Loading video...

Video Failed to Load

There was a problem loading this video. This could be due to a temporary network issue or the video might be unavailable.

expanding on my WebGPU motion tracking effect work this time with optical flow to visualize the motion of pixels as directional vectors in real time :)

Maxime

14,316 subscribers

19,931 views • 4 months ago •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

0 Comments

No comments available

Comments from the original post will appear here

Related Videos

blob tracking effect built with WebGPU had fun playing with motion-based effects alongside sound over the past few weeks :)

blob tracking effect built with WebGPU had fun playing with motion-based effects alongside sound over the past few weeks :)

Maxime

19,933 views • 5 months ago

Netflix just dropped Go-with-the-Flow Motion-Controllable Video Diffusion Models Using Real-Time Warped Noise

Netflix just dropped Go-with-the-Flow Motion-Controllable Video Diffusion Models Using Real-Time Warped Noise

AK

95,564 views • 1 year ago

I just made the most expressive real-time motion harness for the Reachy Mini. It was created by: > synthesizing a massive motion dataset using Astra (over 10000 samples) > fine-tuning Qwen 3.5 4B using such dataset > then training a flow matching transformer head to transform sparse motion plan (<1hz) into dense expressive motion (25Hz) The average inference time for a full motion is less than 200ms, enabling real-time interaction. Pollen Robotics and Hugging Face only recorded 85 emotions for the reachy mini. With my work, you can now prompt and animate any motion you want. Everything is open-source; write-up and links attached below. Here is a video of Reachy explaining itself through my harness.

I just made the most expressive real-time motion harness for the Reachy Mini. It was created by: > synthesizing a massive motion dataset using Astra (over 10000 samples) > fine-tuning Qwen 3.5 4B using such dataset > then training a flow matching transformer head to transform sparse motion plan (<1hz) into dense expressive motion (25Hz) The average inference time for a full motion is less than 200ms, enabling real-time interaction. Pollen Robotics and Hugging Face only recorded 85 emotions for the reachy mini. With my work, you can now prompt and animate any motion you want. Everything is open-source; write-up and links attached below. Here is a video of Reachy explaining itself through my harness.

Binh

145,957 views • 1 month ago

100 AI Directing Techniques - Cinematic Motion 01｜Tracking Motion Side Tracking · Matched Speed · Motion Blur · Foreground Wipe · Parallax · Handheld Breathing · Dynamic Light 02｜Opposing Motion Opposing Motion · Speed Contrast · Side Tracking · Sharp Subject · Background Streaks · Motion Blur 03｜Anchored Motion Anchored Subject · Vehicle Tracking · Moving Scenery · Near-Far Parallax · Motion Blur · Wind Interaction 04｜Locked-off Motion Locked-off Camera · Static Background · Slow Shutter · Motion Trails · Directional Blur · Still-Motion Contrast 05｜Still Subject Still Subject · Flowing Crowd · Locked-off Camera · Multi-directional Motion · Long Exposure · Foreground Occlusion · Temporal Contrast Use official invite code VIDUAIGUST1 to get 200 free credits, and become a director yourself.

100 AI Directing Techniques - Cinematic Motion 01｜Tracking Motion Side Tracking · Matched Speed · Motion Blur · Foreground Wipe · Parallax · Handheld Breathing · Dynamic Light 02｜Opposing Motion Opposing Motion · Speed Contrast · Side Tracking · Sharp Subject · Background Streaks · Motion Blur 03｜Anchored Motion Anchored Subject · Vehicle Tracking · Moving Scenery · Near-Far Parallax · Motion Blur · Wind Interaction 04｜Locked-off Motion Locked-off Camera · Static Background · Slow Shutter · Motion Trails · Directional Blur · Still-Motion Contrast 05｜Still Subject Still Subject · Flowing Crowd · Locked-off Camera · Multi-directional Motion · Long Exposure · Foreground Occlusion · Temporal Contrast Use official invite code VIDUAIGUST1 to get 200 free credits, and become a director yourself.

Vidu AI

947,386 views • 2 months ago

Behind & Effect Usage of my ‘resonance’ Motion Graphic Work

Behind & Effect Usage of my ‘resonance’ Motion Graphic Work

sharpbin

22,437 views • 23 days ago

Time to put those deformable voxels to work. Here is a very quick softbody test. There are some artefacts and missing motion vectors but there's potential!

Time to put those deformable voxels to work. Here is a very quick softbody test. There are some artefacts and missing motion vectors but there's potential!

Dennis Gustafsson

140,575 views • 2 years ago

UE5.6 AI Real-Time Motion Generation Plugin — powered by a 1-billion-parameter motion model trained on NVIDIA CUDA. It runs locally in real time with 8GB of VRAM and converts webcam or video footage into real-time XYZ skeletal point data and rotation values.

UE5.6 AI Real-Time Motion Generation Plugin — powered by a 1-billion-parameter motion model trained on NVIDIA CUDA. It runs locally in real time with 8GB of VRAM and converts webcam or video footage into real-time XYZ skeletal point data and rotation values.

CYANPUPPETS

28,871 views • 10 months ago

UE5.6 AI Real-Time Motion Generation Plugin — powered by a 1-billion-parameter motion model trained on NVIDIA CUDA. It runs locally in real time with 8GB of VRAM and converts webcam or video footage into real-time XYZ skeletal point data and rotation values.

UE5.6 AI Real-Time Motion Generation Plugin — powered by a 1-billion-parameter motion model trained on NVIDIA CUDA. It runs locally in real time with 8GB of VRAM and converts webcam or video footage into real-time XYZ skeletal point data and rotation values.

CYANPUPPETS

15,968 views • 9 months ago

UE5.6 AI Real-Time Motion Generation Plugin — powered by a 1-billion-parameter motion model trained on NVIDIA CUDA. It runs locally in real time with 8GB of VRAM and converts webcam or video footage into real-time XYZ skeletal point data and rotation values.

UE5.6 AI Real-Time Motion Generation Plugin — powered by a 1-billion-parameter motion model trained on NVIDIA CUDA. It runs locally in real time with 8GB of VRAM and converts webcam or video footage into real-time XYZ skeletal point data and rotation values.

CYANPUPPETS

29,029 views • 9 months ago

I forgot to share this reel of my work I had to make for the Young Guns ceremony earlier this month—it was really fun to see some of my work in motion for the first time! (By "had to make" I obviously mean "had to hire someone to make"—huge ty to motion designer Devon Burgoyne!)

I forgot to share this reel of my work I had to make for the Young Guns ceremony earlier this month—it was really fun to see some of my work in motion for the first time! (By "had to make" I obviously mean "had to hire someone to make"—huge ty to motion designer Devon Burgoyne!)

Elizabeth Goodspeed

76,730 views • 1 year ago

Hooke's law in motion: a dancer's joints anchor a real-time spring-mass lattice. Every extension, contraction, and twist propagates tension through the system; rendering these hypnotic, radiating dashed vectors as emergent force fields.

Hooke's law in motion: a dancer's joints anchor a real-time spring-mass lattice. Every extension, contraction, and twist propagates tension through the system; rendering these hypnotic, radiating dashed vectors as emergent force fields.

Mathematica

1,582,040 views • 7 months ago

Tracking Everything Everywhere All at Once paper page: present a new test-time optimization method for estimating dense and long-range motion from a video sequence. Prior optical flow or particle video tracking algorithms typically operate within limited temporal windows, struggling to track through occlusions and maintain global consistency of estimated motion trajectories. We propose a complete and globally consistent motion representation, dubbed OmniMotion, that allows for accurate, full-length motion estimation of every pixel in a video. OmniMotion represents a video using a quasi-3D canonical volume and performs pixel-wise tracking via bijections between local and canonical space. This representation allows us to ensure global consistency, track through occlusions, and model any combination of camera and object motion. Extensive evaluations on the TAP-Vid benchmark and real-world footage show that our approach outperforms prior state-of-the-art methods by a large margin both quantitatively and qualitatively.

Tracking Everything Everywhere All at Once paper page: present a new test-time optimization method for estimating dense and long-range motion from a video sequence. Prior optical flow or particle video tracking algorithms typically operate within limited temporal windows, struggling to track through occlusions and maintain global consistency of estimated motion trajectories. We propose a complete and globally consistent motion representation, dubbed OmniMotion, that allows for accurate, full-length motion estimation of every pixel in a video. OmniMotion represents a video using a quasi-3D canonical volume and performs pixel-wise tracking via bijections between local and canonical space. This representation allows us to ensure global consistency, track through occlusions, and model any combination of camera and object motion. Extensive evaluations on the TAP-Vid benchmark and real-world footage show that our approach outperforms prior state-of-the-art methods by a large margin both quantitatively and qualitatively.

AK

280,678 views • 3 years ago

🚀 Introducing MegaFlow: Zero-Shot Large Displacement Optical Flow! 🔥 Code, models, and demo — all available now! 🎯 Large displacement motion estimation (optical flow, point tracking) has been a long-standing challenge. MegaFlow proposes a simple solution by leveraging pre-trained vision priors. Combined with a global matching formulation, we tackle both optical flow and point tracking in a unified structure, delivering incredibly strong results. 🏆 SOTA zero-shot performance across multiple benchmarks and in-the-wild videos. Check out the links below to see it in action! 👇

🚀 Introducing MegaFlow: Zero-Shot Large Displacement Optical Flow! 🔥 Code, models, and demo — all available now! 🎯 Large displacement motion estimation (optical flow, point tracking) has been a long-standing challenge. MegaFlow proposes a simple solution by leveraging pre-trained vision priors. Combined with a global matching formulation, we tackle both optical flow and point tracking in a unified structure, delivering incredibly strong results. 🏆 SOTA zero-shot performance across multiple benchmarks and in-the-wild videos. Check out the links below to see it in action! 👇

Haofei Xu

32,340 views • 7 months ago

In the past, AI motion capture was too unstable to form a reliable workflow. Now, with a 1-billion-parameter AI real-time motion model, you can connect a 1080P camera or upload videos to achieve highly stable, low-latency real-time capture. It supports UE, Unity, and Blender, and requires 8GB of VRAM for real-time processing.

In the past, AI motion capture was too unstable to form a reliable workflow. Now, with a 1-billion-parameter AI real-time motion model, you can connect a 1080P camera or upload videos to achieve highly stable, low-latency real-time capture. It supports UE, Unity, and Blender, and requires 8GB of VRAM for real-time processing.

CYANPUPPETS

19,145 views • 7 months ago

Booba in motion this time!

Sensitive content

Booba in motion this time!

Jasmine Stacks

47,743 views • 4 months ago

Blender AI Motion Plugin — a 1-billion-parameter motion model trained on NVIDIA CUDA, running locally in real time with 8GB of VRAM. It converts webcam or video footage into real-time XYZ skeletal point data and rotation values, transmitting them to Blender, Unity, and UE for retargeting.

Blender AI Motion Plugin — a 1-billion-parameter motion model trained on NVIDIA CUDA, running locally in real time with 8GB of VRAM. It converts webcam or video footage into real-time XYZ skeletal point data and rotation values, transmitting them to Blender, Unity, and UE for retargeting.

CYANPUPPETS

125,423 views • 10 months ago

Blender AI Motion Plugin — a 1-billion-parameter motion model trained on NVIDIA CUDA, running locally in real time with 8GB of VRAM. It converts webcam or video footage into real-time XYZ skeletal point data and rotation values, transmitting them to Blender, Unity, and UE for retargeting.

Blender AI Motion Plugin — a 1-billion-parameter motion model trained on NVIDIA CUDA, running locally in real time with 8GB of VRAM. It converts webcam or video footage into real-time XYZ skeletal point data and rotation values, transmitting them to Blender, Unity, and UE for retargeting.

CYANPUPPETS

18,317 views • 11 months ago

Blender AI Motion Plugin — a 1-billion-parameter motion model trained on NVIDIA CUDA, running locally in real time with 8GB of VRAM. It converts webcam or video footage into real-time XYZ skeletal point data and rotation values, transmitting them to Blender, Unity, and UE for retargeting.

Blender AI Motion Plugin — a 1-billion-parameter motion model trained on NVIDIA CUDA, running locally in real time with 8GB of VRAM. It converts webcam or video footage into real-time XYZ skeletal point data and rotation values, transmitting them to Blender, Unity, and UE for retargeting.

CYANPUPPETS

32,540 views • 1 year ago

I got to remake Batman Returns as a stop motion animation for work! 🦇🐈🐧 This actually marks my first time animating the world of Batman on an official WB approved level. #stopmotion #animation #batmanreturns

I got to remake Batman Returns as a stop motion animation for work! 🦇🐈🐧 This actually marks my first time animating the world of Batman on an official WB approved level. #stopmotion #animation #batmanreturns

D.M. Galloway | DGDX Animation

13,605 views • 5 months ago

The free UE5.5 AI motion capture plugin is compatible with Unity, Blender, and VMC. It supports both real-time motion capture via webcam and video uploads to generate 3D motion data. Powered by a 1-billion-parameter motion model running locally on your computer, it requires 8GB of VRAM for real-time processing. Compatible with Nvidia, AMD, and Intel GPUs, it can leverage NVIDIA CUDA to significantly boost graphics computation efficiency.

The free UE5.5 AI motion capture plugin is compatible with Unity, Blender, and VMC. It supports both real-time motion capture via webcam and video uploads to generate 3D motion data. Powered by a 1-billion-parameter motion model running locally on your computer, it requires 8GB of VRAM for real-time processing. Compatible with Nvidia, AMD, and Intel GPUs, it can leverage NVIDIA CUDA to significantly boost graphics computation efficiency.

CYANPUPPETS

17,842 views • 1 year ago