Chris Offner's banner

Chris Offner

@chrisoffner3d • 3,973 subscribers

3D computer vision, spatial AI, and synthetic data @Google XR. visual computing, machine learning, robot perception

Shorts

"Sora is a data-driven physics engine."

"Sora is a data-driven physics engine."

1,743,159 görüntüleme

My team at Google XR just started open-sourcing GNM, our ecosystem of parametric human models and perception stacks, starting with GNM Head. Exciting to see the community adopt and experiment with it so enthusiastically! 🥳

My team at Google XR just started open-sourcing GNM, our ecosystem of parametric human models and perception stacks, starting with GNM Head. Exciting to see the community adopt and experiment with it so enthusiastically! 🥳

136,856 görüntüleme

A high-dimensional Gaussian is a soap bubble: the densest point is the emptiest place. In high dimensions, density and mass diverge completely. For a d-dimensional standard Gaussian distribution, nearly all probability mass is concentrated in a "thin shell" or annulus with radius √d, rather than near the center. The mode has the highest density, but every sample you'll ever draw lives on a thin shell at radius √d. The typical point and the most probable point are in completely different places. Today was the n-th time this has tripped me up, so to commemorate it, I just created this visualization. 🫠

A high-dimensional Gaussian is a soap bubble: the densest point is the emptiest place. In high dimensions, density and mass diverge completely. For a d-dimensional standard Gaussian distribution, nearly all probability mass is concentrated in a "thin shell" or annulus with radius √d, rather than near the center. The mode has the highest density, but every sample you'll ever draw lives on a thin shell at radius √d. The typical point and the most probable point are in completely different places. Today was the n-th time this has tripped me up, so to commemorate it, I just created this visualization. 🫠

112,933 görüntüleme

GNM head model + MediaPipe face landmark tracking

GNM head model + MediaPipe face landmark tracking

84,007 görüntüleme

Doesn't look like the super low-res depth maps from the iPhone Pro's LiDAR sensor would be too useful as supervision (or sanity check) for recent monocular depth estimators.

Doesn't look like the super low-res depth maps from the iPhone Pro's LiDAR sensor would be too useful as supervision (or sanity check) for recent monocular depth estimators.

66,076 görüntüleme

Here's a comparison of Depth Pro [arXiv:2410.02073], Marigold-E2E [arXiv:2409.11355], and Metric3D v2 [arXiv:2404.15506]. Looks like Marigold is attempting to look through the glass while the two metric models aren't. Dmytro Mishkin 🇺🇦

Here's a comparison of Depth Pro [arXiv:2410.02073], Marigold-E2E [arXiv:2409.11355], and Metric3D v2 [arXiv:2404.15506]. Looks like Marigold is attempting to look through the glass while the two metric models aren't. Dmytro Mishkin 🇺🇦

56,434 görüntüleme

Is there an existing iPhone Pro app that simultaneously records the LiDAR stream and a standard RGB video? Apps like Record3D App seem to only store RGB for the points in the LiDAR point cloud but without recording a standard RGB video.

Is there an existing iPhone Pro app that simultaneously records the LiDAR stream and a standard RGB video? Apps like Record3D App seem to only store RGB for the points in the LiDAR point cloud but without recording a standard RGB video.

20,411 görüntüleme

Videos

Anya Rossi

sweetdream.ai

SweetDream.ai•Sponsored•Livecam

Watch Anya Live

Anya is streaming live right now! Join her private show and enjoy exclusive content.

Exclusive private shows

1.2k viewers online

Private Show

Join now for exclusive access

Free preview available • Premium content

Most non-technical people around me are extremely wary of robots, having been conditioned by decades of dystopian sci-fi to see them as threats to people and their humanity. Showing them THE LINE's "Dear Alice" animation often helps them see a more hopeful vision.

Most non-technical people around me are extremely wary of robots, having been conditioned by decades of dystopian sci-fi to see them as threats to people and their humanity. Showing them THE LINE's "Dear Alice" animation often helps them see a more hopeful vision.

4,468,017 görüntüleme • 1 yıl önce

"Sora is a data-driven physics engine."

"Sora is a data-driven physics engine."

1,743,159 görüntüleme • 1 yıl önce

"A Generative Motion Rig for Artist Driven Motion Authoring" from Disney Research Zurich and ETH Zurich So cool! Full video: Paper:

"A Generative Motion Rig for Artist Driven Motion Authoring" from Disney Research Zurich and ETH Zurich So cool! Full video: Paper:

37,722 görüntüleme • 1 ay önce

Christian Rupprecht explains their interpretability research in 3D computer vision, testing if (and where in the model) multi-view transformers like VGGT, DepthAnything 3, and DUSt3R use point/patch correspondences to make sense of 3D scene geometry.

Christian Rupprecht explains their interpretability research in 3D computer vision, testing if (and where in the model) multi-view transformers like VGGT, DepthAnything 3, and DUSt3R use point/patch correspondences to make sense of 3D scene geometry.

74,628 görüntüleme • 5 ay önce

I just tried out playing Counter-Strike in a neural network on my MacBook. In my first run, it diverged into mush pretty quickly. The recording is sped up 5x.

I just tried out playing Counter-Strike in a neural network on my MacBook. In my first run, it diverged into mush pretty quickly. The recording is sped up 5x.

135,866 görüntüleme • 1 yıl önce

"Society should get to a place where we feel like an equitable solution has been found and a bunch of AI companies taking all the money is not really a very acceptable version of that, but that's what's happening right now." – Casey Muratori

"Society should get to a place where we feel like an equitable solution has been found and a bunch of AI companies taking all the money is not really a very acceptable version of that, but that's what's happening right now." – Casey Muratori

49,027 görüntüleme • 8 ay önce

"As roboticists and computer vision people [outside of big tech], do we just wait for the next foundation model?" I share the frustration. It's disempowering when most major progress recently is downstream of "foundation models" that you don't have the compute or data to train.

"As roboticists and computer vision people [outside of big tech], do we just wait for the next foundation model?" I share the frustration. It's disempowering when most major progress recently is downstream of "foundation models" that you don't have the compute or data to train.

68,742 görüntüleme • 1 yıl önce

Good luck 3D-reconstructing this one.

Good luck 3D-reconstructing this one.

77,011 görüntüleme • 1 yıl önce

The iPhone LiDAR depth is pretty stable but low-res and low detail. Monodepth is highly detailed but unstable ("depth flickering"). For more stable and detailed metric depth, I solve for the per-frame affine transform that optimally "anchors" the monodepth to the LiDAR.

The iPhone LiDAR depth is pretty stable but low-res and low detail. Monodepth is highly detailed but unstable ("depth flickering"). For more stable and detailed metric depth, I solve for the per-frame affine transform that optimally "anchors" the monodepth to the LiDAR.

43,137 görüntüleme • 1 yıl önce

Daha fazla içerik yok.