from the future
9.3K posts
from the future
@nk
Plasticity https://t.co/QKSvYoG0yy
Paris, France Katılım Kasım 2007
1K Takip Edilen62.6K Takipçiler
hard to believe how well SAM3 segmentation works even on this kind of basically-featureless image ... no texture or detail at all. Prompts are "head" and "face".
English
@TheNextDecade if `2d snapping` is on then all snapping will projected to the current construction plane... if not that is a bug...
English
The same approach works for trimmed surfaces since the fragment shader uses interpolated UVs to evaluate the Bezier normal. The extra compute and VRAM cost is modest - certainly smaller than a denser mesh of equivalent quality. Hopefully I’ll have time to write this up soon.
English
Something I’ve been researching for a bit: smoother CAD body shading without generating expensive, dense meshes.
For Bezier surfaces, the fragment shader can evaluate the surface normal analytically instead of using interpolated mesh normals. So even a coarse mesh can have clean highlights.
English
Blog post before merge would have been marginally better comms. however there was no way to avoid becoming the main character on twitter because of this change. too many hot button issues (ai, manual memory management, zig vs rust, etc.). I'm impressed with your courage - I hope it works out.
English
@Shreyassanthu77 @thdxr yeah I should’ve waited before pressing merge
English
@CongoCart @ibuildthecloud It is pretty sad how people keep making mediocre versions of electron (using the buggy and slow system webview), claim it’s more efficient (which it won’t be when you inevitably need to use CEF) and get 3,000 likes on twitter.
English
Actually it isn't JS that is a memory hog, it is WebViews -- the minute you strap a webview onto something it is going to eat > 200mb at the minimum!
I did an entire blog series with detailed measurements. Tauri, Wails, Electron, etc -- they all use about the same memory because of the Webviews.
The interesting thing is that Native doesn't always fix the issue, for example; Flutter Desktop is also just as big of a memory hog as Electron/Tauri/Wails.
However the biggest memory issues are caused by poor coding practices and not caring.
x.com/CongoCart/stat…
English
I don't get why people keep reinventing electron. The problem with electron is JavaScript, not the container. JavaScript is an absolute pig when it comes to memory and CPU. I would love for someone to really fix this for desktop apps.
Chris Tate@ctatedev
Introducing zero-native Build native desktop + mobile apps with web UI and Zig → Tiny binaries, low memory usage → Selectable web engines (WKWebView, WebKitGTK, WebView2, Chromium/CEF) → Next.js, Vue, Svelte, Vite, React → macOS, Linux, Windows, iOS, Android
English
@jarredsumner I’m curious how good mythos (and other internal models) are at statically finding leaks and crashes. Not yet a replacement for a more rigorous type system
English
why: I am so tired of worrying about & spending lots of time fixing memory leaks and crashes and stability issues. it would be so nice if the language provided more powerful tools for preventing these things.
English
99.8% of bun’s pre-existing test suite passes on Linux x64 glibc in the rust rewrite
English
@emm0sh @broblas @getPlasticity External references are sort of “easy” but the question is what to do when the original ref file is gone. We can keep a copy in the .plasticity file but then I kind of wonder what the value is.
English
@broblas @getPlasticity plasticity? it’s fast direct modeling and really affordable. runs on linux and mac too
English
Instances are more powerful than ever. You can create instances with arrays, mirror commands, and more.
Try Plasticity for free at plasticity.xyz
English
@mferraro89 qCreatedBy and skSolve seems a bit close. My own internal API is also somewhat similar but I'm a bit nervous about IP issues.
English
@brotzky I could certainly style the blank content better and animate the window size. But in my case, would you agree that the blank content looks weird?
English
@nk It's because the webview I insert loads basically instantly :)!
I thought of the did-finish-load but felt that masks the problem a bit. Personal preference, but I prefer to see something "working" instead of nothing.
x.com/brotzky/status…
Brotzky@brotzky
"AI slop" or "vibe coding" is so wrong. I spent the weekend making my site stupid fast. Results? Instant page loads, no stale data, and the fastest app I've ever built. All the little details 🧵
English
@brotzky One thing you can also look into is bytenode. It will complicate your build process but it will reduce your javascript parse/eval time by ~50%. I think for plasticity the difference is 800ms parse without vs 400ms with bytenode.
English
For people who’s cured or significantly managed their migraines, what did you do? My gf has really bad ones and we’re throwing everything at the wall seeing what sticks
English
@kevinakwok Pretty strong proof that dwarkesh is the best tech interviewer of the age. I think very few podcasters could have evoked that moment.
English
Jensen Dwarkesh podcast was a true scissor statement
Haven't watched it yet but so funny how everyone I know agrees one of them was so good and the other so bad. Just can't agree on who
English
@onirenaud Are render bundles actually as fast as multidraw cpu side though? I thought they still do a lot of validation
English
Taking a shot at this we never know 👀
BatchedMesh powered by MultiDrawIndirect has been a dream for a long time.
#issuecomment-4259276305" target="_blank" rel="nofollow noopener">github.com/gpuweb/gpuweb/…
Renaud@onirenaud
What is happening with Multi Draw in WebGPU? Chrome 131 shipped experimental support for multi-draw indirect that didn't work for 10+ versions, and even 15 releases later there is still no update? Multi Draw is widely supported by backends and is an obvious win.
English
@DSculptors No it’s a new technology developed by jorg peters. He is currently working on creasing.
English
@nk The PloySpline looks amazing. Congratulations! Does it support crease full/semi as well? Is it part of the Parasolid functionality?
English
I talk here a bit about Polysplines, a new kind of spline surface, which is a generalization of the b-spline and (hopefully) a successor technology for Catmull-Clark.
youtube.com/watch?v=FFEIhp…
YouTube
English