X / Twitter
Introducing StableRecon: The fast food of photogrammetry! 📹➡️3D
https://t.co/9UY5xH3yph
Turn your videos into glb and splat “in a snap”. It's fast & accessible, with room for growth.
Try it out and share your results - the good, the bad, and the blocky! 🏛️🏠🗿 https://t.co/7wslos17uI
If you need relevancy evals, boostrap them with an llm as a judge
https://t.co/5dOWRD1qeo
gave a private presentation in nyc about generating useful ideas for ai applications for nontechnical people (aka How To Talk To Your AI Engineer)
really liking the idea of "power adjectives":
take X, where X = thing you already know and do and love and use
and add an adjective to it - "automated", "personalized", "proactive", "creative", ____?
that adjective is the AI piece, but you dont care that is AI really, you want the end result. spell that out in as much detail as possible, and the ai engs got it from there
A new version of 3D Tiles Renderer from @nasajpl has been released! This release adds components for react-three/fiber so you can easily drop 3d tiles into r3f scenes & integrate with ecosystem components, including postprocessing & drei!
1/2
#threejs #r3f #drei #react #3dtiles https://t.co/T5koGcAzkw
M͓̽i͓̽n͓̽e͓̽r͓̽u͓̽n͓̽n͓̽e͓̽r͓̽ 🏃💣1⃣2⃣3⃣ https://t.co/WaqErOuK9M
Most case studies are thinly veiled advertorials
but this @Altera_AL piece with @openai seems to contain some actual alpha on cognitive architecture for very long running game-NPC-level agents with memory/personality/emotional intelligence.
destroys @drjimfan's voyager on their own minecraft benchmark - 4hrs of autonomous operation
TouchOSC spotted in this @MKBHD video visiting Disney Imagineering... running on a Steam Deck no less!
Full video here:
https://t.co/DcGe3U2AIJ
TouchOSC Steam Deck install instructions:
https://t.co/rgdD53jVao https://t.co/SQQpXCg83c
Super easy to generate speech with F5 TTS + MLX locally thanks to @lllucas
1. pip install f5-tts-mlx
2. python -m f5_tts_mlx.generate --text "Hello world"
3. afplay output.wav (🔊)
@NVIDIAAI recently released InstantSplat which allows for fast 3D reconstruction, so I decided I wanted to make a demo of it using @rerundotio and @Gradio! It's one of the fastest ways to go from sparse images -> Gaussian splat I've come across. https://t.co/KIK5a7yCUP https://t.co/sTgE9ot8D7
Google is on fire with their open source releases! 🔥
Today, they dropped Gemma-APS, a collection of Gemma models for text-to-propositions segmentation. The models are distilled from fine-tuned Gemini Pro models applied to multi-domain synthetic data! 👇
https://t.co/DNO6ezDFnY
LOTUS: Diffusion-based Visual Foundation Model for High-quality Dense Prediction
Monocular depth estimation is so hot right now!
https://t.co/z6YK6a7xgC
hooray, the "Bayesian Models of Cognition" book is out, for all your Bayesian models of cognition needs :)
(I contributed to 3 chapters in the book -- development, intuitive physics, theory of mind -- that can be read on the ol' website) https://t.co/wzDBlVdlk6
I just came across one of the coolest free fonts I’ve seen in a while.
Think Helvetica, but with pixel alternates. Every time you type, different pixels emerge.
Download Free Font (№76) below.
Design by @Fontfabric https://t.co/Wefwu2wHKy
Extremely helpful advice I got for the baby-toddler-preschool years: never try to make a happy child happier
👄 TANGO: Co-Speech Gesture Video Reenactment with Hierarchical Audio Motion Embedding and Diffusion Interpolation 🔥 Jupyter Notebook 🥳
🔥 This ongoing work is planned for release by the end of November or early December. 🔥
Thanks to @HaiyangLiu3 ❤ Xingchao Yang ❤ Tomoya Akiyama ❤ Yuantian Huang ❤ Qiaoge Li ❤ Shigeru Kuriyama ❤ Takafumi Taketomi ❤
🌐page: https://t.co/eVpP4Frcq9
📄paper: https://t.co/RKzcbxCBKO
🧬code(coming soon...): https://t.co/K32BkQNdXS
🍊jupyter(prototype model): https://t.co/s7naJeZyc4