正在加载视频...

视频加载失败

加载此视频时出现问题。这可能是由于临时网络问题，或视频可能不可用。

3/ One emergent capability I find remarkable is long-term consistency, especially because we don’t use any explicit 3D representations or priors. Simply training the model to generate the next frame auto-regressively teaches it to maintain physical consistency across time

Agrim Gupta

5,616 subscribers

41,474 次观看 • 11 个月前 •via X (Twitter)

科学技术教育

Anya Rossi• Live Now

Private livecam show

0 条评论

暂无评论

原始帖子的评论将显示在这里

相关视频

Another use case with Nano Banana Pro (Video is fully narrated so don't forget to turn on the sound 🔊) Image to 3D Model For 3D Printing 1 - Generate an image of a vinyl toy character. I generated a character called Peelbert. 2 - Generate multiple outfits for Peelbert while retaining consistency. Nano Banana Pro's consistency and understanding of outfit changes is superb. 3 - Use Hunyuan 3.0 to do image to 3D model 4 - Save as STL and 3D Print. Nano Banana Pro to the real world 👍

Another use case with Nano Banana Pro (Video is fully narrated so don't forget to turn on the sound 🔊) Image to 3D Model For 3D Printing 1 - Generate an image of a vinyl toy character. I generated a character called Peelbert. 2 - Generate multiple outfits for Peelbert while retaining consistency. Nano Banana Pro's consistency and understanding of outfit changes is superb. 3 - Use Hunyuan 3.0 to do image to 3D model 4 - Save as STL and 3D Print. Nano Banana Pro to the real world 👍

Travis Davids

44,133 次观看 • 7 个月前

Do you find it difficult to discern God's voice or maintain consistency in the place of prayer? Watch this...

Do you find it difficult to discern God's voice or maintain consistency in the place of prayer? Watch this...

Naijasermon.com

111,930 次观看 • 1 年前

📢WorldAgents: 3D worlds only from 2D image models - without any training! We propose an agentic approach with a Director (VLM) to plan the scene, a Generator (Flux or NanoBanana) for new views, and a Verifier (VLM) for selection / 3D consistency. -> High-fidelity 3D worlds from a single text prompt. What's remarkable: our agents find consistent views from 2D image models to obtain 3D-consistent worlds; this shows that image models contain world priors - agents just need to find them! Great work by Ziya Erkoç Angela Dai

📢WorldAgents: 3D worlds only from 2D image models - without any training! We propose an agentic approach with a Director (VLM) to plan the scene, a Generator (Flux or NanoBanana) for new views, and a Verifier (VLM) for selection / 3D consistency. -> High-fidelity 3D worlds from a single text prompt. What's remarkable: our agents find consistent views from 2D image models to obtain 3D-consistent worlds; this shows that image models contain world priors - agents just need to find them! Great work by Ziya Erkoç Angela Dai

Matthias Niessner

18,967 次观看 • 3 个月前

🔘 Real-time capabilities Genie 3 is our first world model to allow live interaction, while also improving consistency and realism compared to Genie 2. It can generate dynamic worlds at 720p and 24 FPS, with each frame created in response to user actions.

🔘 Real-time capabilities Genie 3 is our first world model to allow live interaction, while also improving consistency and realism compared to Genie 2. It can generate dynamic worlds at 720p and 24 FPS, with each frame created in response to user actions.

Google DeepMind

224,441 次观看 • 11 个月前

The new Gen 4.5 update feels like an incredible unlock, especially for narrative work and consistency across scenes. Whenever I test new models internally, I find that the best way to evaluate performance is to use them to build a story. Here’s one I’ve always wanted to see brought to life.

The new Gen 4.5 update feels like an incredible unlock, especially for narrative work and consistency across scenes. Whenever I test new models internally, I find that the best way to evaluate performance is to use them to build a story. Here’s one I’ve always wanted to see brought to life.

Cristóbal Valenzuela

38,484 次观看 • 5 个月前

1/ Tencent releases HY World 1.5 (WorldPlay), an open-source real-time world model with 24fps A streaming video diffusion model that enables real-time, interactive world modeling with long-term geometric consistency In direct comparison to Genie 3, it's technically inferior, and there's no information on how long the consistency will last. However, Tencent has released a significant world model as open source for the first time, making it available to everyone. A few more examples, a thread 🧵

1/ Tencent releases HY World 1.5 (WorldPlay), an open-source real-time world model with 24fps A streaming video diffusion model that enables real-time, interactive world modeling with long-term geometric consistency In direct comparison to Genie 3, it's technically inferior, and there's no information on how long the consistency will last. However, Tencent has released a significant world model as open source for the first time, making it available to everyone. A few more examples, a thread 🧵

Chubby♨️

37,776 次观看 • 6 个月前

I think people underestimate how rare it is for a franchise to maintain The Last of Us level of storytelling consistency.

I think people underestimate how rare it is for a franchise to maintain The Last of Us level of storytelling consistency.

Nyx

10,853 次观看 • 2 个月前

3D editing is hard: you need to ground an image + instruction and generate a faithful 3D shape in one forward pass -- no test-time optimization. So, we steer pretrained image-to-3D representations to do text-guided 3D edits; no massive 3D edit-pair dataset needed. Key trap: the “no-edit” solution is a nasty local minimum. We fix it with preference optimization, pushing the model to actually edit. Steer3D is the second work that adapts alignment ideas from LLMs to the 3D modality. SAM 3D also used DPO to improve its 3D generations.

3D editing is hard: you need to ground an image + instruction and generate a faithful 3D shape in one forward pass -- no test-time optimization. So, we steer pretrained image-to-3D representations to do text-guided 3D edits; no massive 3D edit-pair dataset needed. Key trap: the “no-edit” solution is a nasty local minimum. We fix it with preference optimization, pushing the model to actually edit. Steer3D is the second work that adapts alignment ideas from LLMs to the 3D modality. SAM 3D also used DPO to improve its 3D generations.

Georgia Gkioxari

116,061 次观看 • 6 个月前

next time u use the term goatfield make sure u use it to refer to the right one

next time u use the term goatfield make sure u use it to refer to the right one

valentine

290,986 次观看 • 4 个月前

✨ 🖼️ Generate images with consistent characters without any fine tuning nor training. Consistory is a training-free approach to maintaining subject consistency between text-to-image generations on pretrained models. #NVIDIAResearch 🎨 Test it here:

✨ 🖼️ Generate images with consistent characters without any fine tuning nor training. Consistory is a training-free approach to maintaining subject consistency between text-to-image generations on pretrained models. #NVIDIAResearch 🎨 Test it here:

NVIDIA AI Developer

11,600 次观看 • 1 年前

Next AI short I am dropping is about : Character Consistency. To tell a powerful story, one of the key elements is to control the consistency of the character and the scenario. Using Runway reference, @HeyGen_Official and Adobe After Effects

Next AI short I am dropping is about : Character Consistency. To tell a powerful story, one of the key elements is to control the consistency of the character and the scenario. Using Runway reference, @HeyGen_Official and Adobe After Effects

Stelfie the Time Traveller

17,266 次观看 • 1 年前

We vibe coded this demo in Google AI Studio to show Nano Banana 2's real-world understanding. With each frame in this environment, the model sees only the previous image, and is prompted to imagine what happens next. The consistency is incredible!

We vibe coded this demo in Google AI Studio to show Nano Banana 2's real-world understanding. With each frame in this environment, the model sees only the previous image, and is prompted to imagine what happens next. The consistency is incredible!

Demis Hassabis

69,141 次观看 • 4 个月前

They solved environmental consistency with Genie 3, and this was an emergent capability. You can see the trees remain the same even after being out of line of sight. Visual memory extends back one minute now. Google is on a steady path to a real world simulator.

They solved environmental consistency with Genie 3, and this was an emergent capability. You can see the trees remain the same even after being out of line of sight. Visual memory extends back one minute now. Google is on a steady path to a real world simulator.

Andrew Curran

496,611 次观看 • 11 个月前

John Mozeliak says there's still plenty of time for the #STLCards' lineup to find some consistency: "The one thing I'd ask from our fan base right now is a little bit of patience, because we do think that we do have the talent to win."

John Mozeliak says there's still plenty of time for the #STLCards' lineup to find some consistency: "The one thing I'd ask from our fan base right now is a little bit of patience, because we do think that we do have the talent to win."

FanDuel Sports Network Midwest

674,845 次观看 • 3 年前

V3D Video Diffusion Models are Effective 3D Generators Automatic 3D generation has recently attracted widespread attention. Recent methods have greatly accelerated the generation speed, but usually produce less-detailed objects due to limited model capacity or 3D data. Motivated by recent advancements in video diffusion models, we introduce V3D, which leverages the world simulation capacity of pre-trained video diffusion models to facilitate 3D generation. To fully unleash the potential of video diffusion to perceive the 3D world, we further introduce geometrical consistency prior and extend the video diffusion model to a multi-view consistent 3D generator. Benefiting from this, the state-of-the-art video diffusion model could be fine-tuned to generate 360degree orbit frames surrounding an object given a single image. With our tailored reconstruction pipelines, we can generate high-quality meshes or 3D Gaussians within 3 minutes. Furthermore, our method can be extended to scene-level novel view synthesis, achieving precise control over the camera path with sparse input views. Extensive experiments demonstrate the superior performance of the proposed approach, especially in terms of generation quality and multi-view consistency

V3D Video Diffusion Models are Effective 3D Generators Automatic 3D generation has recently attracted widespread attention. Recent methods have greatly accelerated the generation speed, but usually produce less-detailed objects due to limited model capacity or 3D data. Motivated by recent advancements in video diffusion models, we introduce V3D, which leverages the world simulation capacity of pre-trained video diffusion models to facilitate 3D generation. To fully unleash the potential of video diffusion to perceive the 3D world, we further introduce geometrical consistency prior and extend the video diffusion model to a multi-view consistent 3D generator. Benefiting from this, the state-of-the-art video diffusion model could be fine-tuned to generate 360degree orbit frames surrounding an object given a single image. With our tailored reconstruction pipelines, we can generate high-quality meshes or 3D Gaussians within 3 minutes. Furthermore, our method can be extended to scene-level novel view synthesis, achieving precise control over the camera path with sparse input views. Extensive experiments demonstrate the superior performance of the proposed approach, especially in terms of generation quality and multi-view consistency

AK

31,997 次观看 • 2 年前

Consistency is key 🔐 We’ve been working behind the scenes to bring more consistency to the balls used on Tour – so players can rely on a more familiar feel across event swings. Find out what we’ve been up to. 🎾 #OneVision

Consistency is key 🔐 We’ve been working behind the scenes to bring more consistency to the balls used on Tour – so players can rely on a more familiar feel across event swings. Find out what we’ve been up to. 🎾 #OneVision

ATP Tour

48,707 次观看 • 1 年前

Clavicular explained the REAL PURPOSE of LOOKSMAXXING and how consistency is what actually leads to LONG-TERM RESULTS 😳👀 “Looksmaxxing is a thing that really takes a long time and it’s not linear. Certain days you’re going to look worse, certain days you’re going to look better.” “As long as there’s consistency in your routine... going to the gym, getting leaner over time...you’re going to see progress and that's really good.”

Clavicular explained the REAL PURPOSE of LOOKSMAXXING and how consistency is what actually leads to LONG-TERM RESULTS 😳👀 “Looksmaxxing is a thing that really takes a long time and it’s not linear. Certain days you’re going to look worse, certain days you’re going to look better.” “As long as there’s consistency in your routine... going to the gym, getting leaner over time...you’re going to see progress and that's really good.”

S6BER

37,663 次观看 • 3 个月前

Great work with the EdgeMen, always a blessing and privilege to work with these men, the attention to detail and encouragement amongst the group is one of a kind. The small details mean the most! #TrainingWithChief “Long term consistency beats short term intensity”

Great work with the EdgeMen, always a blessing and privilege to work with these men, the attention to detail and encouragement amongst the group is one of a kind. The small details mean the most! #TrainingWithChief “Long term consistency beats short term intensity”

Chief

23,767 次观看 • 5 个月前

Luke Shaw: "We’re winning one game and losing the next, we need to build on consistency." [Stan Sport Football]

Luke Shaw: "We’re winning one game and losing the next, we need to build on consistency." [Stan Sport Football]

UnitedReds

10,268 次观看 • 8 个月前