Product Development
RankForge
A local, AI-assisted 9:16 ranking and compilation editor for turning many clips or URLs into a reviewable short-form video.
Product DevelopmentRankForgeOverview
RankForge is a local, AI-assisted 9:16 ranking and compilation editor. It takes clips or URLs and helps turn a large source set into a reviewable short-form video.
01 problem → system → outcome
The problem
Reviewing, ranking, titling, labeling, transitioning, commenting on, and rendering many source clips is slow and difficult to keep consistent.
Goals
- Make large clip sets easier to review
- Keep creative control with the user
- Move from sources to a renderable edit efficiently
The solution
The product extracts frames, uses visual AI to score footage, selects the strongest clips, prepares ranking and captions, then gives the user a chance to review and change the edit before rendering.
Workflow
- 01Clips or URLs
- 02Storage
- 03Frame extraction
- 04Visual AI
- 05Scoring
- 06Strongest selection
- 07Ranking
- 08Captions and voice
- 09Editing
- 10Rendering
System design
FFmpeg extracts and assembles media. Qwen3-VL 4B evaluates humor, surprise, visual interest, clarity, and actions. Llama 3.1 8B supports titles and labels, while Ollama and Kokoro TTS provide local inference and voice generation.
Capabilities
- Upload and URL import
- Configurable selection from roughly 100 source clips
- Trim, order, ranking, and labels
- Title styles, numbering, and positioning
- Transitions, mute controls, and media management
- Render retry and MP4 export
Engineering focus
- Making AI scoring interpretable
- Preserving user review before render
- Managing many media assets
- Reliable local rendering
Outcome
A focused editing system that makes it faster to turn a large, messy clip pool into a ranked short-form video without removing the creator from the process.
Next opportunities include Additional scoring presets, More template controls, Expanded editing workflows.
Discuss a similar project