Hey warning, your thumbnail is completely off base. Not even by a little bit. The number "86,163km" genuinely means nothing as the tears of the kingdom map doesn't reach that distance in any possible direction. You might have meant kilometers squared "km^2" as through a google search of "tears of the kingdom map size" that is the first result that shows up.
The tree system is more than just a billboard, they are called imposters. They take into account which angle you are looking at the tree from, and display a png of an angle that is as similar to the viewing angle as possible. So this way you can look at the billboard from even above and it will display the top of the tree, not the weird 2D paper that we come to expect from grass billboards.
Yes thank you!! I don't understand why so many content creators talk about things they haven't researched, like don't talk about things you don't know!
Megatexture is not about texture LoD, not directly at least. It's about avoiding texture tiling, about having a unique texture for the whole map, instead of repeating the same few textures over and over. The optimization of the memory usage is a side effect of the implementation ("virtual texture"). Megatexture pretty much requires virtual texturing to be viable, but the latter can exist without the former.
how is this clickbait? people really taking issue with informative videos now, then proceed to eat up the mindless clickbait everywhere else on the site.
Hardware limitations. It's the same reason you can't load an entire 150G video game into your 32G of memory all at once. How are you watching this and not understanding that already?
it does actually mention it, impostors are basically billboards at it's core. As of lod he doesn't mention HLods and other type of tech around LODs, it's kinda unnecessary in a general explanation video.
When the eventual GDC talk comes out about TOTK I would accept a video about its somehow accurate-feeling complex physics simulation + building mechanics titled "this shouldn't be possible" as I genuinely don't know how they managed that on a mobile CPU. A barebones overview of LODs, with nothing specific to implementation on the Switch itself, doesn't warrant this buy-in
@@johnnydarling8021 That’s already been covered on the channel with this and other games. The “shouldn’t be” isn’t explained or contextualized in how TOTK uses it. Some examples I know from playing the game myself: everything underneath sky islands use low LoD, or how the chasms into the depth are used to allow the game load in the map, or how green space of recall creates a sort of wormhole to load in the next area. It’s easy to recognize when a video is simply filler to make sure the channel doesn’t get dumped by the RU-vid algorithm. Ironically enough, I comment to support engagement
Bro this thumbnail is so pointless, like what do you mean 86,163 km the game's height limit is 3292 meters. Why do you make videos like this, you didn't even say why this "shouldn't be possible"
Quick note, billboards usually rotate in a way to face the player to avoid being able to see the plane from a different angle. This can sometimes cause silly artifacts such as in genshin where if you fly over a tree while it's a billboard it will bend over to face you in the sky or will do crazy rotations.
0:26 What kind of freak would take an executable file, a file meant for _programs_ and is spefiic to Windows and Windows alone.... And then use it as a model which is supposed to e part of one program?????? Atop that, it also has heaps of extra power and processes being used up when you could just use any normal format like a *normal person!*
UE5 is slowly negating the need for LODs whoch I find to be a HUGE time saver on production. With 5.3 now landscapes can utilize this as well which will hopefully work well with randomly generated assets.
UE5 isn't getting rid of LODs; it's just abstracting them so the user doesn't need to manually handle it anymore. Nanite, under-simplified, is an automatic LOD manager.
@@yesno1085 That is true, but it also allows your game to scale much better. Traditional techniques keep up well enough with Nanite until we scale up the distance and number of assets.
@@DaSquyd Yeah, but Nanite also handles super-high-poly meshes much better than traditional LODs. If you put 50 million-poly spheres in a level without Nanite, your PC will crash, while Nanite doesn't even break a sweat.
Does anyone else find 2:36 incredibly nauseating? the way the background slowly scrolls along with the main element on the screen shaking is super disorienting for me.
Hardware limitations. You don't have infinite memory and infinite compute speed. It's the same reason it isn't possible for a 150 gig game to just load into your RAM all at once.
@@sventomasek LODs make possible the thing that shouldn't be possible, which is to have a massive world all visible at once or bunches of AIs all active at once. (I admit it never occurred to me that the other things he mentions would have LODs too.) Just like saying "running a 100-gig program on a computer with 32-gig of RAM shouldn't be possible." But we invented demand paging, so now it's possible.
@@darrennew8211 With Nanite you can have a million million-poly assets in a level and it won't destroy performance. Also comes with zero pop-in. Clearly something isn't being done right with traditional LODs.
@@darrennew8211 Not really, Nanite doesn't use LOD at all. It's completly different system. One of the advantages, among others, of this system is that it does not have to load complete object models but only parts of them. Thus if you see for example 1px of a rock you don't have to load the whole rock as with the LOD system
@@Lamas236 That's cool. I didn't know it was even more sophisticated. I figured it was just calculating LOD on demand, but I guess eventually it gets different enough you can no longer call it LOD. And it makes sense they'd do that, given how it seems to be commonly used, with lots of overlapping geometry.
you would also tell about chunks! Separe a chunk of the world on a single and big draw call and bla bla, and put smal gemometry details on another draw calls to have more control or about the use of draw call in away to do better oclusion of polygons to not redenrig off screen stuff, like rooms inside a bulding a separete draw calls so you do not renderer all the house at same time and this way you can go wild on polygons in a single room, But more draw calls more sad the cpu will be so must be a balance. Or about how to organize the textures, the altas the materia from each level to have reduce streaming, well todays we have sdd so there is no need to care about it anymore.
So 3:40 minutes of still not even remotely answering the "this shouldn't be possible" and the point of the video nowhere to be seen. Yeah we all know about LoD but we haven't seen anything about the point of this video.
people really are missing the point of the video? "this shouldn't be possible" is clearly referring to a fully detailed rendered world, and he then goes on to explain that LODs are used to make it possible. It's a little clickbaity i'll agree but idk how there's loads of commenting complaining about "the video doesn't explain why it's not possible" ? ... it literally starts off with that..
The title is misleading, making people think this video is about some revolutionary tech when in reality it's just something very simple which we've been doing for many years. I would also argue that it isn't possible to render a fully detailed world, since using LODs makes it a very undetailed world. The thumbnail is also misleading, as TOTK is not 86,163 kilometers in height. His intention was clearly to clickbait, because most people already know about LODs and wouldn't have clicked on the video if they knew it's just LODs.
@@AliasA1 Where did I say the video was wrong? I said the title and thumbnail are misleading and clickbait, which they clearly are. If the title was something like "How games render such big worlds with ease", more like his previous videos it would have been fine and explained what the video is about, but this title doesn't explain anything and is purely used to farm clicks, same with the thumbnail with misinformation. Nowhere in the video did he say that it shouldn't be possible to render all of this at once, which makes it sound like the title is referring to LODs (the only thing he talks about in the video), but using LODs should and definitely is 100% possible. If you think the title and thumbnail are not clickbait, sorry but you're coping hard.
UE5 basicly have a system that create LODs itself based on the camera view, angles of the polygons and the pixels of the screen. It's as unnotisable as it can get and you don't have to create LODs yourself.
Offtopic but a cherry on top of this video's topic is UE5's Nanite Foliage which its polygonal clusters smoothly scale with distance but has better topology (in comparison to what LODs do), making even high-detailed trees and grasses million miles away still looks the same plus with better game performance to render real-time. Even low-polys and/or bad graphics also benefited from Nanite Foliage too for even smoother performance. Note: Search "topology" because it's far different from meshes despite its similarities/synonyms
Damn let the man be. If the title says "this shouldn't be possible" and he proceeds to explain how it's possible, it makes sense to think instead "your computer can't do this on its own, so here are the cool techniques people use to give the illusion that it is possible" Like even if you want to argue it's click-bait, which it's not, it's literally in-line with the concept of the video. He's explaining how the impossible was made to seem possible. Like other people pointed out, he forgot to mention imposters, and that's a valid critique, but the rest of the video makes sense you're just not giving it any thought.
There's not really any good alternative, the only reason why games can have so many high-poly assets in them is because it the engine never has to load all of them at once. Swapping assets to a lower LOD is barely an inconvenience to any modern engine.
Super cool! I didn't realize that statemachines could inflict so much issues as well. I guess I could change the amount of raycasts if an enemy is nearer or father away. I wonder having 2 statemachines would benefit. 1 being movement/search type and 2 for computation types like amount of raycasts. Though you wouldn't want a spike of computations on the other side of the map because a few NPCs found each other.. Like if two NPC's are fighting in an arena sort of scenario. Otherwise it's probably not loaded in if it isn't an arena type game.
Hey @StylizedStation, I was just wondering if you could make a short video on an animation that I can't quite figure out, I was wondering how I could create an animation that adds objects that were not there before, E.g spawning something in at anytime and anyplace for a certain animation. Either in blender or UE5 would be great. This would really help and would be appreciated very much. Thankyou
Often for environments specifically tiling textures are used because with a small texture you can cover large amounts of space. Obviously tiling isn't great and sometimes it's really noticeable. Megatextures are basically a texture atlas at it's core but it's also a giant texture (that's why they are called that) with zero tiling
@@FesterThanLight i cant speak on behalf of everyone but i havent learned anything new from his past couple videos. was wondering if anyone else felt the same
Tom, I have a Question for you, And maybe you answer it in The Lazy Handpainter's Guide, But How do I get Spiderverse Style Textures for Substance, your voice doesn't annoy me so this is why I'm asking you.