Video yükleniyor...

Video Yüklenemedi

Bu video yüklenirken bir sorun oluştu. Bu geçici bir ağ sorunundan kaynaklanıyor olabilir veya video kullanılamıyor olabilir.

Ana Sayfaya Dön

Video generation is powerful but too slow for real-world robotic tasks. How can we enable both video and action generation while ensuring real-time policy inference? Check out our work on the Unified Video Action Model (UVA) to find out! (1/7)

Shuang Li

8,290 subscribers

68,008 görüntüleme • 1 yıl önce •via X (Twitter)

Eğitim Bilim & Teknoloji

Anya Rossi• Live Now

Private livecam show

0 Yorum

Yorum bulunmuyor

Orijinal gönderinin yorumları burada görünecek

Benzer Videolar

The Sora 2 feed is fascinating, seeing how everyone is using it for the first time. This generation stood out a mile from all the cameos and pokemon, by jesperagain: > 1960s black-and-white BBC report on Sora 2 video generation model launch. Grainy video to match the time

The Sora 2 feed is fascinating, seeing how everyone is using it for the first time. This generation stood out a mile from all the cameos and pokemon, by jesperagain: > 1960s black-and-white BBC report on Sora 2 video generation model launch. Grainy video to match the time

fofr

283,743 görüntüleme • 1 yıl önce

$0.003/sec changes the economics of running real-time AI video. I’ve been testing Vivix W1, and that number is hard to ignore. The comparison shows visual quality on par with the models alongside W1. About a second of generation time per second of footage, too. But W1 isn’t just another video generator. It’s built for real-time interactive video, where the model keeps generating as the experience moves forward. Every second of that experience adds to the generation bill. At $0.003/sec, you can actually start thinking about a story that keeps going as viewers make choices. Vivix.AI already supports real-time generation through the API. The browser playground currently uses offline generation, with real-time browser tools coming soon. Continuous video is one thing. Giving viewers meaningful control is another. If I change my mind halfway through a story, how much can it actually change with me? That’s the interesting challenge. The next plot twist could be yours 👇

$0.003/sec changes the economics of running real-time AI video. I’ve been testing Vivix W1, and that number is hard to ignore. The comparison shows visual quality on par with the models alongside W1. About a second of generation time per second of footage, too. But W1 isn’t just another video generator. It’s built for real-time interactive video, where the model keeps generating as the experience moves forward. Every second of that experience adds to the generation bill. At $0.003/sec, you can actually start thinking about a story that keeps going as viewers make choices. Vivix.AI already supports real-time generation through the API. The browser playground currently uses offline generation, with real-time browser tools coming soon. Continuous video is one thing. Giving viewers meaningful control is another. If I change my mind halfway through a story, how much can it actually change with me? That’s the interesting challenge. The next plot twist could be yours 👇

AI StackLab

221,993 görüntüleme • 4 gün önce

🔥Spatial intelligence requires world generation, and now we have the first comprehensive evaluation benchmark📏 for it! Introducing WorldScore: Unifying evaluation for 3D, 4D, and video models on world generation! 🧵1/7 Web: arxiv:

🔥Spatial intelligence requires world generation, and now we have the first comprehensive evaluation benchmark📏 for it! Introducing WorldScore: Unifying evaluation for 3D, 4D, and video models on world generation! 🧵1/7 Web: arxiv:

Hong-Xing (Koven) Yu

66,876 görüntüleme • 1 yıl önce

🏆 Cosmos 3 just topped 7 physical AI leaderboards. NVIDIA Cosmos™ 3, the open omni-model for physical AI, ranks #1 across world generation, robot action policy, and industrial vision understanding. 🌎 World generation: Artificial Analysis, PAI-Bench, Physics-IQ, R-Bench 🤖 Robot policy: RoboLab 👁️ Vision: VANTAGE-Bench, TAR One model. Every domain of physical AI. Available now on Hugging Face ▶️

🏆 Cosmos 3 just topped 7 physical AI leaderboards. NVIDIA Cosmos™ 3, the open omni-model for physical AI, ranks #1 across world generation, robot action policy, and industrial vision understanding. 🌎 World generation: Artificial Analysis, PAI-Bench, Physics-IQ, R-Bench 🤖 Robot policy: RoboLab 👁️ Vision: VANTAGE-Bench, TAR One model. Every domain of physical AI. Available now on Hugging Face ▶️

NVIDIA

59,237 görüntüleme • 4 ay önce

Brand system in video generation I created this video using 1 brand guide, 1 brand Skill, and 1 video model AI is now turning product videos into commodity, so brands can capture attention more effectively However, harnessing video generation isn't straightforward without specifically preparing your video model to generate brand-locked content I outline this step-by-step process in the Article below - read on and save it for later:

AmirMušić

294,957 görüntüleme • 1 ay önce

Seedance 2.0 is going to be a game changer for video creation. The powerful video generation tool empowers creativity and drives efficiency. Built for social content, product ads, and marketing at scale. Get 60% OFF on Pollo AI! Time is running out. Don't miss it!

Seedance 2.0 is going to be a game changer for video creation. The powerful video generation tool empowers creativity and drives efficiency. Built for social content, product ads, and marketing at scale. Get 60% OFF on Pollo AI! Time is running out. Don't miss it!

Pollo AI

16,533,552 görüntüleme • 6 ay önce

🤖 NVIDIA’s Gr00t N1.5 is now available in LeRobot! This is the result of a great collaboration between the Hugging Face LeRobot team and NVIDIA Robotics ! Gr00t N1.5 highlights: 🦾 Cross-embodiment foundation model for robots 🧠 Multimodal inputs: vision, language, and proprioception 🪛Tested on the Libero benchmark and real-world hardware tasks 🌍Trained on real robot, synthetic, and internet-scale video data ⚙️ Flow matching action transformer for action prediction

🤖 NVIDIA’s Gr00t N1.5 is now available in LeRobot! This is the result of a great collaboration between the Hugging Face LeRobot team and NVIDIA Robotics ! Gr00t N1.5 highlights: 🦾 Cross-embodiment foundation model for robots 🧠 Multimodal inputs: vision, language, and proprioception 🪛Tested on the Libero benchmark and real-world hardware tasks 🌍Trained on real robot, synthetic, and internet-scale video data ⚙️ Flow matching action transformer for action prediction

LeRobot

115,194 görüntüleme • 11 ay önce

We are rolling out access to Runway Aleph, our new in-context video model, setting a new frontier for multi-task visual generation. Aleph has the ability to perform a wide range of edits and generations on existing videos. Here are a few of the things Aleph is capable of doing. (1/7)

We are rolling out access to Runway Aleph, our new in-context video model, setting a new frontier for multi-task visual generation. Aleph has the ability to perform a wide range of edits and generations on existing videos. Here are a few of the things Aleph is capable of doing. (1/7)

Runway

99,594 görüntüleme • 1 yıl önce

ICYM ⚠️ Check out the live and upcoming features on our 12th Development Update video! ⚙️ You can also find the transcript to this video here:

ICYM ⚠️ Check out the live and upcoming features on our 12th Development Update video! ⚙️ You can also find the transcript to this video here:

Mount & Blade

14,621 görüntüleme • 3 yıl önce

introducing FLUX 3. beyond video, this is the first model to support interactions with the real world through a new action-prediction system. available now on Krea, more modalities coming soon.

introducing FLUX 3. beyond video, this is the first model to support interactions with the real world through a new action-prediction system. available now on Krea, more modalities coming soon.

Krea

32,973 görüntüleme • 2 ay önce

Cool new work on real-time video generation with motion control. MotionStream enables sub-second latency streaming on a single GPU. You can also control what parts of an image do & don't move, like the video below. 👏 to Xun Huang, Joonghyuk Shin, and team!

Cool new work on real-time video generation with motion control. MotionStream enables sub-second latency streaming on a single GPU. You can also control what parts of an image do & don't move, like the video below. 👏 to Xun Huang, Joonghyuk Shin, and team!

Justine Moore

21,862 görüntüleme • 11 ay önce

Together with NVIDIA , we've built a real-time version of our video generation model: 30B parameters, steerable mid-stream in natural language, 720p at 24fps. 11.8x faster on a single H100, no major quality drop in our blind eval. We're one step closer to World Language Action Models: AI that doesn't just generate the physical world, it acts within it. Closed beta open now, contact us for access. #RekaLabs #NVIDIA #AIResearch #VideoGeneration #WorldModels

Together with NVIDIA , we've built a real-time version of our video generation model: 30B parameters, steerable mid-stream in natural language, 720p at 24fps. 11.8x faster on a single H100, no major quality drop in our blind eval. We're one step closer to World Language Action Models: AI that doesn't just generate the physical world, it acts within it. Closed beta open now, contact us for access. #RekaLabs #NVIDIA #AIResearch #VideoGeneration #WorldModels

Reka

11,659 görüntüleme • 1 ay önce

. Runway just added a new feature that you may not have noticed and we wanted to compare. You can now upscale your generation to 4K! Our favorite has typically been Topaz Labs, so we wanted to compare the out of both. Here is the original video from Runway

. Runway just added a new feature that you may not have noticed and we wanted to compare. You can now upscale your generation to 4K! Our favorite has typically been Topaz Labs, so we wanted to compare the out of both. Here is the original video from Runway

Curious Refuge

33,320 görüntüleme • 1 yıl önce

World modeling and imitation learning have largely been considered two disparate worlds. In our recent work, Unified World Models, just accepted to #RSS2025, Chuning Zhu provides a dead-simple unifying solution: just train a joint diffusion model over actions and future states, but with *decoupled* diffusion time steps across these modalities. Manipulating these decoupled time steps then allows for marginalization or conditioning on actions or states; a single model can serve as a policy, forward dynamics model, video prediction model, or inverse dynamics model by simply setting diffusion timesteps carefully. The resulting model can leverage video datasets along with robot training data much more effectively, and shows improved robustness, generalization, and flexibility. This is exciting because it is frustratingly simple, scalable, and shows strong improvement on real-world robotics problems. Please refer to Chuning Zhu 's excellent thread for more details! More details/code can be found on our website and in the paper -

World modeling and imitation learning have largely been considered two disparate worlds. In our recent work, Unified World Models, just accepted to #RSS2025, Chuning Zhu provides a dead-simple unifying solution: just train a joint diffusion model over actions and future states, but with decoupled diffusion time steps across these modalities. Manipulating these decoupled time steps then allows for marginalization or conditioning on actions or states; a single model can serve as a policy, forward dynamics model, video prediction model, or inverse dynamics model by simply setting diffusion timesteps carefully. The resulting model can leverage video datasets along with robot training data much more effectively, and shows improved robustness, generalization, and flexibility. This is exciting because it is frustratingly simple, scalable, and shows strong improvement on real-world robotics problems. Please refer to Chuning Zhu 's excellent thread for more details! More details/code can be found on our website and in the paper -

Abhishek Gupta

11,430 görüntüleme • 1 yıl önce

MolmoAct2 is landing in LeRobot! Ai2's open Action Reasoning Model combines a Molmo2-ER vision-language backbone with a flow-matching continuous action expert to predict robot action chunks from images, language instructions, and proprioceptive state. An open robot foundation model built for real-world control, with strong out-of-the-box performance and easy fine-tuning in LeRobot. Pick-and-place inference running on NVIDIA DGX Spark! Blog: Paper: Thanks to Ai2 Jiafei Duan Haoquan Fang

MolmoAct2 is landing in LeRobot! Ai2's open Action Reasoning Model combines a Molmo2-ER vision-language backbone with a flow-matching continuous action expert to predict robot action chunks from images, language instructions, and proprioceptive state. An open robot foundation model built for real-world control, with strong out-of-the-box performance and easy fine-tuning in LeRobot. Pick-and-place inference running on NVIDIA DGX Spark! Blog: Paper: Thanks to Ai2 Jiafei Duan Haoquan Fang

LeRobot

25,163 görüntüleme • 4 ay önce

$0.003 per second for video generation, without compromising on speed or visual quality? The supplied comparisons put Vivix W1 alongside leading models on visual quality. It also delivers eight seconds of video in roughly eight seconds. But the bigger story is that W1 is a real-time interactive video model. Every extra minute of interaction means more video to generate and pay for. Making generation this affordable could make interactive stories and live visuals practical to run, not just impressive demos. Live visuals are what I’m most curious about. Generating quickly is one thing; reacting at the right moment is what would make them feel live. Vivix.AI already offers real-time generation through the API; the browser playground currently produces finished clips. A closer look at what could power those live visuals 👇

$0.003 per second for video generation, without compromising on speed or visual quality? The supplied comparisons put Vivix W1 alongside leading models on visual quality. It also delivers eight seconds of video in roughly eight seconds. But the bigger story is that W1 is a real-time interactive video model. Every extra minute of interaction means more video to generate and pay for. Making generation this affordable could make interactive stories and live visuals practical to run, not just impressive demos. Live visuals are what I’m most curious about. Generating quickly is one thing; reacting at the right moment is what would make them feel live. Vivix.AI already offers real-time generation through the API; the browser playground currently produces finished clips. A closer look at what could power those live visuals 👇

Kalsoom (ghotai )

71,646 görüntüleme • 3 gün önce

🤩Video world models are cool, but it is cooler if they can simulate any 3D physical actions in real time! Introducing RealWonder⚡️: Now you can simulate 3D physical action (robot actions, 3D forces, force fields, etc.) consequences from a single image in real time! 🧵1/6

🤩Video world models are cool, but it is cooler if they can simulate any 3D physical actions in real time! Introducing RealWonder⚡️: Now you can simulate 3D physical action (robot actions, 3D forces, force fields, etc.) consequences from a single image in real time! 🧵1/6

Hong-Xing (Koven) Yu

30,524 görüntüleme • 7 ay önce

We’re amazed: AMD now beats Nvidia in inference for Image-to-Video generation. 20% faster and 35% cheaper when using our Higgsfield DoP model. AMD MI300X also has a very easy setup. Details and tests are below. 🧩 1/8

We’re amazed: AMD now beats Nvidia in inference for Image-to-Video generation. 20% faster and 35% cheaper when using our Higgsfield DoP model. AMD MI300X also has a very easy setup. Details and tests are below. 🧩 1/8

Higgsfield AI 🧩

842,385 görüntüleme • 1 yıl önce