Loading video...

Video Failed to Load

Go Home

Implementing motion imitation methods involves lots of nuisances. Not many codebases get all the details right. So, we're excited to release MimicKit! A framework with high quality implementations of our methods: DeepMimic, AMP, ASE, ADD, and more to come!

xbpeng4's profile picture

Jason Peng

6,336 subscribers

171,725 views • 1 year ago •via X (Twitter)

Science & Technology
LIVE

Anya Rossi• Live Now

Private livecam show

8 Comments

Jason Peng's profile picture
Jason Peng1 year ago

A little historical note, the spin-kick was actually the first motion we got to work with DeepMimic. It has since become a "hello world" for many of our projects.

Justin Chen's profile picture
Justin Chen1 year ago

This is awesome! How do you keep outputting so much research so quickly?

Jason Peng's profile picture
Jason Peng1 year ago

I'm lucky to have great students 🙂

Andrew Carr 🤸's profile picture
Andrew Carr 🤸1 year ago

This is really excellent work - thanks for putting out such high quality code

David Bar's profile picture
David Bar1 year ago

Looking good

5D Solar Shift's profile picture
5D Solar Shift11 months ago

@grok lets create advancement ideas that are innovative and ingenious in a large list format of ideas!

Xiatao Sun's profile picture
Xiatao Sun11 months ago

Nice! Getting motion imitation details right is a pain. Good to see DeepMimic, AMP, and ASE all in one place.

Astrid Wilde 🌞's profile picture
Astrid Wilde 🌞1 year ago

bless you

Related Videos

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.
5:56

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

93,765 views • 3 years ago

We're excited to partner with Multiply Labs. Their cutting edge technology will bolster our efforts to add 10 years to healthy human lifespan. Read the press release for more details:
0:30

We're excited to partner with Multiply Labs. Their cutting edge technology will bolster our efforts to add 10 years to healthy human lifespan. Read the press release for more details:

Retro Biosciences

10,559 views • 2 years ago

We cannot air condition our way out of the heat crisis. Many cooling methods add planet warming emissions, in-turn increasing cooling demand. But solutions exist to make cooling more sustainable & affordable: #COP30
1:00

We cannot air condition our way out of the heat crisis. Many cooling methods add planet warming emissions, in-turn increasing cooling demand. But solutions exist to make cooling more sustainable & affordable: #COP30

UN Environment Programme

12,107 views • 10 months ago

We're so excited to share the first details with you about our new event - SideQuest: A Little Off-season Adventure! January 1st - 5th, 2025, near Minneapolis, MN! Check out some details in this thread, with many more to come. Make sure you're subscribed to the newsletter!
0:28

We're so excited to share the first details with you about our new event - SideQuest: A Little Off-season Adventure! January 1st - 5th, 2025, near Minneapolis, MN! Check out some details in this thread, with many more to come. Make sure you're subscribed to the newsletter!

CAPCon

105,006 views • 2 years ago

The North Western Modelers are VERY excited to announce our summer show will take place at the Railroad Museum of Pennsylvania in Strasburg, PA on July 26th and 27th to celebrate 25 Years of Thomas & The Magic Railroad! There's LOTS more details to come, so stay tuned!
0:49

The North Western Modelers are VERY excited to announce our summer show will take place at the Railroad Museum of Pennsylvania in Strasburg, PA on July 26th and 27th to celebrate 25 Years of Thomas & The Magic Railroad! There's LOTS more details to come, so stay tuned!

North Western Modelers

45,831 views • 1 year ago

Introducing ConceptAttention, an approach to interpreting diffusion transformer models! Write a prompt, choose some concepts, generate an image, and get high-quality heatmaps of text concepts. Our method outperforms existing methods like cross attention. Link to demo 👇
0:30

Introducing ConceptAttention, an approach to interpreting diffusion transformer models! Write a prompt, choose some concepts, generate an image, and get high-quality heatmaps of text concepts. Our method outperforms existing methods like cross attention. Link to demo 👇

Alec Helbling

36,655 views • 1 year ago

Loopy Taming Audio-Driven Portrait Avatar with Long-Term Motion Dependency paper page: With the introduction of diffusion-based video generation techniques, audio-conditioned human video generation has recently achieved significant breakthroughs in both the naturalness of motion and the synthesis of portrait details. Due to the limited control of audio signals in driving human motion, existing methods often add auxiliary spatial signals to stabilize movements, which may compromise the naturalness and freedom of motion. In this paper, we propose an end-to-end audio-only conditioned video diffusion model named Loopy. Specifically, we designed an inter- and intra-clip temporal module and an audio-to-latents module, enabling the model to leverage long-term motion information from the data to learn natural motion patterns and improving audio-portrait movement correlation. This method removes the need for manually specified spatial motion templates used in existing methods to constrain motion during inference. Extensive experiments show that Loopy outperforms recent audio-driven portrait diffusion models, delivering more lifelike and high-quality results across various scenarios.
0:45

Loopy Taming Audio-Driven Portrait Avatar with Long-Term Motion Dependency paper page: With the introduction of diffusion-based video generation techniques, audio-conditioned human video generation has recently achieved significant breakthroughs in both the naturalness of motion and the synthesis of portrait details. Due to the limited control of audio signals in driving human motion, existing methods often add auxiliary spatial signals to stabilize movements, which may compromise the naturalness and freedom of motion. In this paper, we propose an end-to-end audio-only conditioned video diffusion model named Loopy. Specifically, we designed an inter- and intra-clip temporal module and an audio-to-latents module, enabling the model to leverage long-term motion information from the data to learn natural motion patterns and improving audio-portrait movement correlation. This method removes the need for manually specified spatial motion templates used in existing methods to constrain motion during inference. Extensive experiments show that Loopy outperforms recent audio-driven portrait diffusion models, delivering more lifelike and high-quality results across various scenarios.

AK

128,912 views • 2 years ago

In line with this year's #WorldFoodSafetyDay theme of "Science in action," here's how we partner with a third party laboratory to apply scientific rigour to our testing methods and build trust in the safety and quality of food products in our stores.
2:26

In line with this year's #WorldFoodSafetyDay theme of "Science in action," here's how we partner with a third party laboratory to apply scientific rigour to our testing methods and build trust in the safety and quality of food products in our stores.

Massmart

51,462 views • 1 year ago

Introducing: LetzAI V4. We're excited to release the newest version of LetzAI with Video Models, a new Image Editor, a new Upscaler, an improved LetzAI Chat, and a better UI. All at $0.01 per high-quality image. Create what you love.
1:23

Introducing: LetzAI V4. We're excited to release the newest version of LetzAI with Video Models, a new Image Editor, a new Upscaler, an improved LetzAI Chat, and a better UI. All at $0.01 per high-quality image. Create what you love.

LetzAI

113,607 views • 1 year ago

Time to get a move on, because we're about halfway through our Motion Design Tips & Tricks series, with more videos to come each week⚙️ Put your project in motion with our playlist collecting all the learnings shared so far, including today's video on Dynamic Shape Actors:
2:14

Time to get a move on, because we're about halfway through our Motion Design Tips & Tricks series, with more videos to come each week⚙️ Put your project in motion with our playlist collecting all the learnings shared so far, including today's video on Dynamic Shape Actors:

Unreal Engine

12,457 views • 1 year ago

There are many ways to collect high-quality data. GenRobot seems to be closer to human-like methods. It uses a wearable exoskeleton robotic hand: a person simply wears the Gen DAS Dex and can flexibly complete various tasks, while simultaneously capturing the entire human motion trajectory and converting it into high-quality data for use by the humanoid robot. >23 degrees of freedom omnidirectional hand capture with millimeter-level precision >Tactile and visual fusion >Sub-millisecond head-hand synchronization
1:33

There are many ways to collect high-quality data. GenRobot seems to be closer to human-like methods. It uses a wearable exoskeleton robotic hand: a person simply wears the Gen DAS Dex and can flexibly complete various tasks, while simultaneously capturing the entire human motion trajectory and converting it into high-quality data for use by the humanoid robot. >23 degrees of freedom omnidirectional hand capture with millimeter-level precision >Tactile and visual fusion >Sub-millisecond head-hand synchronization

CyberRobo

18,098 views • 6 months ago

Motion Shooting: last week of summer access for us quarter schools. Excited about our improvement and commitment to excellence. Lots to work on. Screening, spacing, and work without the ball need to get better but love our trajectory. Lots of weapons. #GrowthMindset #GoBruins
1:33

Motion Shooting: last week of summer access for us quarter schools. Excited about our improvement and commitment to excellence. Lots to work on. Screening, spacing, and work without the ball need to get better but love our trajectory. Lots of weapons. #GrowthMindset #GoBruins

Cori Close

48,997 views • 2 years ago

Excited to share this demo video of Vincent! Check out the user-owned AI framework in action and get a glimpse of how you can take control of AI. Full blog post with more details in the replies! 👇
1:47

Excited to share this demo video of Vincent! Check out the user-owned AI framework in action and get a glimpse of how you can take control of AI. Full blog post with more details in the replies! 👇

David Sneider 🎆

16,716 views • 1 year ago

WE'RE IN A HIGH ALERT. SO MANY ABDUCTIONS HAPPENING TO YOUNG GIRLS AND WOMEN! REDMAN, LEGENDARY HIP HOP ARTIST IS SOUNDING THE ALARM. HE IS GIVING ADVICE FOR ALL OF YOU TO SHARE WITH YOUR CHILDREN AND FOR WOMEN AND MEN. THERE IS MORE THAN EPSTEIN HAPPENING! LOTS OF BAD GUYS HERE
5:21

WE'RE IN A HIGH ALERT. SO MANY ABDUCTIONS HAPPENING TO YOUNG GIRLS AND WOMEN! REDMAN, LEGENDARY HIP HOP ARTIST IS SOUNDING THE ALARM. HE IS GIVING ADVICE FOR ALL OF YOU TO SHARE WITH YOUR CHILDREN AND FOR WOMEN AND MEN. THERE IS MORE THAN EPSTEIN HAPPENING! LOTS OF BAD GUYS HERE

Peggy Dodson

451,777 views • 1 year ago

Our bodies’ molecules aren’t static so why do many AI models treat them that way? Without considering the physical movement of molecules, many AI tools fail to robustly capture what makes drug candidates promising. AquaGen is a generative AI framework built to close that gap: an all-atom, explicit-solvent model that runs an order of magnitude faster than traditional MD methods. We’re excited about the implications of this work on early-stage drug discovery. Technical report and blog below.
1:07

Our bodies’ molecules aren’t static so why do many AI models treat them that way? Without considering the physical movement of molecules, many AI tools fail to robustly capture what makes drug candidates promising. AquaGen is a generative AI framework built to close that gap: an all-atom, explicit-solvent model that runs an order of magnitude faster than traditional MD methods. We’re excited about the implications of this work on early-stage drug discovery. Technical report and blog below.

Valence Labs

10,379 views • 3 months ago

We're building a UI Engineering team at Shopify. If high quality design translation, motion, and implementation polish is what you're all about, DM Tasha A. to get the conversation started. More detail in the video 👇🏻
2:18

We're building a UI Engineering team at Shopify. If high quality design translation, motion, and implementation polish is what you're all about, DM Tasha A. to get the conversation started. More detail in the video 👇🏻

Derek Briggs

82,869 views • 8 months ago

We're excited to announce that the 1.0 release of shapez 2 is coming on April 23rd! With the Manufacture Mode, full modding support, achievements & major quality of life improvements based on your feedback, there's a ton to look forward to. ➡️ Check the replies for more info!
0:45

We're excited to announce that the 1.0 release of shapez 2 is coming on April 23rd! With the Manufacture Mode, full modding support, achievements & major quality of life improvements based on your feedback, there's a ton to look forward to. ➡️ Check the replies for more info!

tobspr Games | shapez 2 & puzzlez

15,438 views • 7 months ago