobaggi6129]

Dire situation when an ai will be able to decide what we play/watch/listen to... not to talk about lack of originality and coherent vision in any project...

[@theslay66]

@@alessandrobaggi6129 Well it's already the case... it's called KZbin algorithm.

[@alessandrobaggi6129]

@@theslay66 ...and that's exactly why i said the situation will be dire, as we already can see some major errors in that.

[@ghoulchan7525]

​@@theslay66 problem. AI stuff can't be copyrighted thanks to a case where a monkey decided to take a selfie.

[@GANONdork123]

We're reaching the point of diminishing returns when it comes to realistic graphics. As graphical technology advances, the details we're missing become smaller and less noticeable. Think about a painting a picture. At first the changes from each step to the next are massive, but as you get closer to the final painting become smaller and more intricate, to the point where many might not even notice the difference. When it comes to graphics tech, there's also not much we can really do to push things forward either. The last major step in computer graphics is path tracing. This is something film and animation has had access to for years, but up until recently has been too computationally heavy to do in real time to any degree. Once real-time full path tracing becomes a reality, there really isn't anywhere else to go. All that we see in our real world is light bouncing off of, refracting through, and scattering through various materials. Once we can accurately simulate light and surface interactions in real-time, what can we possibly do to improve upon that? The answer is, nothing really. Sure we can pump more polygons into the models, and make 8k or even 16k textures a thing, but at some point, nobody will be able to see the difference. You don't need to simulate every grain of sand to make a convincing beach.

[@jaefrmbk2k]

Great subject/video. I think the next great innovations will come when video game development becomes more democratized with much more affordable & easier to use tools where more independent visionaries & artists can create comparable content visually & mechanically to the Triple A titles. that's where Style & more fresh Concepts will really shine 💫

[@ingwerschwensen8115]

What's wrong with YT's algorithm? Why did it take years upon years before it suggested one of your vids to me? Anyhow, great video! Thanks! I immediately subscribed.

[@dorktales254]

Wait, the compiler doesn't go on the cartridge

[@sainuudotcom]

I think the breakthrough has already occurred and it’s in the PS5. It’s all about Memory management and Machine Learning! The title that really going to showcase the capabilities is going to be Spider Man 2. I have heard that industry people including Marvel producers are just amazed how Spider Man 2 is looking and running.

[@chriscollins3225]

That and Final Fantasy 16

[@bob_0146]

GoW Ragnorok could've looked like an improvement if it was actually next gen

[@techknow9237]

With all the graphical power available today - the problem is that it all looks great but physics and a sense of interaction with the graphical 3D world is missing in many games. Skyrim still stands out as the world is interactive on many layers.

[@b747xx]

They did use assembler, not because they needed to stuff the compiler into cartridge (if they wanted to use C) But more because compiled C code where really not as optimized as hand made assembly code. You compile the C code on you're workstation (for the chosen target), that compiler then generate the machine code. You flash that later on the cartridge. But then that compiled code would take twice the space and run way slower than the human made/optimized assembler code. Fast CPU allow use to add a lot more inefficiency in the final product. That's the difference. Allowing people to concentrate on the content instead of the code. Note that while doom was made in C, part of it was also hand optimized in assembler. Edit : I should have read the OP comment, sorry for that!

[@KINGJADEX]

This was fantastic! I would be interested in delving more into what the "Age of AI" would entail. Very interesting stuff!

[@bamula2]

Currently reading The History of Video Games Volume 2 and this right up my alley with technical reasoning on topics. Great video.

[@ApfelSgt]

Why the hell don’t you have a million subs??!! 😮 Your quality of your videos is so damn high and you out so much work into it! Less then 50k is a shame!!! Thanks for your work, love your videos!

[@Rain1dog]

Hey, I just wanted to say I’ve come to really enjoy your videos. They are very entertaining, informative, and thought provoking. Thank you for your time and effort putting these videos together!

[@jjdizz1l]

This is the most in-depth QUALITY documentary I have had the pleasure to enjoy in a long time. I know perfection is an overused term, but this is PEAK tech content here; not too much for enthused minds to grasp and great detail. I love the direction you're moving in. Keep at it.

[@aliferestored]

So glad you randomly came on my feed, what a great video. I rarely comment on KZbin videos, but wanted to let you know this was so good. Thank you!

[@RubbishRebel]

Amazing video, my G you are criminally underrated. Very informative and entertaining video. I really liked the history part with assembly coding, mario and doom. Would like to see more documentaries.

[@HeirofDacia]

I really hope people stop salivating over hyper realistic games, when some of the best games I've played are super stylish (Guacamelee, Hollow Knight, Papers, please!, Psychonauts 1&2, Enter the Gungeon, Loop Hero, Cuphead.), and I'd play them over any of the AAA games that are touted as having the best graphics in the world or whatever.

[@ΑνδρέαςΠ-ι9ε]

That was a masterpiece! I do not see easily long videos due to time reasons, but you hooked me! Excellent executed!

[@Cpgeekorg]

This was a really well put together piece that does a great job of explaining a good overview of some of the iterative problems that we've had in the past, where we've been, where we're at, and at least some of the elements of where we're going. I completely agree with your points that at the end of the day, computer graphics in video games being good, is very specifically a function of allowing artists who have excellent ideas express those ideas as easily and as quickly and as portably as possible... if you walk up to a person and say "tell me your vision for an excellent video game" (particularly a graphical and or narrative artist), they're likely to draw you a picture of a style and/or describe an overview or a scene in the style that they find interesting enough to explore and iterate on, if we can make the tools as ergonomic as possible to tell the computer what we want it to do so it looks like their vision, we've gotten much of the way there (that's kind of the endgame for modern game engines). This video has some great remarks about stylism and how that can be used to improve games and the storytelling within them, and I completely agree, but there are stories and experiences that *could* be best served by displaying them in real-time photo realistic fashion with realistic physics and systems simulation to the point where you can simulate a whole world (or perhaps someday galaxies) inside of a computer system at a subatomic level with extreme fidelity)... this would be great for scientific computing, but maybe not the best for gaming, HOWEVER, for stuff like virtual reality (and that hardware is getting better all the time), or volumetric holography systems that are being developed now, I think it could be fantastic to have a spy thriller or a drama or a fantasy rpg or an action-movie style game that looks and feels completely real until you take off a headset or utter the command to exit the simulation or whatever. I think the greatest hurdle to overcome on our way toward photorealistic systems simulation (at least getting us as far as city-sized fully realized photorealistic environments) has to do with some of the same elements that we've always dealt with - the reason we didn't use C back in the 8 bit era... because compiled code isn't NEARLY as optimized as optimized assembly language code (or even hand-coding machine language), where every single clock cycle, memory read or write, every packet of information sent or received to and from the dedicated accelerators and what goes on there is manually and particularly managed in such a way to do exactly and only what it is that you are trying to accomplish at that very moment. Right now, computers are running on an incredible number of layers of abstraction. at it's base, you've got machine language, then on top of that you've got code that is almost always manually written (at least at first) in assembly language as local routines for whatever high level compiler you're using so when a programmer calls a high level function, they are basically calling from a library of stored high language code which calls from one or more libraries of high but slightly lower level same-language code that is ultimately compiled, assembled, linked, and out pops a binary output. there are HUUUGE optimizations that can be made to the way that these high level languages store and retrieve data, where that data should reside at run time (on the ssd? in vram? in your cpu cache? in the memory of another dedicated accelerator?) and specifically what cpu/gpu core executes what code at what time and how that's orchestrated in the system. a programmer says "i want this graphics model loaded at these coordinents rendered with these properties" NOT "in what order where and when should i be storing these bytes of code or data so they can be optimized to run as best as possible... nobody optimizes that way anymore because it takes too much time that could be better used making games, and it's ok, because for what they're trying to do they can just trust that the next gpu that comes out that takes 600w to run and doubles as a space heater will be able to handle the lack of optimization by delivering more performance because it's really expensive to pay programmers to do that... AND it's really hard for anyone person to know the ins and outs of a very complex modern computer system... machine learning has entered the chat... we are at the infancy of the revolution that is machine learning... this is often referred to by the media and the companies who create accelerators for it as artificial intelligence... they AREN'T the same thing. machine learning, the stuff that's gracing us so far with things like beating chess grand masters and deep fakes, and inserting intelligently interpolated frames to improve frame rates for things like DLSS and FSR aren't ready to form skynet yet... rather, they train and run neural network models (usually convolutionally by giving them information about how to test if the output is what we want or not, feeding them with sample data, and having them score themselves on their rubric (often repeating the cycle millions or billions of times) until the output of sample data matches your expectations... it's a form of teaching a computer how to teach itself to automate it's learning process and repeat until it gets gud. I expect as we get better, more performant dedicated matrix multiply and accumulate accelerator hardware (commonly referred to as ai accelerators, neural engines, or tensor cores depending on which vendor you're talking about), that we will be able to have it go through our mountains of portable high language code and rewrite SIGNIFICANTLY more optimized versions of the underlaying code libraries and their associated assembly language routines that they link to (which can be thought of as being similar to how hardware drivers work as an intermediary between your software and hardware in the programming world) and within several years optimize every last bit of memory and code for any particular computer architecture. you could train it on the existing libraries and assembly code with "this input high level function is supposed to perform this high level output function, here's the old code, now run it billions of different times in billions of different ways with this sample hardware spec and make it the best it can be. you can do this same thing at all abstraction layers up to and including hardware development. you could say "here's all the x86 processors, here's the list of erata, what's the most optimized way to lay these chips out to reduce latency, cost, and improve performance with this big library of software and simply have it iterate in an automated fashion until it finds a bunch of good ways, send it to fabrication, manually validate it, and ship it... being able to automate the INTELEGENT brute-forcing of optimization at many different levels of computer abstraction is THE way forward. once we remove all the bottlenecks in our current systems and make things as good as they can get, we can then get to work on what optimized future systems might look like, have machine learning systems ideate and optimize on supercomputers in simulators distributed across all the computers in an organization... there's a lot of bloat in modern systems because they were coded by hand and were optimized to be "good enough" up and down the software stack from bios to operating system kernals and libraries to software libraries to drivers to game engine libraries, to games themselves... removing those bottlenecks, I suspect we'll see some DRAMATIC uplift in performance even when running the same software on previous generation hardware and huge performance gains once we have ML models building cpus, gpus, and other optimized accelerators for us... and that's just scratching the surface. imagine game design as similar to how people on star trek interact with holodecks... ask the computer for objects, physics systems, describe with words or drawing in space for the things you want, the style you want them, how you intend for them to work and they just appear in vr either ready-to-go from a library of things, or generated from machine learning models (kind of like how ai image or music generators work now), create people with simulated personalities that are fully fleshed out, explain their backstory and their place, and have the digital actors go about their business while other machine learning models optimize the underlaying hardware and software to make it happen at the highest possible performance on the available hardware... couple that with the ever shrinking latency, increasing speeds of modern computer inter-connectedness, and we can all share cpu power with everyone else in a huge decentralized computation and acceleration grid as a universal human resource. add that to robotics automations and we could be looking at a utopia... then again, enough automation and somebody is going to want to direct a machine learning model to create an actual artificial general intelligence which through the automated iteration and access to pretty much every data system will quickly become an artificial super intelligence very quickly... and then i suppose we'll be judged for who we are and see what happens, but I still think that's at least 10 years off.

[@pitchblack5422]

The Documentary style is phenomenal I am looking forward to seeing more of these types of videos

[@gumbilicious1]

I am a fairly old-school programmer with a CS degree and I have no idea why you'd have to put a compiler on a cartridge in order to run C code. nothing else has to have The C Compiler to run C code, it's the whole point of having the compiler: to make a binary executable separate from the text code it was compiled from. You make a.out on a computer, a.out will run on any hardware platform the compiler was made for with no need for any future compiling. now this is different if you want to run a.out on a different hardware platform than the compiler was made for, you'd have to recompile the code to run on a different machine but as long as the hardware is the same then a.out will run. consoles would be an ideal way of controlling the hardware platform, so once again, no idea why you'd need to include a compiler on the cartridge. now, there is a different issue using code that was compiled, and that is bloated machine code. if you are a very *good* assembly coder you can design code that is far smaller than compiled C code, and this would allow you to write efficient code that can fit on tiny ROM carts. this was my understanding of why old games were written in assembly, not because you would need a C compiler on the cart

[@sirhambino596]

I usually don’t comment much. But I have to say I couldn’t not congratulate you. This video was so freaking good!!! I loved it!!!!! Keep ‘em coming!!!!!

[@ea_gaming]

Loved this video! This was excellent- love the shorter content, but the longform was also excellent, especially for a video of this nature. I hope the time between videos moving forward isn't as long as it was to this one though!

[@nathanvarnell5505]

Crazy skill in making this video, man. You deserve a hell of a lot more recognition

[@aldemar.bernal]

The C compiler taking cartridge space? I think there is a mistake there because you actually compile the code in a computer and the only thing that you put in the cartridge is the binary compiled code, not the compiler nor the code in C, If you'd put your C code in the cartridge you'd actually give for free the game source code. So it happens today, no matter what language you use what you do is to compile your code and what you distribute is the binary not your code.

[@beatenplastic]

Can I just say, this was absolutely superb. I watch a lot of videos like this about subjects other than video games and this was truly top shelf. I hope you go far.

[@phant0mdummy]

I wish more games would chase art styles like Hi-Fi Rush. Not to say every game should appear as a cartoon or comic book. If everything is Jet Set Radio, then nothing is Jet Set Radio; but the idea I'm getting at is to have fun again. Make games that will not age with the advancement of graphics. Remember when Gears of War 1 had the best graphics you had ever seen? Hi-Fi Rush will still look incredible in 30 years because it is incredible. It's art in motion. More games should focus on delivering art direction over raw graphical fidelity. If they stop wasting all their energy on the looks of the games, maybe someone will finally innovate on some interesting gameplay again.

[@Drstrange3000]

I agree. Especially with the art is in motion. It is why I still like Nintendo games. Splatoon just looks and feels so great to play. I love the style of it so much and Hi-Fi Rush was my dream game came to life! Both these have amazing animations and are just so lovely and vivid. Fun is rooted in their DNA.

[@HighTechWizard]

Amazing video! Thank you tenfold for not holding back from the technical side of things, I come here for your approachable, but well informed computer science. Incredibly polished documentary, and I love how creative you get connecting what you say to the almost subliminal messaging of what games you're showing visually. Parts that were very much your own, like the visual timeline transitions of innovation cycles and the various montages of gaming and music. I expect to see these video essays in planetarium theaters in the future!

[@adamgr6988]

A rare time where the algorithm just hooks me up. I had never heard of your channel before, but this video was just immaculate work, top to bottom. Easy subscription

[@KainDizengarth]

I love this kind of video. I love learning stuff like this. Thx for the great content ♥

[@greatvigor]

Outstanding video! Well paced, edited, and animated. The algorithm finally put something I'd want to watch in my feed.

[@mistadopeman]

In ten years. Graphics are going to reach a point where frames per second will not even be a conversation. Believe it. That’s when we won’t care about graphics anymore. When frames per second can be nearly Infinite for literally any game.

[@shadeaquaticbreeder2914]

In our society the biggest hindrance to progression/technological innovation is money. Why would they make a car that can fly and run off water or something as readily available when they can make a POS car that you have to fill up at least every week for more money and it inevitably falls apart and you have to fix it.

[@jackyoh971]

You really underestimate AI. one AI can replace graphic designer another one for music etc. 1 person can make way more thing done more quickly with AI than a team before. AI made for being NPC will change the industry. There is only thing that already exist and can have a lot of potential is that.

[@WalnutOW]

12:20 This isn’t true. The compiler doesn’t run on the target system (e.g. NES), it runs on the developer’s PC or devkit. The reason why NES games weren’t written in compiled languages like C is because compilers (at least back then) produce longer, less optimized assembly code compared to what an experienced programmer can do by hand. On systems with limited resources like the NES, the extra performance that could be gained by writing directly in assembly was crucial. Nowadays, it’s negligible, so most game developers never have to even look at assembly.

[@FaisalMahmood91]

I was looking through the comments to see if someone can explain the real reason, because I knew that part was inaccurate. I knew that you'd compile the code before putting it on the cartridge, but didn't think of the optimization problem.

[@thyron]

This video is very, very good. The production quality and the script are very good made, easy to follow, entertaining, visually appealing. Wow!

[@gatsbysgarage8389]

Honestly battlefront 2015 is still one of the best looking games of all time… and crysis 3 still has pretty incredible graphics. I think some games (like cyberpunk) still do push the graphics envelope to their limit even raytraced games in general have trouble standing out from a game like battlefront

[@jerrys.9895]

Editing, composition, and presentation are all so friggin good! You're criminally undersubscribed, friend. Doing my part.

[@anmolagrawal5358]

Awesome video, I like how small independent content creators on KZbin are seemingly upping their game more recently producing more in-dept coverage of specific topics. That or just the algo getting better at rewarding those. Leader to the former outcome. As for the potentially breakthrough future implementation, how the heck did you miss out on mentioning BCIs? Gabe talked about it recently, and they may be able to produce a virtual rendition more colorful than reality itself.

[@dulappen494]

If you draw a circle with 8 corners and improve by 4 corners, you made a 50% increase. If you improve your 12 corners by 5 you now have 17. But that is only a 41% increase despite the fact that you made a bigger improvement. Graphics improvements seem to slow down because improvements described as percentages of the total fidelity decrease even if the actual graphical jumps get bigger from year to year. (This is oversimplified, but it get's the point across.)

[@p0t4toePotato]

uhh why would they put the c compiler on a cartrige? its not an interpreted language, even assembly has to be translated to machine bye code so this means they would have to develop on different machines, convert the source assembly into the machine code and write the machine code to the cartridge memory.... so why not add another step before that? barebone C implementation -> compiled to assembly -> convert to binary machine format-> burn to cartrige -> let the cpu feed itself on the cartrige instructions and data

[@KentonBenfield]

I'm wondering the same thing; either I don't know something unique and important here, or this was a miss on the research/understanding of the reasons for not using C back then.

[@romanograsnick]

My assumption for the next stage is the AI controled game engine which can make a game by just typing in some text and checking some boxes. I do hope traditional, hand crafted games by artists like me will stand out among these. Thanks for the vid!

[@DrewAk49]

Awesome video! Great editing and info! The ray tracing on the marbles looks amazing!

[@ungratefulmango]

This video was so rad. I learned a lot and you really got me thinking. I like this a lot man. So well thought out and well made

[@abrahamj15]

Even tho I hate how Nvidia has promoted it to push high prices, that innovation is raytracing and AI, we have reached a limit in how many polygons and megabytes of texture you need to make something realistic, we just need better lighting and take off work of the artist when it comes to giving detail to the environment which AI has proven amazing at it

[@Drstrange3000]

It is amazing how much great lighting and animations can make something appear realistic without looking photorealistic. Even watching the Replaced trailer it looks so good while still having pixel art. The animations and lighting are top notch.

[@ori5278]

This is the type of video that would hit 1M views. Great vid innocenceii

[@omritahar9048]

My dear God... WHAT A VIDEO!!! The quality of this is UNBELIEVABLE, man. Fascinating subject (please do not spare the tech explanations), with an absolute knockout of execution. I am blown away. I really do. 👏

[@FrankenSama]

This was a great video. The thumbnail fooled into thinking it would just be another rant about how gaming is "dying", but I was so engaged from beginning to end that time just flew by. Great work 👍🏾

[@jasonhurdlow6607]

Video wasn't very technical, but it did hit some good points. As someone who did game development in the '90s I'd say the biggest advance has been hardware acceleration (aka GPUs), before them we had to do everything on the CPU, and it was very limited. Today's CPUs and GPUs are orders of magnitude faster than what we had then. It's simply stunning!

[@ndacostabest]

How have you only got 39K subs man! Seriously uinderrated channel. Keep doing you!

[@Matty002]

the photorealism graphics chase got old immediately after it started, especially when all the military shooters and franchise games started looking the same nobody will ever tire of games looking unique, or at least not deathly generic

[@burningflag3679]

Thank you, very well done. I call myself The Scrubbiest Dev. To remind myself there is always more to learn. As such i spent 2022 consuming a lot of gaming related content on KZbin. There are so many misunderstandings about what goes into game dev. Which leads to unnecessary anger in the community. I've been researching & scripting videos for months, in an effort to take people step by step as i develop a typical 3rd person action game. But videos like this are just as important. Gaming really needs more channels & videos like yours. have a good on & take care.

[@matsnilson7727]

I'm not going to pretend that graphics haven't improved quite a bit in 20 years, but for me personally, the last game that truly made me pick up my jaw from the floor was Gran Turismo 3 on the PS2. I've obviously played lots of graphically impressive games since then, but I've become so incredibly used to good looking games now. Most of us have. There's nothing really special about it any more - at least not in the same way it used to be. This probably sounds negative. Not meant to be. Games generally look fantastic today. I just think it's become a lot harder to be blown away by relatively small visual improvements instead of the bigger leaps we used to get.

[@ProfessorPowGames]

I love your final take on where the industry is headed, I believe that for a while, the "games as a product" mindset has encompassed most of the discourse on gaming and where it's headed, but video games are just as much an art form than they are a product. So personally I think we're heading towards a video game renaissance, where not only game developers but the audience at large appreciate games more as art than a simple commodity to be bought and sold. We can see this happening already in the indie games space where the games that become the most successful are the ones that push a unique concept or message, juxtapose this with the current triple A game climate where more and more large studios seem hellbent on consolidating every release into a gray amorphous blob that is nearly indistinguishable from the gray amorphous blob that was released the year before *ahem* ie, studies particularly like ubisoft. Like you said, videos games as a medium are still in their infancy, so it's almost impossible to be sure how the medium will mature, I just take comfort in the fact that despite where larger industry trends are going, thanks to the ability of game engines to ease the process of making games for passionate creators, we (almost certainly won't truly run out of unique perspectives or experiences.

[@PrincessNine]

I've been saying 1 thing since they announced the ps5. The true test of next gen/the next evolution is physics simulation. We've basically abandoned clothes and hair simulation and it's honestly ridiculous. It's getting weird to see photo real people who's hair stays in place and clothes are pre-wrinkled.(death stranding) I'm tired of seeing hair and jackets not moving.

[@randallrush4962]

The use of solid state drives would be a shining example of a hardware breakthrough. It literally changes the way game developers approach game design in multiple ways. The use of hallways/narrow and blind pathways to hide loading times, the need to double up on files for access speeds, and the limitations of world design directly because of memory speeds, are only a couple of those points. Ray tracing I would only consider a natural progression or iteration especially considering that the style of ray tracing used in real time is still pretty low quality even in comparison to cinematic works years old. Keep in mind that this generation has just begun and the best looking games of any hardware generation appear at the end.

[@matthewlopez8987]

I am a massive fan of informative videos like that and I really appreciate the effort you put into gathering the information presented here. I know they take time to make, but I hope to see more of this

[@MisterSynonym]

I guess progress can come in different forms other than visuals. For example: An Ant-Man videogame could do some really interesting stuff with asset scale. We could also see more detail in environment/asset destruction. Or, modifying assets as a gameplay mechanic e.g. turning water into stone etc. Who knows, when I'm an old man, I might see the first ever Quantum Console! Nice to see you again Innocenceii

[@andrecostafranca]

I was impressed that, throughout the entire video, you haven't touched a single time the VR topic. I know VR is not exactly a new type of graphics, but it is an important new factor on the board when we think about the way people will be playing games in the future. I genuinely think that in 2 or 3 generations, VR will be the staple of gaming, because what does every person that loves videogames wants to do? At least in my perception, it is to be part of that universe, and I think VR is the next step in achieving that, not the photo realism or the style in itself.

[@henryhere]

DANG this video is well made! Excellent job man! I never know what to do when editing, but every second of your video keeps things interesting!

[@Altalus]

That video was the best you made so far. Love the style.

[@genericgorilla]

I dunno, I think the eras are defined a bit too arbitrary, maybe it's the old owl of Minerva stuff, but I'd like to think we aren't living in the "age of the engine", because if we boil it down, an engine is simply an abstraction of the graphics pipeline. The real innovative change and the era which came to define the last 20 years or so was, actually, programmable pixel shaders. And the immediate future points out towards both ray tracing but also increasingly more capable AI upscaling algorithms which bypass hardware limitations by simply lowering the target resolution, like magic. But also yeah, graphics improve asymptomatically, there's diminishing returns to graphic fidelity with every new breakthrough, and I'd like to also step forward the fact that graphics are not only limited by hardware, but also limited by budget, that is, the labortime vs expected returns or how much work you can dump into a game before it stops being profitable. Which in the end points to AI art generation even more

[@saltycynic5770]

Since yous asked. I've not seen any of your other videos, but I subscribed. Informative, well produced, I enjoyed the length/format.