Lie detector test goes horribly wrong. My first animated video with Midjourney and Veo 3!. Here's how I did it 👇

How it feels when your phone dies 💀. Made with @midjourney and @GoogleLabs Veo 3

I've always loved making videos, but only recently revisited the hobby after the Veo 3 launch. Excited to follow this passion and see where it takes me!. Tagging my fav AI video creators / curators for feedback 🫶. @0xFramer @Artedeingenio @Diesol @PJaccetturo @techhalla.

3) Video. I use a combination of Veo 3 and Midjourney Video. For dialogue, Veo 3 is UNMATCHED. It works exceptionally well with photorealistic animation styles, and the voices and lip sync are almost perfect. For action and more abstract scenes, MJ video is a great complement.

Scene consistency is challenging, but I have a few tricks:. A) Reuse first frames where you can. Dialogue-driven story makes this easier. B) Use blurry backgrounds. Let characters in the foreground capture the audience's attention. C) Use ChatGPT for granular editing. GPT

For character consistency, I use Midjourney omni references with a character reference sheet that has a three-view turnaround on a blank background. ChatGPT is great for generating character sheets from a messy image. It sounds obvious, but simple features and clothing make

2) Frames. Generating the first frame for each scene demands the most attention to detail. A good first frame makes video gen 100x easier. Midjourney is still king for image generation imo. I like to keep my prompts simple to maximize variety and creativity in the output.

1) Script. Great storytelling cuts through (even on TikTok and IG). Always has, always will. I started by writing story beats, then turned them into a more detailed list of shots. I reviewed the script 3-4 times with fresh eyes before locking it in. My process here is crude

i built a website that helps me get off my phone and touch some grass 🌿. link in replies

RT @Abhindas1: Every smart person is getting into hardware, and every smart person is finding out they are not that smart.

RT @soft_servo: @arian_ghashghai just get the arms working in unstructured environments first imo. huge market with some consumer potential….

RT @parthingle_x: Pitch: We make this in America.

RT @hollympeck: we're so early

just preordered a friend from 👉👈.

apple really spent 30 minutes talking about backdrop-filter: blur() like they discovered a new element. #WWDC25

seems to be resolved.

is there @vercel outage? everything just redirects our team to this page

RT @Suhail: A chunk of why hardware is hard is seemingly just getting the hardware.

RT @soft_servo: @JulianSaks the analogy for the teleportation approach would be trying to train an LLM without the internet by having human….