Video wird geladen...

Video konnte nicht geladen werden

Zur Startseite

interpolating image embeddings across arbitrary numbers of vertices. using polygons as lenses through n-dimensional latent space.

hturan's profile picture

harley turan

12,956 subscribers

23,143 Aufrufe • vor 1 Jahr •via X (Twitter)

LIVE

Anya Rossi• Live Now

Private livecam show

0 Kommentare

Keine Kommentare verfügbar

Kommentare vom Original-Post werden hier angezeigt

Ähnliche Videos

exploring paths through latent space by interpolating between known embeddings
1:02

exploring paths through latent space by interpolating between known embeddings

harley turan

96,479 Aufrufe • vor 1 Jahr

Shadows of The Shining Exploring the latent space of The Shining is to approach the film as one approaches a memory. Here, artificial intelligence acts as a second consciousness: it does not reproduce — it remembers. It reconstructs The Shining from its traces, its echoes, from what has survived the narrative. The latent space becomes a mental territory, an inner Overlook where the ghosts of the image, the logic of dreams, and the mechanisms of visual memory unfold once again. The aim is to turn the exploration of the latent space into a mnemonic experience of cinema itself.
0:45

Shadows of The Shining Exploring the latent space of The Shining is to approach the film as one approaches a memory. Here, artificial intelligence acts as a second consciousness: it does not reproduce — it remembers. It reconstructs The Shining from its traces, its echoes, from what has survived the narrative. The latent space becomes a mental territory, an inner Overlook where the ghosts of the image, the logic of dreams, and the mechanisms of visual memory unfold once again. The aim is to turn the exploration of the latent space into a mnemonic experience of cinema itself.

Benjamin Bardou

37,527 Aufrufe • vor 10 Monaten

ColPali is changing the game for PDF retrieval by eliminating the need for OCR and chunking methods 🚀 Inspired by ColBERT’s success with text, ColPali splits an image of a document into patches, which are then processed through a vision LLM called PaliGemma. The embeddings for each patch retain contextual information, similarly to text embeddings in methods like ColBERT. During retrieval, user queries are embedded in the same space, and then compared to document patches using the MaxSim operator. ColPali recipe POC Weaviate AI Database: ColPali paper: As always, shoutout to the awesome Daniel Williams for the help! 💙
2:50

ColPali is changing the game for PDF retrieval by eliminating the need for OCR and chunking methods 🚀 Inspired by ColBERT’s success with text, ColPali splits an image of a document into patches, which are then processed through a vision LLM called PaliGemma. The embeddings for each patch retain contextual information, similarly to text embeddings in methods like ColBERT. During retrieval, user queries are embedded in the same space, and then compared to document patches using the MaxSim operator. ColPali recipe POC Weaviate AI Database: ColPali paper: As always, shoutout to the awesome Daniel Williams for the help! 💙

Victoria Slocum

67,862 Aufrufe • vor 1 Jahr

"Alpha" by 𒐪 upgrades the opepen protocol, introducing the possibility of interactive 3-dimensional works—an infinite expansion of the canvas into navigable nothingness. For the first time, the opepen form transcends its two-dimensional bounds, rendered in a .glb environment: across edition sizes, the density of nothingness shifts. Placeholder textures become the commentary: nothingness as form, space as message, and the new dimensions unlocked when the protocol expands both conceptually and mechanically.
0:30

"Alpha" by 𒐪 upgrades the opepen protocol, introducing the possibility of interactive 3-dimensional works—an infinite expansion of the canvas into navigable nothingness. For the first time, the opepen form transcends its two-dimensional bounds, rendered in a .glb environment: across edition sizes, the density of nothingness shifts. Placeholder textures become the commentary: nothingness as form, space as message, and the new dimensions unlocked when the protocol expands both conceptually and mechanically.

jack

42,362 Aufrufe • vor 1 Jahr

Stunning timelapse footage captured from a SpaceX Dragon spacecraft shows brilliant green ribbons of the Aurora Australis dancing across Earth’s atmosphere as the capsule traveled through space. 🎥: Stellarix
0:59

Stunning timelapse footage captured from a SpaceX Dragon spacecraft shows brilliant green ribbons of the Aurora Australis dancing across Earth’s atmosphere as the capsule traveled through space. 🎥: Stellarix

Breaking911

29,402 Aufrufe • vor 2 Monaten

Two views of the Lion Nebula 🦁🦁 Observing across the starry ‘plains’ of space, Webb has captured these new views of NGC 2392, the Lion Nebula. The first view here shows the planetary nebula as seen by the #MIRI on Webb. The second image combines this mid-infrared data with near-infrared data from the #NIRCam instrument. #WebbSeesFarther 📷 NASA, European Space Agency, Canadian Space Agency, STScI, A. Pagan (STScI), N. Bartmann (ESA/Webb) 🎶 Stellardrone - The Night Sky in Motion
0:30

Two views of the Lion Nebula 🦁🦁 Observing across the starry ‘plains’ of space, Webb has captured these new views of NGC 2392, the Lion Nebula. The first view here shows the planetary nebula as seen by the #MIRI on Webb. The second image combines this mid-infrared data with near-infrared data from the #NIRCam instrument. #WebbSeesFarther 📷 NASA, European Space Agency, Canadian Space Agency, STScI, A. Pagan (STScI), N. Bartmann (ESA/Webb) 🎶 Stellardrone - The Night Sky in Motion

ESA Webb Telescope

17,566 Aufrufe • vor 19 Tagen

“Pointillisme,” an immersive installation by Japanese designer Taiju Yamashita, created using thousands of transparent suspended spheres arranged at precise intervals. Paired with mirrored surfaces and controlled lighting, the installation creates shifting patterns of depth, reflection, and perspective as viewers move through the space. The work was presented at Milan’s Salone del Mobile as part of DAFT about DRAFT’s exhibition space.
0:27

“Pointillisme,” an immersive installation by Japanese designer Taiju Yamashita, created using thousands of transparent suspended spheres arranged at precise intervals. Paired with mirrored surfaces and controlled lighting, the installation creates shifting patterns of depth, reflection, and perspective as viewers move through the space. The work was presented at Milan’s Salone del Mobile as part of DAFT about DRAFT’s exhibition space.

Interesting World

94,973 Aufrufe • vor 3 Monaten

Not everything is as it seems 🤖 Evolving #GenerativeAI enables scammers to create ultra-realistic human avatars using voice, image, and text manipulation. #ThinkTwice: Verify identities through multiple channels and be cautious of requests, even from "familiar" faces.
0:36

Not everything is as it seems 🤖 Evolving #GenerativeAI enables scammers to create ultra-realistic human avatars using voice, image, and text manipulation. #ThinkTwice: Verify identities through multiple channels and be cautious of requests, even from "familiar" faces.

INTERPOL

25,493 Aufrufe • vor 1 Jahr

New short course Multimodal RAG: Chat with Videos, developed with Intel and taught by vasudevlal! In this course, you’ll work with LLaVA (Large Language and Vision Assistant), a Large Vision Language Model (LVLM) that can process both images and text. For example, given an image of a person doing a handstand on a skateboard at the beach, LLaVA doesn't just caption the scene, it’s able to predict possible outcomes, like the person losing balance or falling off. By understanding not just what's in a video frame, but what might happen next, your application can provide more insightful answers to questions about video. You'll build a full multimodal RAG pipeline that can chat about video content: - Use the BridgeTower model to create joint text-image embeddings in a 512-dimensional multimodal semantic space. - Learn video processing techniques to extract keyframes, generate transcripts using Whisper, and create captions. - Use the LanceDB vector database to store and retrieve high-dimensional multimodal embeddings. - Integrate the LLaVA model, combining CLIP's (Contrastive Language Image Pretraining) vision transformer with Llama, for advanced visual-textual reasoning. Your final system will ingest video data, generate embeddings for frames and text, perform similarity searches for relevant content, and use the retrieved multimodal context to inform LVLM-based response generation. The result is a system capable of answering nuanced questions about video content, effectively chatting about the video it has processed. Please sign up here!
2:40

New short course Multimodal RAG: Chat with Videos, developed with Intel and taught by vasudevlal! In this course, you’ll work with LLaVA (Large Language and Vision Assistant), a Large Vision Language Model (LVLM) that can process both images and text. For example, given an image of a person doing a handstand on a skateboard at the beach, LLaVA doesn't just caption the scene, it’s able to predict possible outcomes, like the person losing balance or falling off. By understanding not just what's in a video frame, but what might happen next, your application can provide more insightful answers to questions about video. You'll build a full multimodal RAG pipeline that can chat about video content: - Use the BridgeTower model to create joint text-image embeddings in a 512-dimensional multimodal semantic space. - Learn video processing techniques to extract keyframes, generate transcripts using Whisper, and create captions. - Use the LanceDB vector database to store and retrieve high-dimensional multimodal embeddings. - Integrate the LLaVA model, combining CLIP's (Contrastive Language Image Pretraining) vision transformer with Llama, for advanced visual-textual reasoning. Your final system will ingest video data, generate embeddings for frames and text, perform similarity searches for relevant content, and use the retrieved multimodal context to inform LVLM-based response generation. The result is a system capable of answering nuanced questions about video content, effectively chatting about the video it has processed. Please sign up here!

Andrew Ng

107,967 Aufrufe • vor 2 Jahren

NGC 4654, which jostles among many other galaxies in the Virgo Cluster, is a curious spiral galaxy. One side is rounded and the opposite side is stretched into a long trail of gas (beyond this image). This is caused by ram pressure stripping. As NGC 4654 speeds through the cluster 💨 it rams through the intracluster medium, which squashes its leading edge and tugs at its gas, stretching it out across space! 1/2
0:30

NGC 4654, which jostles among many other galaxies in the Virgo Cluster, is a curious spiral galaxy. One side is rounded and the opposite side is stretched into a long trail of gas (beyond this image). This is caused by ram pressure stripping. As NGC 4654 speeds through the cluster 💨 it rams through the intracluster medium, which squashes its leading edge and tugs at its gas, stretching it out across space! 1/2

HUBBLE

31,247 Aufrufe • vor 1 Monat

Get a glimpse of the most powerful and impactful stories and themes of 2023 across News, Sports, Entertainment, Archive and Creative through Getty Images’ 2023 Year In Review! As world events unfolded, Getty Images’ award‑winning photographers, videographers and editors documented moving stories from thousands of events across the globe. From conflicts and devastating climate disasters to record-breaking crowds and audiences in stadiums and theaters worldwide and industry voices championing equity and fair representation. In the creative space, Getty Images’ global content creators embraced authenticity and innovation, producing genuine human connection, as well as the blending of creativity and technology through responsible AI image generation. View more, here: 📷: Getty Images
0:51

Get a glimpse of the most powerful and impactful stories and themes of 2023 across News, Sports, Entertainment, Archive and Creative through Getty Images’ 2023 Year In Review! As world events unfolded, Getty Images’ award‑winning photographers, videographers and editors documented moving stories from thousands of events across the globe. From conflicts and devastating climate disasters to record-breaking crowds and audiences in stadiums and theaters worldwide and industry voices championing equity and fair representation. In the creative space, Getty Images’ global content creators embraced authenticity and innovation, producing genuine human connection, as well as the blending of creativity and technology through responsible AI image generation. View more, here: 📷: Getty Images

Getty Images

22,141 Aufrufe • vor 2 Jahren

Inversion is hiring across ALL disciplines. Including: Accurate Landing As part of the Accurate Landing team, you’ll help Arc after its return from space as it steers through the atmosphere and lands within feet of its target. In short: help us build the most capable skydiver ever. Explore open roles:
0:28

Inversion is hiring across ALL disciplines. Including: Accurate Landing As part of the Accurate Landing team, you’ll help Arc after its return from space as it steers through the atmosphere and lands within feet of its target. In short: help us build the most capable skydiver ever. Explore open roles:

Inversion

25,593 Aufrufe • vor 2 Monaten

There is a path for Elon Musk to become the world’s first trillionaire by the end of 2026. It involves embracing energy-based money that follows the laws of thermodynamics (bitcoin) and using it as collateral to accelerate investment across AI, robotics, and space exploration.
2:50

There is a path for Elon Musk to become the world’s first trillionaire by the end of 2026. It involves embracing energy-based money that follows the laws of thermodynamics (bitcoin) and using it as collateral to accelerate investment across AI, robotics, and space exploration.

Joe Burnett, MSBA

21,867 Aufrufe • vor 8 Monaten

Next mlx-vlm release will ship with continuous batching support on the server 🚀 What's coming: → Continuous batching — new requests join the active batch immediately, no waiting. Mixed image + text batches supported → OpenAI-compatible API — field-for-field match with mlx-lm, reasoning/content split for thinking models, tag-aware streaming → Multi-turn tool calling — full tool use support across streaming and non-streaming, works with Gemma4 and other templates → Vision feature caching — cache image embeddings across turns. Gemma4: 228x speedup, Qwen3.5: 23x on cache hit All running locally on Apple Silicon. Check our this demo running 4 concurrent requests (mixed image + text) to gemma-4-26B-A4B-IT by Google Gemma in bf16 using Pi + MLX-VLM server on my M3 Ultra. One of the requests ingests a 8K resolution image!
0:52

Next mlx-vlm release will ship with continuous batching support on the server 🚀 What's coming: → Continuous batching — new requests join the active batch immediately, no waiting. Mixed image + text batches supported → OpenAI-compatible API — field-for-field match with mlx-lm, reasoning/content split for thinking models, tag-aware streaming → Multi-turn tool calling — full tool use support across streaming and non-streaming, works with Gemma4 and other templates → Vision feature caching — cache image embeddings across turns. Gemma4: 228x speedup, Qwen3.5: 23x on cache hit All running locally on Apple Silicon. Check our this demo running 4 concurrent requests (mixed image + text) to gemma-4-26B-A4B-IT by Google Gemma in bf16 using Pi + MLX-VLM server on my M3 Ultra. One of the requests ingests a 8K resolution image!

Prince Canuma

82,349 Aufrufe • vor 4 Monaten

Guangdong's commercial space industry is reaching for the stars! Since June 27, the "Beacon Over Lingnan" video series has been lighting up major online platforms. Episode 2 features an exclusive dialogue with Yang Yiqiang, Chairman of CAS Space—a company that symbolizes the rapid rise of China's commercial space sector. What's the vision? What's next? Yang shares his thoughts in this must-watch conversation. Season 1 includes 20 episodes across four themes: economy, culture, science & education, and people's livelihood. Twenty guests—from innovators to leaders—tell the stories of Guangdong, the GBA, and China through their unique lenses. Tune in and discover the beacon of Lingnan!
9:03

Guangdong's commercial space industry is reaching for the stars! Since June 27, the "Beacon Over Lingnan" video series has been lighting up major online platforms. Episode 2 features an exclusive dialogue with Yang Yiqiang, Chairman of CAS Space—a company that symbolizes the rapid rise of China's commercial space sector. What's the vision? What's next? Yang shares his thoughts in this must-watch conversation. Season 1 includes 20 episodes across four themes: economy, culture, science & education, and people's livelihood. Twenty guests—from innovators to leaders—tell the stories of Guangdong, the GBA, and China through their unique lenses. Tune in and discover the beacon of Lingnan!

Hi,GBA

112,036 Aufrufe • vor 2 Monaten

What Are These Mysterious Orbs That Turn Up When People Meditate And Ask For Something To Reveal Itself? Higher-dimensional entities, due to their access to a broader spectrum of dimensions, could perceive the quantum states of our consciousness as we perceive visual or auditory stimuli. The meditative state, coupled with focused intention, might create a coherent quantum signal that acts as a beacon into higher-dimensional space (Lamp Vs Laser - MIAS Gateway) This signal could then be interpreted by these entities as an invitation or call, prompting them to respond. The response could take the form of an energy manifestation, using quantum principles to create visual phenomena (Orb Experiences). These entities, understanding the limitations of our perceptual capabilities, may use orbs as a way to make their presence known, while still operating within the bounds of their dimensional realities. The orbs themselves, being self-contained with non-scattering light, might represent a controlled, localized disturbance in the quantum field, a kind of visual “echo” of higher-dimensional interactions.
1:36

What Are These Mysterious Orbs That Turn Up When People Meditate And Ask For Something To Reveal Itself? Higher-dimensional entities, due to their access to a broader spectrum of dimensions, could perceive the quantum states of our consciousness as we perceive visual or auditory stimuli. The meditative state, coupled with focused intention, might create a coherent quantum signal that acts as a beacon into higher-dimensional space (Lamp Vs Laser - MIAS Gateway) This signal could then be interpreted by these entities as an invitation or call, prompting them to respond. The response could take the form of an energy manifestation, using quantum principles to create visual phenomena (Orb Experiences). These entities, understanding the limitations of our perceptual capabilities, may use orbs as a way to make their presence known, while still operating within the bounds of their dimensional realities. The orbs themselves, being self-contained with non-scattering light, might represent a controlled, localized disturbance in the quantum field, a kind of visual “echo” of higher-dimensional interactions.

Jay Anderson

65,400 Aufrufe • vor 2 Jahren

Pitching has been outstanding across Kansas baseball this spring. ⚾️ Every class has difference-makers on the mound right now, power arms, competitors in big moments and guys putting together dominant seasons as postseason play gets closer. Here’s a look at some pitchers with great numbers to watch across Kansas baseball from 2-1A through 6A. Who else joins the conversation? Stats as of May 5th. #sportsinkansas
0:46

Pitching has been outstanding across Kansas baseball this spring. ⚾️ Every class has difference-makers on the mound right now, power arms, competitors in big moments and guys putting together dominant seasons as postseason play gets closer. Here’s a look at some pitchers with great numbers to watch across Kansas baseball from 2-1A through 6A. Who else joins the conversation? Stats as of May 5th. #sportsinkansas

Sports in Kansas

13,414 Aufrufe • vor 3 Monaten

[NeurIPS '24] DreamMesh4D: Video-to-4D Generation with Sparse-Controlled Gaussian-Mesh Hybrid Representation Abstract (excerpt) We introduce DreamMesh4D, a novel framework that combines mesh representation with sparse-controlled deformation technique to generate high-quality 4D object from a monocular video. To overcome the limitation of classical texture representation, we bind Gaussian splats to the surface of the triangular mesh for differentiable optimization of both the texture and mesh vertices. In particular, DreamMesh4D begins with a coarse mesh provided by a single image based 3D generation method. Sparse points are then uniformly sampled across the surface of the mesh, and are used to build a deformation graph to drive the motion of the 3D object for the sake of computational efficiency and providing additional constraint. For each step, transformations of sparse control points are predicted using a deformation network, and the mesh vertices as well as the bound surface Gaussians are deformed via a geometric skinning algorithm. The skinning algorithm is a hybrid approach combining LBS (linear blending skinning) and DQS (dual-quaternion skinning), mitigating drawbacks associated with both approaches. The static surface Gaussians and mesh vertices as well as the dynamic deformation network are learned via reference view photometric loss, score distillation loss as well as other regularization losses in a two-stage manner. Extensive experiments demonstrate that our method outperforms prior video-to-4D generation methods in terms of rendering quality and spatial-temporal consistency.
0:20

[NeurIPS '24] DreamMesh4D: Video-to-4D Generation with Sparse-Controlled Gaussian-Mesh Hybrid Representation Abstract (excerpt) We introduce DreamMesh4D, a novel framework that combines mesh representation with sparse-controlled deformation technique to generate high-quality 4D object from a monocular video. To overcome the limitation of classical texture representation, we bind Gaussian splats to the surface of the triangular mesh for differentiable optimization of both the texture and mesh vertices. In particular, DreamMesh4D begins with a coarse mesh provided by a single image based 3D generation method. Sparse points are then uniformly sampled across the surface of the mesh, and are used to build a deformation graph to drive the motion of the 3D object for the sake of computational efficiency and providing additional constraint. For each step, transformations of sparse control points are predicted using a deformation network, and the mesh vertices as well as the bound surface Gaussians are deformed via a geometric skinning algorithm. The skinning algorithm is a hybrid approach combining LBS (linear blending skinning) and DQS (dual-quaternion skinning), mitigating drawbacks associated with both approaches. The static surface Gaussians and mesh vertices as well as the dynamic deformation network are learned via reference view photometric loss, score distillation loss as well as other regularization losses in a two-stage manner. Extensive experiments demonstrate that our method outperforms prior video-to-4D generation methods in terms of rendering quality and spatial-temporal consistency.

MrNeRF

12,323 Aufrufe • vor 1 Jahr