Hey 👋
I'm writing this from Bangkok.
Waiting for a friend outside, I took two photos from the exact same spot.
Same angle. same spot. one of is zoomed a bit.
I thought: "What if I made this into a day-to-night timelapse... without waiting for night?"
Here's what I did:
Used Gemini to convert the second photo to a night version.
Fed both frames (day + AI-night) to 4 video tools with the same prompt.
The tools I tested (all via fal.ai):
• Wan-2.1 • Veo 3.1 • Veo 3.1 Fast • Kling O1
The Results
Same input. Same prompt. Very different outputs.
(I put them in a Notion page so you can compare without loading heavy files.)
💡 Takeaway
First-frame-last-frame prompting is the new hack for AI video. Even still I’m not satisfied with the results. But.
You don't need complex scripts but two anchor points.
Give the AI the "before" and "after", and let it figure out the journey.
See you next week,
Yunus 🚀
P.S. Working from Bangkok for a while and will head to Vietnam soon. If you're here too, say hi 👋
