Amir Zarkesh’s Post

You’re just chilling, watching a video, and thinking, “Wow, wouldn’t it be cool if I could add insane visual effects just by saying what I want?” Well, that’s pretty much what AutoVFX is all about! It’s like the magic wand for video editing. 🎥✨ AutoVFX takes one video, mixes it with natural language instructions (yep, you just tell it what you want), and BAM! You’ve got crazy realistic and dynamic visual effects. No complicated tools, no endless tweaking just pure editing awesomeness. It combines some next-level tech, like neural scene modeling, AI-powered code generation, and even physics simulations, to make sure the results are super photorealistic and accurate. Apparently, it’s so good it leaves other methods in the dust, whether it’s about the quality, how well it follows your instructions, or just how versatile it is. AutoVFX makes pro-level editing feel like a breeze for anyone. How cool is that? 😍 #AutoVFX #AI #VideoEditing #VFX #TechInnovation #NaturalLanguageProcessing #CreativeTech

"HOLODECK: Language Guided Generation of #3D #EmbodiedAI Environments" - Abstract:" 3D simulated environments play a critical role in Embodied AI, but their #creation requires expertise and extensive manual effort, restricting their diversity and scope. To mitigate this limitation, we present #HOLODECK, a system that generates 3D environments to match a user-supplied prompt fully automatedly. HOLODECK can generate diverse #scenes, e.g., arcades, spas, and museums, adjust the #designs for styles, and can #capture the #semantics of #complex queries such as “apartment for a researcher with a cat” and “office of a professor who is a fan of #StarWars”. HOLODECK leverages a #LargeLanguageModel #LLM (#GPT4) for common sense knowledge about what the scene might look like and uses a large #collection of #3D #assets from Objaverse to populate the scene with diverse #objects. To address the challenge of positioning objects correctly, we prompt GPT-4 to generate #spatial relational #constraints between objects and then optimize the layout to satisfy those constraints. Our large-scale human evaluation shows that annotators prefer HOLODECK over manually designed procedural baselines in residential scenes and that HOLODECK can produce high-quality outputs for diverse scene types. We also demonstrate an exciting application of HOLODECK in Embodied #AI, #training agents to #navigate in #novel #scenes like #music rooms and daycares without human-constructed data, which is a significant step forward in developing general-purpose #embodied #agents. " https://lnkd.in/e6C69D5Y #space #sciences #research #technology #gaming

DeepMind's Genie 2 generates interactive 3D worlds from single images. These simulated environments run at 720p (interactive version has reduced quality) for up to a minute and include physics and complex lighting. Genie 2 even remembers previously viewed scenes, which was a weakness of previous systems. DeepMind sees applications in game development for rapid prototyping. But there's a more interesting use case. Combined with Deepmind's SIMA agent, which navigates digital worlds using only natural language, Genie 2 can generate unlimited training scenarios for embodied AI. This could accelerate the development of agentic AI capable of complex actions in both real and digital scenarios. Summary: https://lnkd.in/gCXsx2_c Blog: https://lnkd.in/g-Z8NnAu Follow THE DECODER - EVERYTHING AI for daily #AI news.

FLUX.1 → Luma AI Last Frame is WOW 🤯 We have new players catching up and they are DAMN good! These are is state-of-the-art generative deep learning models for media that push the boundaries of creativity, efficiency and diversity. Black Forest Labs released the FLUX.1 suite of models that push the frontiers of text-to-image synthesis. The best part? It is open source! It excels at: +Composition, organic textures, art styles and text output. +Truly impressive language understanding and text control. +Lighting fast generation speed. 10 times faster than MJ V6! Luma AI's Dream Machine needs no intro. Their 'Keyframe' feature let's you get the perfect video while also leaving plenty room for creativity. I but them together to turn this image into a killer 4 second animation. Where do you think we will see in the next 6 months? Got your hands on them yet? -- P.S. Want to become AI pro in just 3 weeks and master AI in creative workflows? Comment below and I will send you a link to our AI x Creative Accelerator program. #ai #aivideo #aianimation #midjourney #chatgpt #runway #flux1

"I'm thrilled to share my ongoing work with image morphing powered by AI and CodeFormer. It's exciting to witness how tools like Stable Diffusion are shaping the future of visual creativity, making it easier than ever to bring imagination to life. Stay tuned for updates as I continue to explore this evolving space!" What is Stable Diffusion? Stable Diffusion is a revolutionary AI model designed for generating high-quality, photorealistic images from text prompts. Developed as an open-source project, it leverages advanced deep learning techniques, specifically latent diffusion models, to interpret textual descriptions and produce corresponding images. Its applications span across creative design, media, gaming, and more. Why Stable Diffusion is a Game-Changer Creative Freedom: Allows users to visualize ideas with text prompts. Versatility: Handles a wide range of styles, from photorealistic images to abstract art. Open-Source Innovation: Enables developers and researchers to customize the model for various applications. My Journey in Image Morphing Using AI-Powered CodeFormer As part of my work, I've been diving into the fascinating world of image morphing through CodeFormer, an AI-powered framework. This cutting-edge technology allows for seamless transformation of one image into another, creating visually stunning outputs. By combining Stable Diffusion with the advanced capabilities of CodeFormer, I've been exploring: Dynamic image transitions. Creative morphing techniques for storytelling. Innovative solutions for design and media applications. Let's keep on working with current technology... #stablediffusion #openai #ai #imagegeneration #technology #videoediting #opensource #trend #gamechanger..

LLaMA-Mesh: Unifying 3D Mesh Generation with Language Models Something super exciting just dropped in the 3D world! Ever wondered what would happen if you could chat with AI about creating 3D models? Well, these folks just made it possible! They've taken those smart language models we all know and love, and taught them to understand and create 3D meshes The coolest part? They figured out how to turn complex 3D stuff into simple text that AI can understand it's like teaching AI to speak "3D" in its own language! #3DAI #ArtificialIntelligence #TechInnovation #3DModeling #AIResearch #FutureTech #ComputerGraphics #AIart #MachineLearning #TechBreakthrough #3DDesign #Innovation #ResearchAndDevelopment #AIcommunity #ComputerScience

just a few months has gone by and what AI has evolved into is truly amazing. 🤖 Want to create games but don't know where to start? Genie by Google have created model that is able to generate an endless diverse of playable game environments created from images, photographs, or even sketches from us! 🚀 isn't this truly amazing? Genie can be prompted with a given image, enabling us to be able to interact with the creation of our own virtual world. what's amazing is that Genie is able to learn fine-grained controls without any training data with action controls! Genie really opens up a door to a new world of video game generation. 🎮 I believe this is just the beginning of what is possible in generation of the virtual world. 🌍 Check out more here: https://lnkd.in/gUuMMtge #deeplearning #machinelearning #generativeai #ai #artificialintelligence

Skillshare this week is featuring HuggingFace 🤗 and a Unity text-to-image tutorial. You can learn how to do prompt-to-image using HuggingFace (Stable Diffusion API). Don't miss out on this opportunity to expand your skill set. Join us: https://lnkd.in/gfeGV9f9 #Unity #huggingface #AI #TexttoImage #XReality

A New Development in Generative AI Sora is Here! Imagine typing a few words and instantly getting a video in return. Sounds like something out of science fiction, right? But it’s real. OpenAI’s Sora has officially launched, and it’s ready to turn your text into video magic. After months of being in the hands of artists and filmmakers, this tool is now open to everyone. You simply type a prompt, and voilà—a video appears. And the best part? You can remix it to your heart’s content, adjusting the video with additional text prompts. But here’s the catch like any new tech, it’s not perfect. Marques Brownlee, who tested it, said it’s “horrifying and inspiring at the same time.” The tool can generate breathtaking landscapes and stylistic effects, but it still struggles with realistic physics objects sometimes pass through each other or disappear altogether.🤯 So, here’s the question: Is this the future of creativity, or are we rushing things too fast? Let’s talk about it! What do you think about AI taking over video creation? Share your thoughts below. 👇 #Sora #OpenAI #GenerativeAI #AIinVideo #TextToVideo #AIContentCreation #VideoCreation #FutureOfAI #TechInnovation #AIRevolution #CreativeAI #AIFuture #TechTrends #AIArt #StorytellingAI #AItools #VideoTech #AIandCreativity #AIVideoGeneration #technology

Turning hair into tentacles 🦑: How is it made? We made this magic happen with a mix of computer vision segmentation, creative coding, generative AI and some classic compositing techniques. What crazy transformation should we try next? Drop your ideas in the comment below 💬👇 AI art by Nestor Rubio + Federico Guardabrazo • • • #specialguestx #sgxperiment #ai #generativeai #artificialintelligence #computervision #aisegmentation #creativecoding #genai #stablediffision #stablediffusion #stablediff #stablediffusionart #aigenerated #promptforum #digitalart #aivideo #aiartcommunity #aicccreators