EV comparison videos for TikTok
Head-to-head EV comparisons that send viewers to TrimIndex.
What it is
The point of this build is traffic for TrimIndex. Each video compares two electric cars, reads the comparison aloud and closes on a “TrimIndex Verdict” card. A Remotion scene shows each car's figures, and a pipeline turns a script into a voiceover, word timings and captions. Hermes Agent built the first comparison from TrimIndex's local database.
By the numbers
What it means for your team
A business with its own data can turn it into short videos the same way: the figures come from its records, voice and captions run locally with no per-minute bill, and every video ends by sending the viewer to the business's site. A person still reviews each video before it is posted.
Status
Both renders so far are landscape, 1920 × 1080; the vertical 9:16 frame TikTok uses is first on the to-do list. The cars' figures are fixed in the scene rather than loaded per video, the script writer is a placeholder, and nothing posts to TikTok. No video has been posted, so there is no traffic to report.
Comparison scene
A Remotion scene sets two cars side by side, each with its year, make, model, trim, EPA range and efficiency, then closes on the TrimIndex Verdict card. The first one compared a 2022 Ford Mustang Mach-E with a 2023 Mercedes-Benz EQS SUV.
Voice
The pipeline reads the script with Piper, a text-to-speech voice that runs on the Mac, so there is no per-minute voice bill.
Captions
Whisper transcribes the voiceover to get a timestamp for every word. A caption component highlights each word as it is spoken, with line breaks sized for a vertical phone screen.
Render
Remotion renders the frames from React components and FFmpeg encodes the result. A Python orchestrator runs the steps in order.
Checked on 2026-10-08
Each stage was run again: Piper wrote a voiceover, Whisper returned every word with its timing, Remotion rendered and encoded a test clip in 8 seconds, and the comparison scene rendered a frame.