Agency made a 60-second client video almost entirely with AI — 20+ revision rounds, tools weren't the problem
247Labs_Inc · reddit · 2026-10-15
An agency lead produced a 60-second fintech explainer almost entirely with AI: Higgsfield for stills and clips, ElevenLabs for voice, ffmpeg for editing, plus Python scripts via Claude Code to catch frozen frames and shake, letting him rebuild the whole video in minutes.
The tools weren't the problem — revisions were: a signed-off storyboard got overridden by a script-first approach; a client image meant as final was treated as direction; models forget context, so the character swapped trousers for a skirt and changed suitcases (fixed by generating every shot from one approved image with locked props); real airline tails appeared creating legal risk; usable footage ran 48s against a 71s script forcing shot reuse; and a frame-by-frame-checked stabilized clip still shook badly in motion.
Next time: lock the storyboard before animating, label every client image as guide or exact, one reference image per character, agree voice and pronunciations early, budget music licensing from day one, and review every cut full-size while playing.
More from Multimodal
- Meshy offers Hermes users a month of Pro with API access for $1 — Teknium · 2026-10-23
- Customuse launches Smart Mesh AI with geometry-aware adaptive character clothing tool — SimplyAnnisa · 2026-10-23
- AI-generated movie poster 'The Chase' crams a wolf, a plane and muscle cars into one chase — Aggravating-Sir-115 · 2026-10-23
- SteerTTS-edge: an 11.84MB steerable TTS with unchanged WER running 19x realtime on CPU — Working_Hat5120 · 2026-10-23
- ElevenLabs' Chimera pairs Eleven v4 voice with Seedance video to make character images speak — AIwithGhotai · 2026-10-23
- Open-Source Tennis Models Plus Claude Turn Video Into a World-Class Coach — tweetsatpreet · 2026-10-23