TL;DR

  • Used Claude Fable 5.2 to auto‑generate a music video for the track Soul in Motion on Suno.
  • Hit the usage limit after ~2 hrs; had to pause before the cooldown.
  • Weekend packed with tennis, football, basketball, cricket, and a live‑streamed Jagran event.
  • Despite Delhi smog and back pain, finished the video and got it ready for a Monday launch.

The Friday Night AI Experiment

I was at my desk, nursing a cold from Delhi’s winter air, when the idea hit: turn the new track Soul in Motion into a video without pulling a full production crew. I set up a small pipeline of AI agents:

  1. Claude Fable 5.2 – the main creative engine.
  2. A prompt‑generation agent that fed Fable with visual themes, color palettes, and narrative beats.
  3. A post‑processing agent that stitched the generated clips, added transitions, and synced the audio. 
# Rough outline of the agent orchestration python orchestrate.py \ --track "Suno/Soul_in_Motion.mp3" \ --output "video/Soul_in_Motion.mp4" \ --agent ClaudeFable5.2

The first two hours produced a surreal montage: neon‑lit streets, slow‑motion dance moves, and a subtle nod to Delhi’s winter chill. I tweaked the prompts, adjusted the pacing, and was ready to polish the final cut when the usage counter hit zero.

Claude’s free tier caps at $X of compute per month. Once the limit was reached, the API returned:

{ "error": "Usage limit exceeded", "next_available_in_seconds": 3600 }

I had to wait an hour before I could resume. The cooldown was a good reminder that even the most creative AI tools have hard constraints.

The Sports Marathon

Tennis at 10 PM

A friend pinged me to play tennis at 10 PM. I thought I’d be a natural athlete, but the court turned into a comedy show. My backhand was a disaster, my footwork was guesswork, and I loved every minute. I finally made it home around 12:30 AM and crashed.

Football at 7 AM

Saturday morning, alarm before sunrise, I laced up for a 2½‑hour sprint session. Two and a half hours of sprinting through the crisp, polluted air later, I was back home at 9:30 AM, showered, and straight into the office.

Basketball at 9 PM

Tonight at 9 PM I’ll hit the basketball court. The plan is to keep the heart rate up and get a quick cardio burst before the next creative sprint.

Cricket at 10 AM

Tomorrow morning at 10 AM I have a cricket match—apparently I’m trying to speed‑run every sport before Monday.

The Grand Finale: Jagran (Mata Ki Chauki)

Tomorrow evening I’ll swap sneakers for a mic at a Jagran hosted by my landlord to celebrate his son’s upcoming wedding. I’ll sing, cough, limp, and somehow still find time to finish the Eli music video with Fable 5.2 and get it ready for a Monday launch.

The live‑stream setup is simple:

stream: platform: YouTube resolution: 1080p bitrate: 4500k audio: "Stereo, 48kHz" overlay: - logo: "assets/logo.png" - text: "Live from Jagran"

I’ll use ffmpeg to mix the live audio from the mic with the pre‑recorded track, ensuring the final mix is clean and punchy.

ffmpeg -i mic.wav -i track.mp3 -filter_complex "[0:a][1:a]amix=inputs=2:duration=first:dropout_transition=3" output.mp3

Wrapping Up

My back is screaming, my lungs are battling the Delhi smog, and every muscle feels like it’s filing a complaint. But when Monday arrives I’ll have a finished video and a weekend I actually lived—and that’s worth a little limp.

The key takeaways:

  • AI pipelines can be powerful but are bounded by usage limits. Plan for cooldowns or higher‑tier plans if you’re on a tight deadline.
  • Physical activity is a great way to reset creative energy, even if it turns into a comedy show.
  • Live‑streaming a music performance is surprisingly straightforward once you have your overlay and audio mix scripts ready.

That’s the end of the weekend log. Stay tuned for the Monday launch!