Pinned Tweet
This is the last year I create content in 2D.
To prove it I’m releasing an 11-minute 4DGS video in partnership with @4DV_ai where every single shot is dynamic 3D. Over the next 12–18 months, I'll be moving my filming exclusively to dynamic 3D.
Thank you so much to @NVIDIAAI and @Dell for making this video possible.
The weekly wrap up of gaussian splatting is now out!
Watch here: youtu.be/fTGJ5WaW_Qs
Radiance Fields retweeted
5/ What’s next: Gaussian splats.
Splats capture a real person, object or place from photos or video and turn it into millions of tiny 3D points.
The result is real-world footage you can drop into a 3D scene, then light, animate and render like CGI. Artists are already using them in music videos and installations.
Been deep in pixel-art gaussian splats the last couple days.
People asking for "prompts" so here is how it actually works instead... it's not just a single prompt, each one is a single full spec but it encompasses all of the following.
The core is a bunch of 3D gaussians, little oriented ellipsoids, all built in code.
No images, no models loaded. Each one gets projected to the screen, sorted back to front every frame, and drawn with a firm core so they read as crisp pixels instead of fuzz.
Render to a 640x360 buffer, scale up 3x, snap to a palette, light dither. That's where the pixel art look comes from.
Lighting every spat makes them look much better, warm light (like fire with the tree for example) vs cool light like the moonlight. Flat colors splats look terrible (like the qwen 27b example).
Radiance Fields retweeted
Gaussian Splat relight in Blender 5.3 + Cycles
using @theworldlabs Atlas to generate the splat
Custom delighting + normals rebuilt from traced depth
All tv shows & movies should be scanning and distributing splats for marketing at this point.
This took me <10 minutes with a Hesai @KosmoAIOfficial
New Spirula Studio update.
Address rolling shutter, focus breathing, and small pose errors.
Experimental GLOMAP and FlexGlue feature matcher join the internal SfM that was already crazy fast.
Article: radiancefields.com/spirula-s…
Article: radiancefields.com/unreal-en…
2
487
Radiance Fields retweeted
Meet Iris-Z: the first general image model to combine appearance and geometry, beating task-specific depth models like MoGe-3.
Generate images + depth; estimate and condition on depth. One backbone.
9B/32B. Fully open: weights, code, demo (Apache 2.0)
🌐 speridlabs.com/research/iris…
The ArcGIS Maps SDK for JavaScript 5.2 from @Esri just added Viewshed and Shadow Cast analysis for Gaussian Splats
Full Article: data.radiancefields.com/esri…
267
Radiance Fields retweeted
Cutting floaters off of Gaussian Splats like this is unironically a really meditative experience!
Thrilled to release InfiniSplat v3 (SplatAnything)!
We teamed up with @TencentHunyuan to scale up InfiniSplat v2’s training data & introduced Gaussian Spacing Normalization for high-quality rendering across all scales, from object to aerial.
Code is now open-source👇
Radiance Fields retweeted
📢📢📢 Introducing Spaghetti Splatting
3DGS is a convolution of a Dirac with a Gaussian Kernel. We provide a strict generalization of 3DGS, extending its Dirac mass to lines and curvilinear structures
project page: spaghetti-splatting.github.i…
arXiv: arxiv.org/abs/2611.12716
🔊 on!
Article: radiancefields.com/blender-5…
262









