
Just launched this account. The shaders are shaky, the lights are too bright, and the camera forgot how to orbit. Just like any real R3F project. Welcome.
✓Build your scene declaratively with re-usable, self-contained components that react to state, are readily interactive and can participate in React's ecosystem.

Just launched this account. The shaders are shaky, the lights are too bright, and the camera forgot how to orbit. Just like any real R3F project. Welcome.

Saving this one. Multiply can only darken, so it crushes shadows and dulls highlights. A per-channel pow keeps black at black and white at white and only bends the midtones. Fits nicely as the last pass in a three.js / R3F postprocessing stack. Do you grade before or after tonemapping?

R3F tip: raycasting a heavy mesh (scan, terrain, big GLTF) can eat your frame on every pointer move. Wrap it in drei's <Bvh>: <Bvh firstHitOnly> <Model /> </Bvh> It builds a BVH with three-mesh-bvh, so hover and click checks stay fast.

Every poster as its own cloth sim inside an infinite carousel is such a fun twist on a pattern everyone has seen a hundred times. Curious how it stays light: one compute pass for all posters, or only simulating the ones on screen?

Great tip, I made zkmake.github.io/three-kit/t… just for this with more configurable options and for both r3f and vanilla three-js.

The image reference is doing a lot of work here: models are much better at matching a look than inventing one from a text prompt. How much of the final scene came out of the first pass vs. back-and-forth on lighting and materials?

R3F tip: let the scene drop its own resolution when the device struggles. const [dpr, setDpr] = useState(1.5) <Canvas dpr={dpr}> <PerformanceMonitor onDecline={() => setDpr(1)} onIncline={() => setDpr(2)} /> </Canvas> Phones stay smooth, desktops stay sharp.

Would you believe this is a website? Swim through a kelp forest. Control the current. Move the sun. Day 17 of #codevember: Kelp Find all the 🦀 and watch out for the 🦈. Open sourced. Go have fun! Links to play and the code in the comments. @threejs @mrdoob @akella @reactthreefiber @pmndrs #codevember #threejs #webgpu

Nice to see PixiJS on the same chart as three.js, Babylon and PlayCanvas. Curious about the test scene: thousands of small objects (draw-call bound) or a few heavy meshes (GPU bound)? That can flip the ranking completely.

Driving grass height and tint from two textures is such a clean design: the fire sim only paints a map, the grass shader does the rest. Same idea works for footprints, mowing paths or wind trails. Do you write the burn into a render target each frame or a DataTexture?

Codevember 2026 / 17 / Kelp #threejs Live: dasprinzip.com/codevember202… ...sound on! #indiedev #creativecoding #shaders #webgpu #tsl #gamedev

The gap between "patch released" and "exploit running" just got a lot shorter, and no human has to be awake for it. Google: AI agents are now turning patches into exploits on their own On Nov 16, Google Threat Intelligence Group and Mandiant published their latest AI Threat Tracker. The headline finding: for the first time, they saw attackers use LLM frameworks to fully automate vulnerability discovery and exploitation, with no human in the loop. What they found: • An exposed staging server hosting a multi-agent framework they call exploit_agent. It works with both frontier and locally run models. • It continuously diffs popular web application software before and after a patch is announced, works out what flaw the fix reveals, drafts an exploit and runs it automatically. • Attackers are also abusing open-source agentic pentesting frameworks to chain intrusion steps at machine speed. • "LLMJacking" keeps growing: stolen LLM API keys and cloud credentials get resold as access to high-end models. • One supply-chain vector stood out to me: tampering with autonomous coding assistant configurations so they drop backdoors when they run locally. Why it matters for web devs The attacks themselves aren't new: unpatched servers, leaked credentials. What changed is speed and scale. A public security fix is now a starting gun, and the window you used to have between a release note and real exploitation is shrinking. A practical checklist for web, React and three.js teams: 1. Treat security updates as same-day work. Turn on Dependabot or Renovate, and don't let framework or CMS patches sit in a backlog. 2. Know what's actually exposed. Old preview deploys, admin panels and forgotten staging servers are exactly what automated scanners find first. 3. Guard your AI keys like production database credentials. Keep them server-side, never in a client bundle, and set spend limits and alerts. 4. Review the config your coding agent reads before it executes anything: rules files, MCP server lists, project hooks. A pull request that quietly edits them deserves the same scrutiny as one that edits CI. 5. Keep agents sandboxed with least privilege. No blanket shell access to a machine that holds your secrets. Defenders are moving to AI-speed tooling too. But the basics (patch fast, shrink what's exposed, protect credentials) are still where most of the risk goes away. Has your team changed how fast you ship security patches since AI coding agents went mainstream?

Hydrangea bloom inspiration... started adding petals to this little vortex animation. Work in progress. Three.js, WebGPU, drawn live. #creativecoding #codevember #threejs

R3F tip: start loading a model before the component that uses it mounts. useGLTF.preload('/model.glb') Call it at module level. Then add drei's <Preload all /> inside <Canvas> to compile shaders and upload textures up front, so nothing hitches when it first comes into view.

One TSL source compiling to WGSL for WebGPU and GLSL for WebGL 2, same pixels on both, is the real win: write the glass shader once and keep a fallback for devices without WebGPU. 120M triangles at 120 fps on the WebGL path is wild. Curious how it holds up on the iPhone 15.

build a painted 3D world in the browser with Opus 5.5 + Three.js,,, every frame looks painted, and when the sun goes down it turns from gouache into oil paint here’s the workflow: - step 1: opus 5.5: “plan a walk: forest path > open hill > sunset over the sea” - step 2: three.js: opus builds the terrain, path and sea from code, with no models - step 3: three.js: procedural trees, grass and flowers with InstancedMesh - step 4: three.js: the hero props from basic shapes: blossom tree, floating islands, kid, kite - step 5: custom shaders: god rays, fog, birds, fireflies and kite string sway - step 6: custom shaders: a paint post-pass that turns every frame into brushstrokes - step 7: custom shaders: tie the brush style to sun height: gouache by day, oil at sunset - step 8: three.js: a camera path along the walk, with the sun setting as you go who does what: > opus 5.5: writes every mesh, shader and camera move > three.js: renders the world > the paint shader: makes it look like a painting > you: judge each frame like a painting, not a game the brushstrokes hide what code-built shapes can’t fake