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.

VideoJAM is our new framework for improved motion generation from AI at Meta We show that video generators struggle with motion because the training objective favors appearance over dynamics. VideoJAM directly adresses this without any extra data or scaling 👇🧵

Hila Chefer

4,431 subscribers

168,643 views • 1 year ago •via X (Twitter)

Science & Technology Education

Anya Rossi• Live Now

Private livecam show

11 Comments

Hila Chefer1 year ago

Why do video generators struggle with motion? We found that the pixel-based loss barely changes when video frames are shuffled—showing it is nearly **invariant to temporal incoherence**. This leads models to ignore motion and prioritize appearance

Hila Chefer1 year ago

Our Solution: VideoJAM VideoJAM instills an explicit motion prior by modifying the objective: the model predicts both appearance and motion from a **single learned representation.** This forces the model to capture both visuals and dynamics, improving motion understanding.

Hila Chefer1 year ago

Inner-Guidance: Improving Motion at Inference At inference, we introduce **Inner-Guidance**—a method that leverages the **model’s own motion predictions** as a dynamic guidance signal, steering the generation toward coherent, realistic motion.

Hila Chefer1 year ago

🎬 Results VideoJAM fine-tunes a pretrained video generator (DiT) on just 3M samples from its own training set—yet achieves remarkable motion coherence. It even outperforms highly competitive proprietary models like Sora and Kling in motion quality

Hila Chefer1 year ago

This work was done during my internship at @AIatMeta 🎉 Huge thanks to my amazing collaborators @urielsinger @amit_zhr @YKirstain @adam_polyak90 Yaniv Taigman @liorwolf and @ShellySheynin Check out the project page for many more results and details:

Hila Chefer1 year ago

Now on Huggingface daily papers 🤗 And arxiv 🥳

Akool Inc1 year ago

Need AI avatars or voiceovers? HeyGen and AKOOL both offer powerful AI video tools, but AKOOL delivers superior results with more precise body movements and lip sync. See the difference!

Alex Nasa1 year ago

@AIatMeta A motion prediction auxiliary model make so much sense, did you run it on this benchmark by any chance?

Hila Chefer1 year ago

@AIatMeta Interesting, thanks so much for the reference! I actually think physics is an interesting issue that is far (far) from being solved. Our representation is motion-based but I think it’ll be so cool if we could add an actual physics-driven representation to the mix 🙏🙏

Everett World1 year ago

@AIatMeta That's super impressive - can't wait to generate with it!

Hila Chefer1 year ago

@AIatMeta Thanks Everett! 🫶

Related Videos

New #NVIDIA Paper We introduce Motive, a motion-centric, gradient-based data attribution method that traces which training videos help or hurt video generation. By isolating temporal dynamics from static appearance, Motive identifies which training videos shape motion in video generation. 🔗 1/10

New #NVIDIA Paper We introduce Motive, a motion-centric, gradient-based data attribution method that traces which training videos help or hurt video generation. By isolating temporal dynamics from static appearance, Motive identifies which training videos shape motion in video generation. 🔗 1/10

Xindi Wu

139,325 views • 5 months ago

🔥 Introducing MVLift: Generate realistic 3D motion without any 3D training data - just using 2D poses from monocular videos! Applicable to human motion, human-object interaction & animal motion. Joint work w/ Jiajun Wu & Karen 💡 How? We reformulate 3D motion estimation as generating consistent multi-view 2D pose sequences. Our framework uses 2D motion diffusion to progressively establish multi-view consistency, requiring only single-view 2D pose sequences for training. Project: Video with demonstration: Paper:

🔥 Introducing MVLift: Generate realistic 3D motion without any 3D training data - just using 2D poses from monocular videos! Applicable to human motion, human-object interaction & animal motion. Joint work w/ Jiajun Wu & Karen 💡 How? We reformulate 3D motion estimation as generating consistent multi-view 2D pose sequences. Our framework uses 2D motion diffusion to progressively establish multi-view consistency, requiring only single-view 2D pose sequences for training. Project: Video with demonstration: Paper:

Jiaman Li

15,880 views • 1 year ago

$Example-based Motion Synthesis via Generative Motion Matching paper page: present GenMM, a generative model that "mines" as many diverse motions as possible from a single or few example sequences. In stark contrast to existing data-driven methods, which typically require long offline training time, are prone to visual artifacts, and tend to fail on large and complex skeletons, GenMM inherits the training-free nature and the superior quality of the well-known Motion Matching method. GenMM can synthesize a high-quality motion within a fraction of a second, even with highly complex and large skeletal structures. At the heart of our generative framework lies the generative motion matching module, which utilizes the bidirectional visual similarity as a generative cost function to motion matching, and operates in a multi-stage framework to progressively refine a random guess using exemplar motion matches. In addition to diverse motion generation, we show the versatility of our generative framework by extending it to a number of scenarios that are not possible with motion matching alone, including motion completion, key frame-guided generation, infinite looping, and motion reassembly.$

Example-based Motion Synthesis via Generative Motion Matching paper page: present GenMM, a generative model that "mines" as many diverse motions as possible from a single or few example sequences. In stark contrast to existing data-driven methods, which typically require long offline training time, are prone to visual artifacts, and tend to fail on large and complex skeletons, GenMM inherits the training-free nature and the superior quality of the well-known Motion Matching method. GenMM can synthesize a high-quality motion within a fraction of a second, even with highly complex and large skeletal structures. At the heart of our generative framework lies the generative motion matching module, which utilizes the bidirectional visual similarity as a generative cost function to motion matching, and operates in a multi-stage framework to progressively refine a random guess using exemplar motion matches. In addition to diverse motion generation, we show the versatility of our generative framework by extending it to a number of scenarios that are not possible with motion matching alone, including motion completion, key frame-guided generation, infinite looping, and motion reassembly.

AK

93,694 views • 3 years ago

Google presents Still-Moving Customized Video Generation without Customized Video Data Customizing text-to-image (T2I) models has seen tremendous progress recently, particularly in areas such as personalization, stylization, and conditional generation. However, expanding this progress to video generation is still in its infancy, primarily due to the lack of customized video data. In this work, we introduce Still-Moving, a novel generic framework for customizing a text-to-video (T2V) model, without requiring any customized video data. The framework applies to the prominent T2V design where the video model is built over a text-to-image (T2I) model (e.g., via inflation). We assume access to a customized version of the T2I model, trained only on still image data (e.g., using DreamBooth or StyleDrop). Naively plugging in the weights of the customized T2I model into the T2V model often leads to significant artifacts or insufficient adherence to the customization data. To overcome this issue, we train lightweight Spatial Adapters that adjust the features produced by the injected T2I layers. Importantly, our adapters are trained on "frozen videos" (i.e., repeated images), constructed from image samples generated by the customized T2I model. This training is facilitated by a novel Motion Adapter module, which allows us to train on such static videos while preserving the motion prior of the video model. At test time, we remove the Motion Adapter modules and leave in only the trained Spatial Adapters. This restores the motion prior of the T2V model while adhering to the spatial prior of the customized T2I model. We demonstrate the effectiveness of our approach on diverse tasks including personalized, stylized, and conditional generation. In all evaluated scenarios, our method seamlessly integrates the spatial prior of the customized T2I model with a motion prior supplied by the T2V model.

Google presents Still-Moving Customized Video Generation without Customized Video Data Customizing text-to-image (T2I) models has seen tremendous progress recently, particularly in areas such as personalization, stylization, and conditional generation. However, expanding this progress to video generation is still in its infancy, primarily due to the lack of customized video data. In this work, we introduce Still-Moving, a novel generic framework for customizing a text-to-video (T2V) model, without requiring any customized video data. The framework applies to the prominent T2V design where the video model is built over a text-to-image (T2I) model (e.g., via inflation). We assume access to a customized version of the T2I model, trained only on still image data (e.g., using DreamBooth or StyleDrop). Naively plugging in the weights of the customized T2I model into the T2V model often leads to significant artifacts or insufficient adherence to the customization data. To overcome this issue, we train lightweight Spatial Adapters that adjust the features produced by the injected T2I layers. Importantly, our adapters are trained on "frozen videos" (i.e., repeated images), constructed from image samples generated by the customized T2I model. This training is facilitated by a novel Motion Adapter module, which allows us to train on such static videos while preserving the motion prior of the video model. At test time, we remove the Motion Adapter modules and leave in only the trained Spatial Adapters. This restores the motion prior of the T2V model while adhering to the spatial prior of the customized T2I model. We demonstrate the effectiveness of our approach on diverse tasks including personalized, stylized, and conditional generation. In all evaluated scenarios, our method seamlessly integrates the spatial prior of the customized T2I model with a motion prior supplied by the T2V model.

AK

40,474 views • 2 years ago

We asked Angus about the future of robotics and AI training. “I built an industrial-grade kinematic solver for 6 degrees of freedom motion, with full path planning and joint control." "You can stream AI model outputs into it, or run it like a traditional industrial robot." "The benefit of that is all of that information you can stream and record from those joints… you can use that as the training data." "Frankly, moving a robot with a lever is really crap data for training robots. But this gives you smooth, high-quality motion, without needing two robots.”

We asked Angus about the future of robotics and AI training. “I built an industrial-grade kinematic solver for 6 degrees of freedom motion, with full path planning and joint control." "You can stream AI model outputs into it, or run it like a traditional industrial robot." "The benefit of that is all of that information you can stream and record from those joints… you can use that as the training data." "Frankly, moving a robot with a lever is really crap data for training robots. But this gives you smooth, high-quality motion, without needing two robots.”

TBPN

18,811 views • 1 year ago

Pure Precision. Move AI is pushing 3D character animation and motion capture forward with these powerful releases: – Dex by Move AI: significantly improved hand & finger tracking – New desktop apps: Move Pro now available on Windows, macOS, and Linux – Editor (Beta): fine-tune motion input data with precision Built for animators, developers, and researchers who demand lifelike motion — with total control, without suits. Watch the full video here →

Pure Precision. Move AI is pushing 3D character animation and motion capture forward with these powerful releases: – Dex by Move AI: significantly improved hand & finger tracking – New desktop apps: Move Pro now available on Windows, macOS, and Linux – Editor (Beta): fine-tune motion input data with precision Built for animators, developers, and researchers who demand lifelike motion — with total control, without suits. Watch the full video here →

Move AI

33,144 views • 1 year ago

VIGGLE'S NEW MODEL IS HERE! Controllable video generation is our mission, and character is where we start. Animate any character just as you want, simply with a prompt, or by uploading a reference video featuring clear motion. Try Viggle for FREE:

VIGGLE'S NEW MODEL IS HERE! Controllable video generation is our mission, and character is where we start. Animate any character just as you want, simply with a prompt, or by uploading a reference video featuring clear motion. Try Viggle for FREE:

ViggleAI

37,506 views • 2 years ago

Video Generation is LIVE on imgnAI Discord and Web in Beta! You can now bring your AI-generated images to life with short videos - right inside our Discord bot and web app. This is just the beginning of a new era in creative motion. 👇 🧵

Video Generation is LIVE on imgnAI Discord and Web in Beta! You can now bring your AI-generated images to life with short videos - right inside our Discord bot and web app. This is just the beginning of a new era in creative motion. 👇 🧵

imgnAI by IMGN Labs

19,496 views • 1 year ago

PointOdyssey: A Large-Scale Synthetic Dataset for Long-Term Point Tracking paper page: introduce PointOdyssey, a large-scale synthetic dataset, and data generation framework, for the training and evaluation of long-term fine-grained tracking algorithms. Our goal is to advance the state-of-the-art by placing emphasis on long videos with naturalistic motion. Toward the goal of naturalism, we animate deformable characters using real-world motion capture data, we build 3D scenes to match the motion capture environments, and we render camera viewpoints using trajectories mined via structure-from-motion on real videos. We create combinatorial diversity by randomizing character appearance, motion profiles, materials, lighting, 3D assets, and atmospheric effects. Our dataset currently includes 104 videos, averaging 2,000 frames long, with orders of magnitude more correspondence annotations than prior work. We show that existing methods can be trained from scratch in our dataset and outperform the published variants. Finally, we introduce modifications to the PIPs point tracking method, greatly widening its temporal receptive field, which improves its performance on PointOdyssey as well as on two real-world benchmarks.

PointOdyssey: A Large-Scale Synthetic Dataset for Long-Term Point Tracking paper page: introduce PointOdyssey, a large-scale synthetic dataset, and data generation framework, for the training and evaluation of long-term fine-grained tracking algorithms. Our goal is to advance the state-of-the-art by placing emphasis on long videos with naturalistic motion. Toward the goal of naturalism, we animate deformable characters using real-world motion capture data, we build 3D scenes to match the motion capture environments, and we render camera viewpoints using trajectories mined via structure-from-motion on real videos. We create combinatorial diversity by randomizing character appearance, motion profiles, materials, lighting, 3D assets, and atmospheric effects. Our dataset currently includes 104 videos, averaging 2,000 frames long, with orders of magnitude more correspondence annotations than prior work. We show that existing methods can be trained from scratch in our dataset and outperform the published variants. Finally, we introduce modifications to the PIPs point tracking method, greatly widening its temporal receptive field, which improves its performance on PointOdyssey as well as on two real-world benchmarks.

AK

122,533 views • 3 years ago

Any movement, any subject, any environment. 📸 That's the creative freedom of Meshcapade's Markerless MoCapade. 🚀 We deliver the industry's most accurate 3D motion and body shape data from any single video source. Our AI transforms footage—from subtle social gestures to intense dynamic movements—into precise, high-fidelity 3D data ready for immediate integration (FBX, GLB). Cut the studio time. Get the motion you need, right now. It doesn't get SMPL-er than this. 😉 #MarkerlessMocap #GameDev #3DAnimation #Meshcapade #SMPLX

Any movement, any subject, any environment. 📸 That's the creative freedom of Meshcapade's Markerless MoCapade. 🚀 We deliver the industry's most accurate 3D motion and body shape data from any single video source. Our AI transforms footage—from subtle social gestures to intense dynamic movements—into precise, high-fidelity 3D data ready for immediate integration (FBX, GLB). Cut the studio time. Get the motion you need, right now. It doesn't get SMPL-er than this. 😉 #MarkerlessMocap #GameDev #3DAnimation #Meshcapade #SMPLX

Meshcapade

31,157 views • 8 months ago

Meta announces Movie Gen A Cast of Media Foundation Models We present Movie Gen, a cast of foundation models that generates high-quality, 1080p HD videos with different aspect ratios and synchronized audio. We also show additional capabilities such as precise instruction-based video editing and generation of personalized videos based on a user’s image. Our models set a new state-of-the-art on multiple tasks: text-to-video synthesis, video personalization, video editing, video-to-audio generation, and text-to-audio generation. Our largest video generation model is a 30B parameter transformer trained with a maximum context length of 73K video tokens, corresponding to a generated video of 16 seconds at 16 frames-per-second. We show multiple technical innovations and simplifications on the architecture, latent spaces, training objectives and recipes, data curation, evaluation protocols, parallelization techniques, and inference optimizations that allow us to reap the benefits of scaling pre-training data, model size, and training compute for training large scale media generation models. We hope this paper helps the research community to accelerate progress and innovation in media generation models

Meta announces Movie Gen A Cast of Media Foundation Models We present Movie Gen, a cast of foundation models that generates high-quality, 1080p HD videos with different aspect ratios and synchronized audio. We also show additional capabilities such as precise instruction-based video editing and generation of personalized videos based on a user’s image. Our models set a new state-of-the-art on multiple tasks: text-to-video synthesis, video personalization, video editing, video-to-audio generation, and text-to-audio generation. Our largest video generation model is a 30B parameter transformer trained with a maximum context length of 73K video tokens, corresponding to a generated video of 16 seconds at 16 frames-per-second. We show multiple technical innovations and simplifications on the architecture, latent spaces, training objectives and recipes, data curation, evaluation protocols, parallelization techniques, and inference optimizations that allow us to reap the benefits of scaling pre-training data, model size, and training compute for training large scale media generation models. We hope this paper helps the research community to accelerate progress and innovation in media generation models

AK

62,719 views • 1 year ago

Tired of teleoperation? One human video → 1,000s of robot demos. (📍GitHub ) Scaling Robot Data Without Dynamics Simulation or Robot Hardware Real2Render2Real (R2R2R) is a new way to scale robot data without physics simulation or hardware. You take a phone scan + a single monocular human demo. It tracks the motion, renders photorealistic scenes, and generates diverse, robot-agnostic trajectories ready for training. > No teleop, no sim, no robot, just a phone and a video > Train VLA models and diffusion policies directly on the output > Supports multiple robot embodiments with kinematic consistency > 1000s of demos in 1/27 the time of real-world collection Thank you, Max Fu, for sharing!! Project: Paper: Code coming soon: It shows that with the right pipeline, you can scale robot learning data without touching a robot. One of the most interesting directions in scalable robotics today. —— Weekly robotics and AI insights. Subscribe free:

Tired of teleoperation? One human video → 1,000s of robot demos. (📍GitHub ) Scaling Robot Data Without Dynamics Simulation or Robot Hardware Real2Render2Real (R2R2R) is a new way to scale robot data without physics simulation or hardware. You take a phone scan + a single monocular human demo. It tracks the motion, renders photorealistic scenes, and generates diverse, robot-agnostic trajectories ready for training. > No teleop, no sim, no robot, just a phone and a video > Train VLA models and diffusion policies directly on the output > Supports multiple robot embodiments with kinematic consistency > 1000s of demos in 1/27 the time of real-world collection Thank you, Max Fu, for sharing!! Project: Paper: Code coming soon: It shows that with the right pipeline, you can scale robot learning data without touching a robot. One of the most interesting directions in scalable robotics today. —— Weekly robotics and AI insights. Subscribe free:

Ilir Aliu

42,804 views • 5 months ago

🚨 Jensen Huang says everyone panicked about the AI data when MOST training data was never REAL to begin with. Ilya Sutskever told the industry pre-training was over. "Ilya said, 'We're out of data,' or something like that. 'Pre-training is over,' or something like that," Huang says. "The industry panicked, you know, that this is the end of AI." "And of course, of course that's obviously not true. We're gonna keep on scaling the amount of data that we have to train with." "A lot of that data is probably gonna be synthetic." That's where the panic came from — synthetic data sounds like cheating. "Most of the data that we are training, that we teach each other with, inform each other with, is synthetic." "It's synthetic because it didn't come out of nature." "You created it. I'm consuming it. I modify it, augment it, I regenerate it, somebody else consumes it." The textbook in your hand is synthetic. The post you're reading is synthetic. The lecture you took is synthetic. Nature didn't make any of it. Humans did. AI just learned to do the same thing — faster. "Training is now limited by compute," Huang says. "Data is now limited by compute." The data wall wasn't a wall. It was a mirror. If you're new here, follow @AiEvolutio for the latest on ChatGPT, Claude, and the AI tools shaping how we work and create. — Jensen Huang ( NVIDIA ), NVIDIA CEO, on Lex Fridman's ( Lex Fridman ) podcast

🚨 Jensen Huang says everyone panicked about the AI data when MOST training data was never REAL to begin with. Ilya Sutskever told the industry pre-training was over. "Ilya said, 'We're out of data,' or something like that. 'Pre-training is over,' or something like that," Huang says. "The industry panicked, you know, that this is the end of AI." "And of course, of course that's obviously not true. We're gonna keep on scaling the amount of data that we have to train with." "A lot of that data is probably gonna be synthetic." That's where the panic came from — synthetic data sounds like cheating. "Most of the data that we are training, that we teach each other with, inform each other with, is synthetic." "It's synthetic because it didn't come out of nature." "You created it. I'm consuming it. I modify it, augment it, I regenerate it, somebody else consumes it." The textbook in your hand is synthetic. The post you're reading is synthetic. The lecture you took is synthetic. Nature didn't make any of it. Humans did. AI just learned to do the same thing — faster. "Training is now limited by compute," Huang says. "Data is now limited by compute." The data wall wasn't a wall. It was a mirror. If you're new here, follow @AiEvolutio for the latest on ChatGPT, Claude, and the AI tools shaping how we work and create. — Jensen Huang ( NVIDIA ), NVIDIA CEO, on Lex Fridman's ( Lex Fridman ) podcast

AI Evolution

15,565 views • 1 month ago

Excited to share our latest work on 🎧spatial audio-driven human motion generation. We aim to tackle a largely underexplored yet important problem of enabling virtual humans to move naturally in response to spatial audio—capturing not just what is heard, but also where the sound is coming from. To this end, we introduce the Spatial Audio-Driven Human Motion (SAM) dataset—the first comprehensive dataset featuring paired high-quality human motion and spatial audio recordings. For benchmarking, we develop a generative framework for human MOtion generation driven by SPAtial audio, termed MOSPA, which learns to synthesize realistic and diverse human motions conditioned on spatial audio input. We hope this research could provide a foundation for future research in spatial perception, virtual characters, and embodied AI. The dataset and model will be open-sourced soon. A big thank you to our intern, Shuyang Xu, for the wonderful collaboration! Congratulations, Shuyang! Project page: Paper: Video: #Animation #CG #CV #AIGC #DL #Deeplearning #Motion #Graphics #AI #GenerativeAI

Excited to share our latest work on 🎧spatial audio-driven human motion generation. We aim to tackle a largely underexplored yet important problem of enabling virtual humans to move naturally in response to spatial audio—capturing not just what is heard, but also where the sound is coming from. To this end, we introduce the Spatial Audio-Driven Human Motion (SAM) dataset—the first comprehensive dataset featuring paired high-quality human motion and spatial audio recordings. For benchmarking, we develop a generative framework for human MOtion generation driven by SPAtial audio, termed MOSPA, which learns to synthesize realistic and diverse human motions conditioned on spatial audio input. We hope this research could provide a foundation for future research in spatial perception, virtual characters, and embodied AI. The dataset and model will be open-sourced soon. A big thank you to our intern, Shuyang Xu, for the wonderful collaboration! Congratulations, Shuyang! Project page: Paper: Video: #Animation #CG #CV #AIGC #DL #Deeplearning #Motion #Graphics #AI #GenerativeAI

Zhiyang (Frank) Dou

14,610 views • 1 year ago

Turning video into humanoid robot motion! 🤳🏼 Training humanoid robots needs huge amounts of motion data, but real-world capture doesn’t scale. Mocap is expensive, dangerous edge cases are rare, and you can’t ask humans to repeatedly fall or crash. Video2Robot tackles this by converting videos into physics-grounded humanoid simulations. Motion is generated to respect balance, inertia, ground contact, and joint limits, then directly retargeted to robot simulators. One prompt can generate a full humanoid motion sequence, including multi-agent interactions and failure cases like falls or collisions, scenarios that are hard or impossible to capture safely in the real world. The pipeline is model-agnostic and works with existing video generators, making it a practical way to scale data for robots. If robots are going to operate in the real world, they need to be trained on the failures too, not just the perfect demos. Here's the GitHub: ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →

Turning video into humanoid robot motion! 🤳🏼 Training humanoid robots needs huge amounts of motion data, but real-world capture doesn’t scale. Mocap is expensive, dangerous edge cases are rare, and you can’t ask humans to repeatedly fall or crash. Video2Robot tackles this by converting videos into physics-grounded humanoid simulations. Motion is generated to respect balance, inertia, ground contact, and joint limits, then directly retargeted to robot simulators. One prompt can generate a full humanoid motion sequence, including multi-agent interactions and failure cases like falls or collisions, scenarios that are hard or impossible to capture safely in the real world. The pipeline is model-agnostic and works with existing video generators, making it a practical way to scale data for robots. If robots are going to operate in the real world, they need to be trained on the failures too, not just the perfect demos. Here's the GitHub: ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →

Lukas Ziegler

59,708 views • 6 months ago

This work is not about a new technique. GMT (General Motion Tracking) shows good engineering practices that you can actually train a single unified whole-body control policy for all agile motion, and it works in the real world, directly with sim2real without adaptation. This is different from many existing works that trains/adapts one individual policy for each specific motion. Instead of just showing the beautiful motions/dances, this is actually practical. Because now you can convert all types of commands into real actions in real time without tuning anything. Visit to check out how we handle the following challenges: ✅ Partial Observability. ✅ Hardware Limitations. ✅ Unbalanced Data Distribution. ✅ Model Expressiveness. Work led by Zixuan Chen Mazeyu Ji

This work is not about a new technique. GMT (General Motion Tracking) shows good engineering practices that you can actually train a single unified whole-body control policy for all agile motion, and it works in the real world, directly with sim2real without adaptation. This is different from many existing works that trains/adapts one individual policy for each specific motion. Instead of just showing the beautiful motions/dances, this is actually practical. Because now you can convert all types of commands into real actions in real time without tuning anything. Visit to check out how we handle the following challenges: ✅ Partial Observability. ✅ Hardware Limitations. ✅ Unbalanced Data Distribution. ✅ Model Expressiveness. Work led by Zixuan Chen Mazeyu Ji

Xiaolong Wang

83,261 views • 1 year ago

SCAIL-2 is a new SOTA AI that transfers motion from one video onto another. Works with multiple characters, no need for pose skeletons or masks. Free & open-weights. Interestingly, the author affiliations include Z AI, the team behind GLM.

SCAIL-2 is a new SOTA AI that transfers motion from one video onto another. Works with multiple characters, no need for pose skeletons or masks. Free & open-weights. Interestingly, the author affiliations include Z AI, the team behind GLM.

⚡AI Search⚡

42,964 views • 1 month ago

Just a few months after releasing Scoot, which we feel redefined motion generation quality, our team has made substantial improvements in speed, quality, subtlety, and instruction following. This comes in concert with our newly scaled data collection and labeling efforts, compute increase, substantial architectural improvements, and more stable / scalable training dynamics. Swing is our new flagship movement model. It's live today in Cartwheel studio with infinite batch generation coming soon to our API. Treating Movement as a fundamentally new data type has really unlocked our ability to rapidly make improvements. As it turns out there are far more use cases than you might imagine from entertainment to robotics.

Just a few months after releasing Scoot, which we feel redefined motion generation quality, our team has made substantial improvements in speed, quality, subtlety, and instruction following. This comes in concert with our newly scaled data collection and labeling efforts, compute increase, substantial architectural improvements, and more stable / scalable training dynamics. Swing is our new flagship movement model. It's live today in Cartwheel studio with infinite batch generation coming soon to our API. Treating Movement as a fundamentally new data type has really unlocked our ability to rapidly make improvements. As it turns out there are far more use cases than you might imagine from entertainment to robotics.

Andrew Carr 🤸

150,288 views • 5 months ago

Large-scale 3D Scene Generation (all scenes are real-time rendered)!! Physically-grounded generative data without hallucinations is the missing link for robot learning and testing at scale. We introduce a method that directly generates large-scale 3D driving scenes with accurate geometry, allowing for causal view synthesis and generation with object permanence and explicit 3D geometry. This also allows for extreme trajectory extrapolation without failure! We also show that we can build fully data-driven simulators for end-to-end learning with this approach. Project: with the amazing team of Julian Ost, Amogh Joshi , Andrea Ramazzina, Maximilian Bömer, Mario Bijelic.

Large-scale 3D Scene Generation (all scenes are real-time rendered)!! Physically-grounded generative data without hallucinations is the missing link for robot learning and testing at scale. We introduce a method that directly generates large-scale 3D driving scenes with accurate geometry, allowing for causal view synthesis and generation with object permanence and explicit 3D geometry. This also allows for extreme trajectory extrapolation without failure! We also show that we can build fully data-driven simulators for end-to-end learning with this approach. Project: with the amazing team of Julian Ost, Amogh Joshi , Andrea Ramazzina, Maximilian Bömer, Mario Bijelic.

Felix Heide

27,779 views • 10 months ago

Over the last few months, we’ve been thinking about how to learn from “off-domain” data - data from non-robot sources like video or simulation. These data sources are not quite good enough to learn policies (even monolithic VLA models) directly, but they still contain lots of information that can be useful for generalizable robot control. How can we develop robot learning models that are able to make use of this type of data for generalizable control? In new work, that we call HAMSTER, we show that VLMs can be useful for enabling robotic learning from off-domain data, but specifically when used through hierarchical VLA architectures. We show that this class of models can learn generalizable robot policies for the real world from large-scale, off-domain data. A 🧵 (1/10)

Over the last few months, we’ve been thinking about how to learn from “off-domain” data - data from non-robot sources like video or simulation. These data sources are not quite good enough to learn policies (even monolithic VLA models) directly, but they still contain lots of information that can be useful for generalizable robot control. How can we develop robot learning models that are able to make use of this type of data for generalizable control? In new work, that we call HAMSTER, we show that VLMs can be useful for enabling robotic learning from off-domain data, but specifically when used through hierarchical VLA architectures. We show that this class of models can learn generalizable robot policies for the real world from large-scale, off-domain data. A 🧵 (1/10)

Abhishek Gupta

11,994 views • 1 year ago