David Lake retweeted
There's something weird in Zip-NeRF (jonbarron.info/zipnerf/) that I'd like feedback about. As part of the project, we needed a way to turn conical frustums into small sets of points, which we use for multisampling iNGP codes. Here's what they look like for 4 frustums ("frusta"?)
David Lake retweeted
Introducing TAPIR & RoboTAP, our latest research from @GoogleDeepMind.
It focuses on spatial intelligence via point tracking, outlining how it enables applications from robotics to video generation to augmented reality, and more! deepmind-tapir.github.io/blo… robotap.github.io/
David Lake retweeted
Shader programming best practices, some target NVidia GPUs but most are widely applicable developer.nvidia.com/blog/ad…
David Lake retweeted
Excited to kick off a brand new course at @CarnegieMellon: Monte Carlo Methods and Applications
Monte Carlo uses random sampling to solve tough problems throughout science & engineering. We cover everything from fundamentals to real-world implementation:
geometry.cs.cmu.edu/montecar…
David Lake retweeted
Fuzzy Metaballs+ use 3D Gaussians for differentiable rendering, integrating optical flow and producing watertight meshes.
These techniques ensure speedy, reliable reconstructions on both GPU and CPU.
github.com/leonidk/fmb-plus
David Lake retweeted
In our first lecture in #MCMA2023, we give a broad overview of the Monte Carlo method, introduce the integration & sampling problems, and take a brief look at how it helps with applications from image generation to solving PDEs to finance to natural language to generative AI!
David Lake retweeted
Another 3D Gaussian Splatting test with 360 video footage, this time with a bigger scene.
Hôtel Benoît de Sainte Paulle, Paris captured with the @insta360 RS 1", and running in real-time.
#NeRF #GaussianSplatting
David Lake retweeted
289 chars #つぶやきGLSL
for (O *= t;
O.a++ < 30.;
O += (1. + cos(k+k+t+vec4(0,1,2,0))) / 2e2 / L
)
p = R - vec3(F+F, R.y),
p = t/L*p - 3./R,
M(p.zx) M(p.yx)
p.x -= 2., t -= L - .1;
Real-time: shadertoy.com/view/dlc3Wl
David Lake retweeted
Smith: you were attacking my family😡🤬
Walker: WHAT!😳😨
teapaw.org/aowclip/status/1657489…
David Lake retweeted
now that this is over, what was Anthony Smith talking about here???? teapaw.org/aowclip/status/1657489…
David Lake retweeted
Prof. Gil Strang's last lecture is ... Monday! Watch it via live stream: grinfeld.org/strang/
David Lake retweeted
The value of reading, and of writing things down that we read, or hear, cannot be overstated. Two expert guests (who specialize in speech and memory) on the Huberman Lab podcast explained that when we read text or listen to something and then write key aspects/takeaways down by hand- not typing, it engages our motor control centers in ways that deeply embed that information to our memory. Taking notes, however cursory, turns out to be the best way to remember and implement information later.
David Lake retweeted
PSA for LaTeX users: GPT-4 almost makes TikZ usable (almost).
Behold, this TikZ diagram of a simple perceptron with three input nodes, generated zero-shot:
David Lake retweeted
After 1.5 years of hard work, I am thrilled to share with you Φ-SO - a Physical Symbolic Optimization package that uses deep reinforcement learning to discover physical laws from data . Here is Φ-SO discovering the analytical expression of a damped harmonic oscillator👇
[1/6]
David Lake retweeted
Air Force 1 collabs in Midjourney
A sneaker prompt for you to remix today. Try changing the shoe brand/model, the collab, the colorway, the materials, or all of the above.
🧵 Thread w/ prompts
David Lake retweeted
A short🧵about Grounded Decoding
grounded-decoding.github.io
(by @wenlong_huang et al.)
There is a concept here, building on similar work (e.g., LM-Nav, SayCan, etc.) about how affordances, dynamics, and LMs interact to get grounding. It's a general idea that can be used broadly 👇
David Lake retweeted
#つぶやきGLSL
float e,i,a,w,x,g;for(;i++<1e2;){vec3 p=vec3((FC.xy-.5*r)/r.y*g,g-3.);p.zy*=rotate2D(.6);i<1e2?p:p+=1e-4;e=p.y;for(a=.8;a>.003;a*=.8)p.xz*=rotate2D(5.),x=(++p.x+p.z)/a+t+t,w=exp(sin(x)-2.5)*a,o.gb+=w/4e2,p.xz-=w*cos(x),e-=w;g+=e;}o+=min(e*e*4e6,1./g)+g*g/2e2;
























