Mythica’s Post

Legend has it that an interview with Walt Disney would require an artist to draw a perfect circle. Yet, the perfect circle is a mathematical abstraction. For now. No human or machine can create the perfect circle 🛞 Nevertheless, the Midpoint Circle Algorithm helps machines create circles that appear perfect in shape to the human eye. This is achieved despite computer generated images being made up of pixels, which aren't even curved. If the wheel is mankind's first invention, the imperfect circle is our signature. #ethicalai #web3

It's been a couple months now since I started an internal AI Engineering Guild ⚔️. I wanted to create a dedicated safe space for the team to really dive into the nitty gritty technical details and to share all the knowledge! Preparing for this weeks session I thought I'd jazz things up a bit and generate a cool new title image. Here's the result! Crazy, how aesthetic and prompt adherent our new Phoenix model is 🤯 #GenerativeAI #Phoenix #LeonardoAI

Building on my last post the next competition challenge was to take the characters and world we created and do a :30 scene. The dialogue and set were defined by the challenge. Well the first four lines of dialogue but if we had time to add we could... and the blown fuse is an inside joke from the challenge discord. Anyway. Learned a lot on this one. On the prior video I used LivePortrait via hugging face to get the lipsync. It wasn't good enough for me for an actual scene. I tried runway but was underwhelmed. Then I remember Pika does lip sync from video. And it was pretty pretty good (not prefect but good enough). The thing is with AI vid it's still really hard to control the action. Yes runway has great brushes but it's still all a roll of the dice. But I came up with a workflow that worked for me. I recorded a scratch track so I knew how long the dialogue would take and where the words would fall. Then rolled a lot of video in Luma. I looked for stuff where action seemed to be able to land on good bits of the dialogue so it would feel more natural. Next I headed to ElevenLabs and rerecorded the audio while watching the video so it paired up better. Changed the voice and popped it all together in Pika. Basically character VO. So it's not perfect... again because of that I was hesitant to show before. But it's a good experiment... and it got me in the finals. Tech stack: Images = Midjourney tweaked in Adobe Photoshop Video = Luma AI VO and GFX = ElevenLabs Lipsynch = Pika Music = Suno Editing = After Effects and Premier Pro Upscaling = Topaz Labs The thing I'm forgetting = ¯\_(ツ)_/¯ #aivideo

Generative Al is a terrific tool that can help save time and costs and spare us from tedious work necessary so we can dedicate our energies to higher-level work. But what is it doing to our own abilities to generate ideas, think critically, and wonder? How is it impacting human cognition and creativity? Are the minds behind generative Al thinking about ethical and societal implications? Are laws keeping up to ensure creatives' rights? Join us for a dynamic panel discussion as experts unravel the complexities of the relationship between the ever-evolving landscape of technology and imagination. https://lnkd.in/g4MvJ8G6

Published my first Comic "EP 1 Cosmic Heroes" on Dashtoon today. This is me dabbling with what studio has to offer and accuse me, i am not an Ai Artist but i sure have found my new hobby to perfection. #genai #aiart #aicomic #webcomic

Created this video with In"Video ai" about SVA and my book. Very interesting technology!

I love vision language models My favorite is KOSMOS-2, because it's a grounded model (it doesn't hallucinate). In this demo you can ask a question about the image, do detailed/brief captioning, and also let it localize the objects! Link in comments 💬

Mona Lisa speaks using Hedra Labs, creepy or not? 😄 PS. service wouldn't let me upload "original" Mona Lisa image, so I need to trick it with Leonardo.Ai image upscaler 👌 #monalisa #lipsync #hedra #leonardoAI

Midjourney 6.1 is here, and the results are mind-blowing! 🤯 Can you spot the difference between these AI-generated and real images? We put it to the test and honestly, the results are pretty epic. But can you tell the difference? Let us know in the comments and share your thoughts on how far AI art has come! #Midjourney #AIImages #AIGenerated #MidjourneyV6 #StockImages #AITools #DesignerToolkit

S1 - EP6 - Computer Vision(ary) Brad Dwyer 🎙 https://lnkd.in/eUGvMgac Brad Dwyer is a nerd you need to know. His expressions through software are thoughtful, award-winning, and world changing. Whether it’s building into #computervision, multimodality, augment reality, #machinelearning, AI, game development, Y Combinator, or raising $20M to support the growing team at Roboflow, this technologist leads the way and this episode of #YDNTP reminds us to stay optimistic as we all keep building!