.png "back to home page")
Meta는 텍스트-동영상 생성 AI 프로그램 개발했다고 발표. 프롬터 자막에 맞춰 5초 길이 동영상 자동 생성. 아직 일반엔 미공개. https://t.co/qfgjPvOcBt
— Journey (@atmostbeautiful) September 30, 2022
Thread/ What to do about deepfake and synthetic media panic and hype mode? It's back big time with the latest AI-based media production news: text-to-image like Stable Diffusion/DALL-E opening up access + simple text-to-video via Meta's Make-A-Video https://t.co/U96GDsE7uS
— Sam Gregory (@SamGregory) September 30, 2022
MIT: @Meta’s New #AI Can Turn Text Prompts Into Videoshttps://t.co/AnuFvRDVrH via @AI_Advice#5G #100DaysOfCode #Analytics #ArtificialIntelligence #BigData #Cloud #Data #DataScience #IoT #MachineLearning #ML #NLP #OpenSource #Python #Serverless #TScottClendaniel #WomenWhoCode pic.twitter.com/LEPQVRlQ3b
— #AI Advice- #ArtificialIntelligence! 😀 (@AI_Advice) September 30, 2022
AI2 research scientist @tanmay2099 weighs in on Meta’s new text-to-video AI for @Melissahei in @techreview: https://t.co/SMWnjcV3mX
— Allen Institute for AI (@allen_ai) September 29, 2022
After intense text-to-image AI hype, Meta is the first big AI lab to unveil its new *text-to-video* AI. It's going to be harder and harder to know what's real online. https://t.co/toyD4xdfIo pic.twitter.com/SUTBw7HsXF
— Melissa Heikkilä (@Melissahei) September 29, 2022
This is pretty amazing, but also Meta’s track record of thinking through possible harms isn’t exactly reassuring. https://t.co/N0AIqNpahQ
— mat honan (@mat) September 29, 2022
Meta’s Got a New #AI Tool that Enables Users to Make Freaky Mish-Mash #Videoshttps://t.co/Lm1YvHHkex#midjourney #dalle2 #Web3 #digitalpainting #100DaysOfCode #OpenSource #bot #python #Digital #javascript #CodeNewbie #NFTs #digitalart #nftart #NFTartists #art #Artists #DL #RT
— 𝔸𝕞𝕚𝕥𝕒𝕧 𝔹𝕙𝕒𝕥𝕥𝕒𝕔𝕙𝕒𝕣𝕛𝕖𝕖 (@bamitav) September 30, 2022
This is going way too fast. https://t.co/yHUlFUFz4h
— Tim Soret (@timsoret) September 29, 2022
I loaded 10M videos that were used as part of the training for this new text-to-video model - from WebVid-10M (https://t.co/lnxd0uU0cX) - into a Datasette instance and made them searchable:https://t.co/clCAFcH0BQ https://t.co/vwNbr20Nkm pic.twitter.com/81vFOpxr7O
— Simon Willison (@simonw) September 29, 2022
✨Introducing Make-A-Video: a powerful text-to-video generative AI system by @MetaAI.
— elvis (@omarsar0) September 29, 2022
Say hello to generating videos with text prompts!
Paper: https://t.co/VnewLT4hZP
Website: https://t.co/9MIDn7OARF
1/5 pic.twitter.com/PZAm9Rid6A
This is insane $META
— Matthew Cochrane (@Matt_Cochrane7) September 29, 2022
"As the name suggests, this AI model allows users to type in a rough description of a scene, and it will generate a short video matching their text." https://t.co/CwSDtlh3oK
Really crazy just how quickly these tools have developed in recent years. Yesterday it with GPT3 with text, then Dall-e with pictures, and now video.#AI https://t.co/SmEbwr0U8r
— Ulrike Franke (@RikeFranke) September 29, 2022
If you squint and consider the evolution of this product over the next ~5 years, you can better understand Meta’s strategy with Reels / the open graph / short-form video as a foundational re-imagination of its ads business https://t.co/Pqe76f8PUz
— Eric Seufert (@eric_seufert) September 29, 2022
Can’t wait to find out they lifted source images from private photos in users accounts. https://t.co/MRVeIp0Oph
— Tony Stark (@Iron_Man_Actual) September 29, 2022
Meta is the most trusted untrustworthy company I can think of. https://t.co/ExYw253YVc
— drew olanoff (@yoda) September 29, 2022
Prompt: Mark Zuckerberg apologizing to Rohingya refugees for enabling genocide in Myanmar https://t.co/sMQMim69tt
— Parker Molloy (@ParkerMolloy) September 29, 2022
Generative AI systems like Make-A-Video are pushing creative expression forward by giving people the freedom to shape and control the AI content they create.
— Meta AI (@MetaAI) September 29, 2022
Learn more about Make-A-Video: https://t.co/qyvIi7SSrT
welp. Generative reality replacement is one step closer https://t.co/cxOPewPlnC
— bryson (@Bryson_M) September 29, 2022
Meta's new text-to-video AI was trained on 10.7 million 15-second-long watermarked sample videos scraped from Shutterstock. https://t.co/BjgeDeaPLc
— Andy Baio (@waxpancake) September 29, 2022
oh great, great, i see absolutely no potential problems that could come from this
— Charles Bergquist (@cbquist) September 29, 2022
but also it is v neat https://t.co/Z9KmeGTufk
AI researchers at Facebook's parent company Meta have unveiled Make-A-Video — an early stages text-to-video AI model that generates five second video clips like those below based on text prompts. details here: https://t.co/ZCZULw9P8a pic.twitter.com/Ruf7Qh62iU
— James Vincent (@jjvincent) September 29, 2022
Meta’s new AI generator can create videos like this just from text. Here’s “a teddy bear painting a portrait”https://t.co/nTK3d77mL6 pic.twitter.com/jIMKMtSQsR
— Tom Warren (@tomwarren) September 29, 2022
If you view $META's mind-blowing text-to-video AI tool (https://t.co/np5vVdZsX6) in conjunction with the marketing strategy automation piece by @eric_seufert then maybe, just maybe, you can get a glimpse into the future. https://t.co/vOQok9mYEj
— George Hadjia (@GHadjia) September 29, 2022
Imagine if there was a regulatory requirement to include the worst case scenario in these kinds of research examples….. https://t.co/du0NW0bkSr
— emily bell (@emilybell) September 29, 2022
The legs on the horse/unicorn things are mesmerizing https://t.co/vyHz5op6uH
— Janelle Shane (@JanelleCShane) September 29, 2022
Here's another sample video, generated from the prompt "horse drinking water." It's obvious that it's artificial, yes, but think how quickly text-to-image programs have advanced in just the past few years. The quality of this output will likely improve a lot in the near future. pic.twitter.com/xv0Kp3jsWo
— James Vincent (@jjvincent) September 29, 2022
I've seen some text-to-video AI art already, though it's more panned static images than anything else. Meta's Make-A-Video looks like a substantive jump into short AI videos. Good stuff. https://t.co/wFFkP84wR3
— Mark Hachman (@markhachman) September 29, 2022
It's exciting stuff but we're gonna need regulations on this stuff a.s.a.p.
— Karla Ortiz (@kortizart) September 29, 2022
I also hope all creatives realize this isn't an issue for just concept artists or illustrators, it'll soon be an issue for all of us. These technologies will only get better. https://t.co/JDjEzwOMgJ
This was clearly coming, but it's hard no to be wowed by not only the quality of the videos but also what this means for the future of media 😝 https://t.co/NZQG7zqRC4
— Xavier Amatriain (@xamat) September 29, 2022
These AI models are impressive, but a future full of misleading videos and imagery feels increasingly likely pic.twitter.com/V8cGiEZRjK
— Tom Warren (@tomwarren) September 29, 2022
First I thought wow. Then I thought haha doggo.
— Irene Solaiman (@IreneSolaiman) September 29, 2022
But all I can think about now is how tf do we create and run robust behavioral/social/safety evals for these systems???
Just when I was getting ideas for text-to-image. AI progress moves *fast*. https://t.co/ei12ODU1Du
We are in for a tsunami of artificially-generated images and video, and it is going to get better and better until it's almost indistinguishable from the real thing, and that is going to have a range of unforeseen consequences https://t.co/ASAwTSdYXb via @Verge
— Mathew Ingram (@mathewi) September 29, 2022
I truly love technology. but this is going to go terribly. https://t.co/MnMJWm84XN
— drew olanoff (@yoda) September 29, 2022
Facebook’s release of previous AI tools like its Blenderbot 3 chatbot has been pretty responsible and transparent. But this looks like a misinformation nightmare down the line — let’s see how FB rolls it out, if ever. https://t.co/4hsXX9zxSn
— Parmy Olson (@parmy) September 29, 2022
I hope to heck the prompt on the lower right was "A herd of charging equoids about to feast on the blood of off-screen victims", right @cstross? https://t.co/8hHnlIqJPL
— Nicholas Weaver (@ncweaver) September 29, 2022
Meta just published „Make a Video“.
— Marcel Pociot 🧪 (@marcelpociot) September 29, 2022
An AI model that turns text into videos 🤯
“A fluffy baby sloth with an orange knitted hat trying to figure out a laptop close up highly detailed studio lighting screen reflecting in its eye”https://t.co/sdjx0Hv0Rn pic.twitter.com/W80zAEWW1i
We’re pleased to introduce Make-A-Video, our latest in #GenerativeAI research! With just a few words, this state-of-the-art AI system generates high-quality videos from text prompts.
— Meta AI (@MetaAI) September 29, 2022
Have an idea you want to see? Reply w/ your prompt using #MetaAI and we’ll share more results. pic.twitter.com/q8zjiwLBjb
Barely a month since @StabilityAI released #StableDiffusion and we have a @MetaAI text to video generator: https://t.co/EsgZvqPk7m #syntheticmedia #deepfakes
— Nina Schick (@NinaDSchick) September 29, 2022
Slippery slope ... #Meta announces Make-A-Video, which generates video from text #AI #Ethics
— Theodora (Theo) Lau - 劉䂀曼 🌻 (@psb_dc) September 29, 2022
cc @ChrisGGarrod @MiaD @hessiejones @AlaricAloor @pierrepinna https://t.co/tFLKK44AWM via @benjedwards @arstechnica pic.twitter.com/0RNm3D6WNT
There's now a tool that generates short video clips from text descriptions—an unsettling, albeit inevitable, next step for the world of AI image generation. https://t.co/7oXCojAZvB
— Chloe Xiang 向舒锦 (@chloexiang) September 29, 2022
Computer, load Celery Man https://t.co/oiJMe5eJSB
— PS1開始音は美しいレズとしての (@janusrose) September 29, 2022