I've been quietly experimenting with AI for months and I think it's time to stop being quiet about it.
Starting from zero here. Going to share what I'm learning, building, and figuring out.
Let's see where this goes.
Playing with my son’s toy cars and magnetic tiles for some Seedance 2.5 testing.
The idea was simple: use real toy movement as a motion reference, then see how Seedance translates it into a cinematic real-world scene.
A traffic jam made from toy cars became a full-scale road scene. A toy excavator clearing the jam became a massive construction vehicle pushing through real cars. A magnetic-tile house became an actual building attempt.
What’s interesting is that Seedance doesn’t just copy the action. It picks up on camera angle, object dominance, timing, contact points, weight, direction, and even accidental scale cues.
In one clip, the toy excavator looked huge because it was close to the camera and cropped at the edge of frame. Seedance translated that almost literally, making the real excavator tower over the cars.
That’s the fascinating bit.
The reference video becomes more than choreography. It becomes a kind of physical prompt: perspective, momentum, gravity, friction, hierarchy, and staging all baked into a few seconds of footage.
Basically: tabletop previs -> real-world cinematic scene.
Still messy. Still unpredictable. But honestly, that’s what makes it fun.
Feels less like prompting now, and more like directing motion with whatever is lying around the house.
Share this with someone who still thinks AI video is just about writing better prompts.
POV: someone made a fake you and it's selling clay protein online
🎭 TikTok's testing a tool that scans for AI clones of your face. Opt-in, US only, still beta.
So it's a start, not a fix.
Follow for the honest version
👉 full breakdown on Slop Reel (bio)
I did something yesterday that pushed me way out of my comfort zone.
I hosted my first online webinar on making AI videos, to followers I've built up myself, on a subject I only picked up 6 months ago. It's amazing how I went from knowing nothing to running live sessions and handing out my full workflows.
So I'm going further. On August 13th I'm running a hands-on workshop where I sit with participants and help them turn their own ideas into polished AI videos, for their brand or their socials.
If that sounds like your kind of thing, all the details are at https://t.co/RegJtMVa5A.
And while you're there, sign up to the newsletter for the latest in the world of generative AI.
#generativeai #VideoMarketing #CreatorEconomy #LearnAI
I've been quiet on X. Meanwhile on Instagram and Threads, my AI videos (IG: @apoorvabuilds) started doing numbers I didn't expect. One reel crossed 629K views, and on Threads the replies kept asking the same question: "How do you make these?"
So I put together a free 45-minute live session on exactly that.
I'll walk through my full workflow:
→ How I plan a video before touching any tool (this is where most people go wrong)
→ The models I actually use for footage and voiceovers
→ Editing in CapCut so the result doesn't look like AI slop
→ The small giveaways that scream "AI made this" and how to remove them
No fluff and no upsell at the end. It's genuinely free, no card required, and there's a replay if you can't attend live.
Two sessions so timezones are covered:
Jul 14 for India, UK and Europe
Jul 16 for US and Australia
Sign up: https://t.co/2dbFuVnfy0
With this prompt you can sit in any stadium, any game you want. Live your dream of being captured on camera in a stadium.
Another day another trend!
gpt image 2 and seedance 2.0 it is
#rcb#ipl20206
When you have no budget for ads, you literally get under a car to promote.
Comment if you would like to try out this #dashcam concept using Seedance 2.0
We put a camera inside a dishwasher.
The dishes started talking.
One mocked the other for being dirty.
It got personal.
🧵 Full AI-generated film, voice cast included 👇
I like the video concepts of @chrisfirst. It's super fun to recreate.
Here is my take on the paparazzi concept.
Another Seedance 2.0 and GPT Image 2 production.