![FLUX.1 [dev] Medium shot. A llama laying in bed with a VR headset. The llama is under a cozy quilt blanket with the word "VR" in bold print. The background is a cozy bedroom, a window with snow outside, adding to the overall voyeuristic atmosphere.](https://ethanbholland.com/wp-content/uploads/2024/09/VR.png)
FLUX.1 [dev] Medium shot. A llama laying in bed with a VR headset. The llama is under a cozy quilt blanket with the word “VR” in bold print. The background is a cozy bedroom, a window with snow outside, adding to the overall voyeuristic atmosphere.
Apple Vision Pro
“The coolest use case of the Apple Vision Pro I’ve seen: Using it to operate a humanoid robot in real-time. According to NVIDIA researchers, from the human’s point of view, you feel ‘immersed’ in another body, like in Avatar. Here’s how it works: → Human operators use Apple” / X
“@dr_cintas NVIDIA’s Project GR00T introduced a new approach to scale robot data. It uses the Apple Vision Pro for teleoperation, RoboCasa for environment simulation, and MimicGen for motion, potentially revolutionizing data collection in robotics.
Neuralink rival Synchron offers thought control with Apple Vision Pro
“Neuralink rival Synchron’s brain implant now lets people control Apple’s Vision Pro with their minds Synchron announced on Tuesday it has connected its brain implant to the Apple Vision Pro headset in an industry first. The company is building a brain-computer interface that will
Meta
“Memory Attention: adding object permanence with $50k in compute @AIatMeta continues to lead Actually Open AI. SAM2 generalizes SAM1 from image segmentation to video, releasing task, model, and dataset as Apache 2/CC by 4.0! Notable aspects from reading the paper: – shockingly
Gaussian Splatting and Nerfs
“🚨GaussianEditor: Editing 3D Gaussians Delicately with Text Instructions [CVPR’24] 🌟𝐏𝐫𝐨𝐣:
“There is a zen-like quality to training radiance fields (like this 100 photo buddha scan) > Semi-structured imagery > Nicely posed with photogrammetry > Hitting train and seeing it materialize > Reality is now digitized
3D Modeling
“Cuebric launches Generative Mesh (Alpha). Instant 3D for world-building. Now you can export .usd, .fbx, and .obj and integrate creative AI into your favorite 3D software. We took the most precious commodity on earth, time, and gave it back to you. So you can accomplish far
“Unveiled at #SIGGRAPH2024: @WPP is the first to test & use new @NVIDIA NIM microservices to build gen 3D worlds for clients incl @CocaColaCo & @Ford 👇
“Game development is about to get easy. This is 3D animation powered by AI. No suits or markers needed. Liam Bailey used the Move One single-camera motion capture app with Unreal Engine 5 to bring this character to life.
Introducing Stable Fast 3D: Rapid 3D Asset Generation From Single Images — Stability AI
“Stability AI unveiled Stable Video 4D, its a new AI model that can turn single object videos into multiple videos from eight different angles Lots of potential applications for this, including game development, video editing, and virtual reality
Stability AI steps into a new gen AI dimension with Stable Video 4D | VentureBeat
Robot Training
“LiDAR SLAM is so flipping cool Two different 3D maps captured by robots with an Ouster LiDAR puck + IMU + wheel encoders being merged together into one
World Models
“✨Just announced: Representation agnostics physics simulation. ➡️
Other AR/VR News
Activision Releases Call of Duty®: Warzone™ Caldera Data Set for Academic Use
“In addition to the new model, we’re also releasing SA-V, a dataset that’s 4.5x larger + has ~53x more annotations than the largest existing video segmentation dataset. We hope this work will help accelerate new computer vision research ➡️
“Huge news. Meta just released Segment Anything 2, the most powerful video and image segmentation model. SAM 2 demonstrates significant performance improvements: ▸ Operates at 44 frames per second for video segmentation. ▸ Requires three times fewer interactions for video
“The new AI segment tool from Meta is pretty nifty. One click to select objects in moving scenes. Everything here is me playing with this real time.
SAM 2 Demo | By Meta FAIR
Our New AI Model Can Segment Anything – Even Video | Meta
Introducing SAM 2: The next generation of Meta Segment Anything Model for videos and images
“Introducing Meta Segment Anything Model 2 (SAM 2) — the first unified model for real-time, promptable object segmentation in images & videos. SAM 2 is available today under Apache 2.0 so that anyone can use it to build their own experiences Details ➡️
“Along with the Meta Segment Anything Model 2 (SAM 2), we also released SA-V: a dataset containing ~51K videos and >600K masklet annotations. We’re sharing this dataset with the hope that this work will help accelerate new computer vision research ➡️
“Meta coming in hot with SAM 2 Segment Anything Model (SAM) lets you do real-time promptable image and video segmentation it can do things like track objects to create video effects (left) or segment moving cells in videos captured from a microscope (right) link below
“Meta just released SAM 2, a new version of its video and image segmentation model. They also released a dataset of approximately 51K videos and 600K masklets (spatio-temporal masks) The code and weights are available under the Apache 2.0 license.” / X
“Meta introduced Segment Anything Model 2 (SAM 2) It’s an advanced AI model that can identify and track objects across video frames in real time. Editing tasks like object removal or replacement are going to be as simple as a single click shortly
“Virtual humans also travelling to #ECCV2024: 1) Luvizon et al. Relightable Neural Actor. tl;dr: Video-based @DiogoLuvizon’s avatar relighting. 2) Ghosh et al. ReMoS. tl;dr: Generative Lindy Hop and Ninjutsu!https://twitter.com/VGolyanik/status/1819260230980481431
Leave a Reply