🧪 Light Field Neural Rendering 🧪

👉Two-stage transformer capable of non-Lambertian effects (reflection, refraction, translucency)

😎Review https://bit.ly/3CpIFdm
😎Paper arxiv.org/pdf/2112.09687.pdf
😎Project light-field-neural-rendering.github.io
😎Code github.com/google-research/google-research/tree/master/light_field_neural_rendering

🦩Phenaki: Text-to(LOOONG)Video generation🦩

👉Phenaki is an #AI capable of realistic long video synthesis, given a sequence of textual open prompts

😎Review https://bit.ly/3RwUvXx
😎Project phenaki.video/index.h
😎Paper openreview.net/pdf?id=vOEXS39nOF

🔥 VToonify: Neural Portrait Style Transfer 🔥

👉VToonify for portrait style transfer. Powered by DualStyleGAN backbone, now with #stablediffusion!

😎Review https://bit.ly/3M9wgNP
😎Demo https://t.co/8gXzF3IrpB
😎Paper arxiv.org/pdf/2209.11224.pdf
😎Project mmlab-ntu.com/project/vtoonify
😎Code github.com/williamyang1991/VToonify

🐢 Stable Diffusion for #Pokemon 🐢

👉Fine-tuning the stable diffusion to create a text-to-pokemon generation model

😎Review https://bit.ly/3C9qBTw
😎Tutorial https://lambdalabs.com/blog/how-to-fine-tune-stable-diffusion-how-we-made-the-text-to-pokemon-model-at-lambda/

🔥 Imagen Video by #Google. SICK! 🔥

👉Novel text-conditional video generation via cascade of video diffusion models 🤯

😎Review https://bit.ly/3SH2TVH
😎Project imagen.research.google/video/
😎Paper imagen.research.google/video/paper.pdf

🔥 Human MDM: source code is out! 🔥

👉A classifier-free diffusion-based generative model for human motion domain

😎Review https://bit.ly/3rFhR2G
😎Project guytevet.github.io/mdm-page
😎Paper arxiv.org/pdf/2209.14916.pdf
😎Code github.com/GuyTevet/motion-diffusion-model

⚛️SOTA ALERT! Particles Tracking ⚛️

👉The new SOTA in video particles tracking. "Old school" taste, with neural flavor 🧡

😎Review https://bit.ly/3CaU5Ai
😎Project particle-video-revisited.github.io/
😎Paper arxiv.org/pdf/2204.04153.pdf
😎Code github.com/aharley/pips

🔥 #AIwithPapers: we are 4,500+! 🔥

💙💛 Someone put the smiling 💩 under a few recent posts. But I still love you! 💙💛

😈 Invite your friends -> https://news.1rj.ru/str/AI_DeepLearning

🍋 Long Video via Transformers 🍋

👉TECO is a vector-quantized latent dynamics prediction for long video

😎Review https://bit.ly/3Ch0tWD
😎Project wilson1yan.github.io/teco/
😎Paper arxiv.org/pdf/2210.02396.pdf
😎Code github.com/wilson1yan/teco

🔥SIMPLI: ligh novel-view synthesis🔥

👉Lightweight novel-view synthesis by #Samsung for arbitrary forward-facing scenes

😎Review https://bit.ly/3CivSYZ
😎Project samsunglabs.github.io/MLI
😎Code github.com/SamsungLabs/MLI
😎Paper samsunglabs.github.io/MLI/paper/paper.pdf

🥏 EVA3D: new SOTA in #3D humans 🥏

👉EVA3D: new SOTA for unconditional NeRF-human generation from 2D only

😎Review https://bit.ly/3Th9qX7
😎Code github.com/hongfz16/EVA3D
😎Paper arxiv.org/pdf/2210.04888.pdf
😎Project hongfz16.github.io/projects/EVA3D.html

🍏 f-DM: Diffusion Models by Apple 🍏

👉Spectacular work by #Apple on DMs: HQ generation with better efficiency and semantic

😎Review https://bit.ly/3Tils2u
😎Project https://jiataogu.me/fdm/
😎Paper arxiv.org/pdf/2210.04955.pdf

🏅GENIE by #Nvidia -> Faster Generation🏅

👉Higher-Order Denoising Diffusion Solvers for faster and better synthesis

😎Review https://bit.ly/3CRjtwr
😎Project nv-tlabs.github.io/GENIE/
😎Paper arxiv.org/pdf/2210.05475.pdf
😎Code github.com/nv-tlabs/GENIE

🥬 "Perception Test" by #DeepMind 🥬

👉Huge dataset with obj & point tracks, temporal sounds, multiple & grounded vQA

😎Review https://bit.ly/3Vqh96Q
😎Dataset github.com/deepmind/perception_test
😎Project www.deepmind.com/blog/measuring-perception-in-ai-models

🔥 Matterport 3D Semantics Dataset 🔥

👉#Meta opens HM3DSEM, the largest #3D real-world dataset with dense semantic

😎Review https://bit.ly/3yF4W4G
😎Paper arxiv.org/pdf/2210.05633.pdf
😎Project aihabitat.org/datasets/hm3d-semantics
😎Data github.com/matterport/habitat-matterport-3dresearch

🦑 Instant Map-free Relocalization 🦑

👉#Niantic unveils a novel instant, metric scaled re-localization with one single photo

😎Review https://bit.ly/3S1Gdyh
😎Paper arxiv.org/pdf/2210.05494.pdf
😎Project research.nianticlabs.com/mapfree-reloc-benchmark
😎Data research.nianticlabs.com/mapfree-reloc-benchmark/dataset

🧮 Novel DM for 3D Shapes by #Nvidia 🧮

👉Hierarchical Latent Point Diffusion Model (LION) for 3D shape generation

😎Review https://bit.ly/3yDhZ6I
😎Paper arxiv.org/pdf/2210.06978.pdf
😎Project https://nv-tlabs.github.io/LION/
😎Code(soon) github.com/nv-tlabs/LION

🪲#6D estimation fully in the wild🪲

👉First ever self-supervised 6D pose estimation training in the wild

😎Review https://bit.ly/3yHdHuS
😎Paper arxiv.org/pdf/2210.07199.pdf
😎Project kywind.github.io/self-pose
😎Code (soon)

⛽ Stable Diffusion in #Blender ⛽

👉Render with SuperPowers: novel scene render via text prompt

😎Review https://bit.ly/3s1mEeN
😎Code github.com/benrugg/AI-Render

⚽Markerless Body-Object Interaction⚽

👉Novel whole-bodies/objects interaction method from multi-view RGB-D data

😎Review https://bit.ly/3yO56GY
😎Data intercap.is.tue.mpg.de/login.php
😎Project https://intercap.is.tue.mpg.de
😎Code github.com/YinghaoHuang91
😎Paper intercap.is.tue.mpg.de/media/upload/main.pdf

🔥 Dressing Avatars by #META 🔥

👉Novel deep photorealistic appearance method for physically-simulated clothing in #metaverse

😎Review https://bit.ly/3yRBW9Y
😎Paper arxiv.org/pdf/2206.15470.pdf