Skip to content
All projects
Photo & Video2025AI-powered

AI Video Editor

Generate and edit video with state-of-the-art models.

  • iOS
SeeDance, Kling
Models tested
Credits
Gating
Cost/latency/quality
Axes

About

An AI video editing and generation app. I shipped feature updates and the credit-check system that gates generation for live users, and ran the R&D comparing video generation models — SeeDance Pro and Kling 1.6 Standard among them — on cost, latency and output quality before anything was integrated.

What shipped

  • Credit check and quota system for live generation users
  • R&D and benchmarking across SeeDance Pro and Kling 1.6 Standard
  • Feature updates shipped to production
  • Prompt and parameter tuning for consistent output quality

How it was built

  1. Model R&D

    SeeDance Pro and Kling 1.6 Standard benchmarked on cost, latency and output quality.

  2. Credit system

    Quota and credit checks gating generation for live users.

  3. Tuning

    Prompt and parameter work for output consistent enough to ship.

Challenges & solutions

  • ProblemVideo generation is expensive enough that the wrong model choice is a business problem, not a technical one.

    SolutionModels were benchmarked on cost and latency alongside quality before integration, so the decision was made on all three axes rather than on demo reels.

Built with

  • AI Video Generation
  • SeeDance Pro
  • Kling 1.6
  • Credit System