Artificial Phil Collins sings the Siggraph 2022 papers
Artificial Phil Collins sings the Siggraph 2022 papers
Also: Windowlicker, but everything is the voice of a Youtuber
Siggraph 2022 Papers
Here’s the annual SIGGRAPH Technical Papers Preview for 2022. I went through Ke-Sen Huangs list of papers and noted down the most intriguing ones below. Tons of AI-supported CGI-techniques and stuff you’ll see in applications in the coming years.
Papers
Authentic Volumetric Avatars From a Phone Scan / Meta Wants to Use Phones to Make Realistic 3D Avatars of Users
GANimator, a generative model that learns to synthesize novel motions from a single, short motion sequence.
Ecoclimates: Climate-Response Modeling of Vegetation: we model tree growth interactively in response to gradients of water, temperature and light. As a result, we are able to capture a range of ecoclimate phenomena that have not been modeled before, including geomorphic controls, forest edge effects, the Foehn effect and spatial vegetation patterning.
Procedural Urban Forestry We introduce procedural placement models (PPMs) for vegetation in urban layouts. PPMs are environmentally sensitive to city geometry and allow identifying plausible plant positions based on structural and functional zones in an urban layout.
StyleGAN-NADA: CLIP-Guided Domain Adaptation of Image Generators StyleGAN-NADA converts a pre-trained generator to new domains using only a textual prompt and no training data.
ComplexGen: CAD Reconstruction by B-Rep Chain Complex Generation CAD-Files from Point Clouds
MPS-NeRF Generalizable 3D Human Rendering from Multiview Images
VirtualCube Immersive virtual 3D-Meeting tech, supposedly for cubicles in big space offices. Advanced Zoom basically.
EyeNeRF: A Hybrid Representation for Photorealistic Synthesis Animation and Relighting of Human Eyes Synthetic 3D eyes incl "Gaze Reaction to cinematic environment" - Simulated eyes following visual stimuli including reflections of that stimuli in the eyes and i'm like O_O
Generative GaitNet Understanding the relation between anatomy and gait is key to successful predictive gait simulation. In this paper, we present Generative GaitNet, which is a novel network architecture based on deep reinforcement learning for controlling a comprehensive, fullbody, musculoskeletal model with 304 Hill-type musculotendons. The Generative GaitNet is a pre-trained, integrated system of artificial neural networks learned in a 618-dimensional continuous domain of anatomy conditions (e.g., mass distribution, body proportion, bone deformity, and muscle deficits) and gait conditions (e.g., stride and cadence).
GWA: A Large Geometric-Wave Acoustic Dataset for Audio Deep Learning dataset for simulating accoustics in any room setting. your voice sounds different in you speak in a large room in a concrete building, or in a room in a cabin in the woods.
Holographic Glasses for Virtual Reality Ultra-thin (2.5 mm) glasses-form factor VR display supporting 3D holographic images
Palette: Image-to-Image Diffusion Models We've developed a simple and general framework for image-to-image translation called Palette. We've evaluated Palette on four challenging computer vision tasks, namely colorization, inpainting, uncropping, and JPEG restoration. Palette is able outperform strong task-specific GANs without any task-specific customization or hyper-parameter tuning.
Self-Distilled StyleGAN: Towards Generation from Internet Photos Filtering outliers in datasets.
Sketch2Pose: Estimating a 3D Character Pose from a Bitmap Sketch
VoLux-GAN: A Generative Model for 3D Face Synthesis with HDRI Relighting We propose VoLux-GAN, a generative framework to synthesize 3D-aware faces with convincing relighting. Our main contribution is a volumetric HDRI relighting method that can efficiently accumulate albedo, diffuse and specular lighting contributions along each 3D ray for any desired HDR environmental map.
Neural 3D Reconstruction in the Wild We introduce a new method that enables efficient and accurate surface reconstruction from Internet photo collections in the presence of varying illumination. To achieve this, we propose a hybrid voxel- and surface-guided sampling technique that allows for more efficient ray sampling around surfaces and leads to significant improvements in reconstruction quality.
Artificial Phil Collins
JohnDopamine5s Youtube-Channel is one of the most interesting bits in the AI-music-space. He uses OpenAIs Jukebox-algorithm trained on 1 million songs to generate new versions of songs, with the intro working as a prompt, with some fascinating outcomes. Here’s AI-versions of the Beastie Boys, Daft Punk Live DJ-Sets, no less than 93 AI-Nirvana-Songs, Smashing Pumpkins, Super Mario soundtracks, and, yes, ofcourse: Artificial Intelligence creates new versions of Phil Collins’ "Don't Lose My Number".
[2201.07406] Neural Language Models are Effective Plagiarists it may very well be that good AI-systems are always ahead of the curve when it comes to cheating.
The Qubit Game Googles QuantumAI made a sweet little browser game to explain quantum computing.
Hologram doctors beamed to space station to visit astronauts In 2021, a team of hologram doctors was "holoported" to space to visit astronauts living aboard the International Space Station, NASA has revealed in a new post. The hologram teams, led by NASA flight surgeon Dr. Josef Schmid and Fernando De La Peña Llaca, CEO of software provider Aexa Aerospace, were the first humans to ever be "holoported" from Earth to space.
Researchers develop a paper-thin loudspeaker Resonanzraum, you ain’t seen nothing yet.
Rotary Un-Smartphone The 3D-printed rotary mobile phone that made the rounds some years ago is now a DIY-kit and available for pre-order.
People who hate magic tend to be higher in socially aversive traits, according to new research They also eat little children and voted for Trump.
There's a Reliable Way to Trigger Lucid Dreams an alzheimer drug can reliably induce lucid dreaming in people who are already trained on it.
ArteTV uploaded the Iron Maiden-doc Flight 666 to their YT-channel. Related: Judas Priest's Guide to Heavy Metal
Idle Breakout by Kodiqi Use multiple balls of different strength, speed and special abilities to smash a million bricks!
The Fictional Liveability Index South Park has a higher living standard than Twin Peaks. Now you know.
Windowlicker but everything is my voice check out this guys YT-channel full of remixes where he imitates all instruments with his voice.
Empire Strikes Back - C64 fan version by Megastyle Back in 1982, The Parker Brothers released their first game, The Empire Strikes Back. The game was available on two popular gaming consoles at this time, the Atari 2600 and the Intellivison. (…) 40 years after (…), Megastyle is now proud to present to you: The Commodore 64 fan version.
Average Happiness by Maja Gehrig Beginning as a very traditional PowerPoint presentation displaying some truly riveting information about life-satisfaction in Europe from 1973-2009, as the demonstrator begins to look at linear relationships in the scatter graph the lines begin to pull at the edges of the chart, revealing the world beyond – one made up entirely of visual representations of data. From here on out, everything takes on a very organic feel, as bar charts sprout from the ground like flora, a pie chart rises into the sky like a sun and an area graph breaks free of its cage and starts to travel slug-like across the screen.
Mushroom Color Atlas The Mushroom Color Atlas is a resource and reference for everyone curious about mushrooms and the beautiful and subtle colors derived from dyeing with mushrooms. But it is also the start of a journey and a point of departure, introducing you to the kaleidoscopic fungi kingdom and our connection to it.