Moonvalley: Like MidJourney, but for videos
PLUS: 🌋 LLaVa: New GPT4(Vision) competitor
Hi folks!👋🏻 This is The Prompt!
Let's get it
🔥 top picks
The authors used GPT4 and an image encoder, to train a multimodal language model that can achieve similar results to GPT4(Vision). Their model and code is open-sourced, and you can try the demo here.
The Yearbook AI trend took over TikTok (3min read)
According to Google Trends, the Yearbook AI trend reached pick popularity this weekend. This is a new “avatar” trend trend using AI; this time you can recreate your high school yearbook photos — with a ’90s twist. We wrote a tutorial on how you can create your own photos(like the ones below) using EPIK or for free.
New text2video model: Like MidJourney, but for videos (direct link)
Text to video is getting better! If you pay attention to this space, you will be surprised by how much better the videos you can make with Moonvalley have become.. To try it out, head to their discord and use the /create command in one of their new-moon channels. Here are some examples:
📕 get smarter
[short examples] How to create illustrations with DALLE-3
[tips and tricks] Extracting prompts from images for different image models (SD, DALLE3, and Midjourney)
[paper] Recent paper proved that Perplexity and GPT-4 wwith prompting outperformed Google Search. Who even uses Google Search anymore?
[a must read] AI’s $200B Question
[twitter thread] Create stickers using DALLE3
✍🏼 prompt of the day
35mm, Vintage American dinner, waitress waiting orders, early morning, grazing natural light, pure shadows, Leica, analog --style raw --c 6 --w 10 --ar 2:3
What'd you think of today's edition?