They're achieving nothing, really. The ability to do what AI does for them—copy/paste/adjust—has always existed. The Internet is full of samples, engines other than Three.js, and Three.js content. It’s nothing new. Most are just grifters or having fun with a new tool. Even technically, it is much harder to decide, design, and implement a material system and pipeline, or an asset system in general, than to make a basic renderer. FFT waves, physical atmosphere, foliage, and terrain have existed for a while, and a student can pull them from CryEngine or Unreal. There are thousands of engine and game demo projects all over. Engines are not about your terrain showing up; they’re about working for artists and designers and deploying to all platforms. Games are not only about loading a level. They’re about smoothness, gameplay, story, and feel. You can’t vibe-code a game any more than you can vibe-code a proper engine. The moment you try something off the rails, these models fall apart. Fable or Opus output code they were trained on, not what an intern would write after seeing the existing codebase. In my own tests, my engine uses an unorthodox, cutting-edge approach to rendering with render graphs, bindless resources, vertex pulling, and mesh shaders. Yet the models (including almighty Fable) poisoned my codebase with average Internet code, ignoring both the existing code and guidelines given right before the “go” command. Like all rapid copy/paste/modify tools, AI is fantastic. It sped up my work by tens of times in some areas. But no, they can’t make quality stuff they haven’t seen. Unless a carbon copy of a game from some repo is desired, they’re not going to work.

Looking it rendering stupidly sized assets on the web, first up is the Intel "Jungle" scene intel.com/content/www/us/en/… Progress so far... - Gpu driven rendering with meshlets, occlusion - Nanite style LOD - bindless materials - streaming textures and mesh - sprite impostors (wip)

Yes mostly the same features as Mac OS now, raytracing is supported too and that wouldn’t really work without bindless.

Oh so iphone can do bindless? Nice, did not know that.

It's my own custom pretty naive handwritten (no AI) not-bindless OpenGL backend that's probably a subpar solution in the long run because like even shader stuff is all handled at compile-time and that sucks but hey it works for me for now so why not

Idk about about iphone, so I'll just assume it can't do bindless, and if that's so, how do you resolve a bindless vs non-bindless development pipeline? Do you manually make two alternate sets of shaders and completely different render graphs to facilitate this change?

Is this with bindless resources?

Not yet, haven’t ported over graphics pipelines yet. But the RHI will only support vertex pulling, as part of its bindless nature. Vertex buffers (although I only use meshlets), are a good use of persistent GPU-only memory, which would still use copies. Textures are a handle

Yeah nice, I also went through all this multiple times and this time I wanted to try to forget most and see where it can go. No BSP or Octree but instead GPU culling using compute and I render everything with bindless materials and a single uber shader. Gemini did the concept

The last game with really incredible texture work I remember is onrush and they used bindless graphics to take the load off the cpu and use it for more interesting things. Naturally the game failed because it was actually good.

I’M JUST BEING DRAMATIC IN FACT IM ONLY AT IT AGAIN AS AN ADDICT WITH A PEN WHOS ADDICTED TO THE WIND AS IT BLOWS ME BACK AND FORTH BINDLESS SPINELESS AND PRETEND OF COURSE ILL BE HERE AGAIN SEE YOU TOMORROW BUT ITS THE END OF MY WAYS END OF MY DAYS AS A WALKING DENIAL MY TRIAL W

Yeah. Definitely no pointers. But bindless allows indexing into a buffer descriptor, so that can be used to sidestep the lack of 64-bit GPU pointers.

WebGPU is limited by Vulkan 1.0, Metal 1.0 and DX12.0 feature set. No bindless at all.

Bindless, spineless, and pretend of course I'll be here again, see you tomorrow but it's the end of today

I implemented fully async world loading (not just off the main thread), so the renderer can show the world as fast as possible while other threads update acceleration structures, bindless arrays, and so on. I'm making a game, so iteration speed is critical.

あとはReSTIRやってる時にこういうことができる…とわかったのがBistroの中の街灯を光源化するプログラムをCodexが書けたのでClustered Forwardとかも書かせられるな…とかBindless Resourceもサンプルと解説書かせるかができそうとか

bindless対応してたら急にピンクのバーとかが出なくなって、なんで～～～😭😭😭ってなってたら、 画像ファイルが同じ場合キャッシュしたものを使うようにしてたけど、画像サイズが、キャッシュしたときのものを使ってなくて、アスペクトが 0.0 / 0.0 = ∞みたいになってて壊れてたからだった😭 ようやく治った～😭

bindlessにしたことで、13->24FPSまで上昇!(まだキャラとかちゃんとしてないけど、おそらく大体このくらいの速度以上にできるんじゃないだろうか!)

bindlessへの挑戦(今は草のみbindlessでテクスチャ(画像)を適用している!できた!)

WebGPU makes you do unholy and disgusting things to update resources in a large enough scene. Add to that the lack of bindless, and you have a world of pain.