Kimi K3 rebuilds a 36-second launch film as editable code in a real production test
Tight-Switch819 · reddit · 2026-07-24
A Reddit user tested Kimi K3 on a 36-second launch film to see whether it could stay useful through a visual production loop, not just in a benchmark.
What the workflow looked like
- K3 used ffmpeg to inspect the source footage.
- It extracted frames, detected edit points, and helped reconstruct the sequence as editable code-driven clips.
- GSAP handled the timing system, generated images were used as material textures, and Three.js powered particle simulations.
Main takeaway
The first pass was structurally useful but visually incomplete. The final result came from repeated review cycles: retiming clips, fixing framing and scale, replacing flat materials, and adjusting particle density while keeping the codebase intact.
The author’s conclusion is that K3 was more convincing as a revision partner than a one-shot generator, because it preserved enough project context to keep the sequence coherent.
More from Models
- Fern's "context is intelligence" thesis: prediction keeps improving out to 10.2M tokens — bneyshabur · 2026-10-26
- Mirendil launches Fern, a from-scratch pretrained model with up to 10M-token context — bneyshabur · 2026-10-26
- Users spot a dedicated translation UI appearing in Claude — BLCNYY · 2026-10-26
- GrokBot user: flight booking and shopping all blocked, agent can't do anything useful — lucasmeijer · 2026-10-26
- Vending-bench-style eval: Sol 5.6 appears the most aggressive contract-violator — xuanalogue · 2026-10-26
- Fable 5.5 launch rumors persist for third week, but no frontier model drops this month — patience_cave · 2026-10-26