The story behind Imagine Flash going from research to production in record time. What a fun ride it’s been to work in this project!
A fun backstory: we first demoed this breakthrough to @finkd, @Ahmad_Al_Dahle and others this February on a single node. There was so much excitement about this research that in a mere 2 months we scaled it up and today is in production available to billions via Meta AI #movefast
More Flash examples…
Today you can experience the future of image generation with our latest version of Meta AI. Offering lightning-fast speeds, stunning high-quality images, and a revolutionary new way to create and animate custom images using simple text prompts.
@savitz described it pretty well:
“...text-to-image queries will be created and updated in real time... It’s a remarkable show of computing power, which is going to result in massive time wasting by people everywhere."
Check it out on meta.ai or in @messenger, @WhatsApp, @instagram or @facebook!
Imagine Flash now powers high quality real-time image generation on meta.ai and WA. Check out our paper and try it yourself!
In addition to Llama 3, today we’re also publishing a new paper: Imagine Flash: Accelerating Emu Diffusion Models with Backward Distillation ➡️ go.fb.me/g4r584
This work from GenAI researchers is enabling new image generation features in Meta AI on @WhatsApp & web.
Imagine Flash is out! High quality real-time image generation. Try it on meta.ai and WA in the US, and check out our paper: tinyurl.com/2bbxbwh6
@AIatMeta
📢Introducing Imagine Flash
Buckle up! We're releasing a new model for real-time image generation. Available in meta.ai and WA.
📄 Paper: tinyurl.com/2bbxbwh6
Amazing team: J.Kohler, @schoenfeldedgar, @artsiom_s,
@rsumbaly, P.Vajda, @alitabet @AIatMeta
Ali Thabet retweeted
Thrilled to announce that our Zurich team was directly responsible for optimizing the AI Sticker generative model. Just type in a description, and watch as it creates personalized stickers for you in IG/FB, WA.
A glimpse of this in an excerpt from keynote by MZ at Meta Connect
Ali Thabet retweeted
Since joining Meta GenAI, our team focused on speed advancements in image synthesis.
Exciting news!🚀 We've unlocked high-quality image synthesis in just ~5 sec! MZ showcased our progress at Meta Connect.
Try it out with the /imagine command in our AI chatbot in FB, IG or WA
Ali Thabet retweeted
🚀 Exciting News!!
Hard work paid off! 🙌 In #MetaConnect2023, MZ showcased our tech enabling high-res image generation in <5 seconds! 🖼️💥
Try it out on WhatsApp, Instagram, and Facebook
Huge thanks to our incredible team! @alitabet @artsiom_s @schoenfeldedgar and J. Kohler.
Thrilled to see my team's work highlighted in this year's Meta Connect. The GenAI Media/Efficiency team optimizes image and video generation models to run very fast at inference team.
teapaw.org/AlbertPumarola/status/…
teapaw.org/artsiom_s/status/17074…
Replying to @artsiom_s
Huge shoutout to our amazing team at Meta GenAI Zurich -
@alitabet @AlbertPumarola @schoenfeldedgar and Jonas Kohler!
3
11
1,129
Ali Thabet retweeted
We have released the code and weights for our #CVPR2023 paper "Avatars Grow Legs: Generating Smooth Human Motion from Sparse Tracking Inputs with Diffusion Model"!
code: github.com/facebookresearch/…
abs: arxiv.org/abs/2304.08577
project: dulucas.github.io/agrol/
The demo is below:
Ali Thabet retweeted
Our CVPR2023 paper is finally out on arxiv🤩 this is a joint work with @artsiom_s @AlbertPumarola @alitabet @robin_kips @blacksquirrel__ during my internship at Meta. Amazing team and fantastic experience!!! Stay tuned for the code.
Ali Thabet retweeted
VisCo Grids: Surface Reconstruction with Viscosity and Coarea Grids
The goal of this work it show that network-free grid-based implicit representations can achieve INR (Implicit Neural Representation)-level 3D reconstructions when incorporating suitable priors.
1/n
Ali Thabet retweeted
In 1 hour at #NeurIPS2022 I will be presenting VisCo-Grids, a grid-based surface reconstruction method incorporating Viscosity and Coarea priors.
Joint work at
@MetaAI with @AlbertPumarola,
@YarivLior, @alitabet and @lipmanya
research.facebook.com/public…
Details in the thread 🧵
Ali Thabet retweeted
This week in #NeurIPS2022 we will be presenting VisCo-Grids, a grid-based surface reconstruction method incorporating Viscosity and Coarea priors.
Work developed at @MetaAI with @artsiom_s, @YarivLior, @alitabet and @lipmanya.
📄PDF: research.facebook.com/public…
Ali Thabet retweeted
16 Days left for CVEU Workshop!
Remember that the workshop will happen one week before the official conference. Our workshop will have an amazing line-up of speakers and two great panel discussions about the future of creative AI. Don't miss it out!
Info: cveu.github.io
Ali Thabet retweeted
Honored to be organizing this event with @guocheng_qian, Jesus Zarzar, @GiancolaSilvio, @alitabet, Matthias Müller, @fedassa and @BernardSGhanem!
Ali Thabet retweeted
1. Combating Adversaries with Anti-Adversaries
Joint work with @MotassimF, @juancprzs, @alitabet, Phil, and @BernardSGhanem
arxiv.org/abs/2103.14347
Ali Thabet retweeted
The #ICCV2021 work from our @KaustVision group be @PardoAlejo Juan , @alitabet @BernardSGhanem is featured in the computer vision monthly magazine
rsipvision.com/ComputerVisio…

















