GKM platformWhitepaperBack to website
Contents

Project documentation / Whitepaper

AI Short-Drama Production

Model Architecture

GKM platform uses open-source large models to train GKM Dramatist, an AIGC model dedicated to short-drama production.

The model provides the following core capabilities:

  • Script generation: Trained on extensive short-drama script datasets, it automatically generates scripts suited to different cultural contexts.
  • Multilingual adaptation: Supports script and subtitle generation in more than 20 languages, including English, Spanish, Indonesian and Arabic.
  • Video generation: Integrates the latest video generation technologies to enable end-to-end production from script to finished video.
  • Style transfer: Automatically adjusts visual styles to target-market preferences, including realism for European and American audiences, a fresh aesthetic for Southeast Asia and ornate styles for the Middle East.

Industrialized Production Workflow

GKM platform has established a standardized workflow for industrialized AI short-drama production:

Story concept → AI script generation → Storyboarding → AI video generation → Voice-over and subtitles → Post-production editing → Quality review → Platform release.

Computing Resource Scheduling

GKM platform dynamically allocates computing resources according to task priority:

  • Training tasks: Prioritize high-performance GPU clusters.
  • Inference tasks: Use the distributed computing network.
  • Rendering tasks: Use low-cost computing nodes.

Computing resource usage is recorded on-chain to ensure transparency and traceability.

AI Studio: Creator Tools

AI Studio makes AI short-drama production tools available to creators worldwide, enabling third-party creators to publish their work and participate in revenue sharing. By combining production tools with its own streaming platform, GKM platform supports content creation, publishing and monetization within one ecosystem.

Dual-Track Model-Layer Decoupling and the Platform Routing Gateway

The model layer and platform layer implement physical-grade isolation at the logical level. The model layer establishes a Trusted Pipeline using Story Protocol's copyright standards. Only authorized creators may access LoRA model assets certified by official IP publishers, and finished productions must contain immutable cryptographic watermarks carrying the IPA identifier. The platform layer uses common underlying smart contracts to drive its frontend portals, operating compliant streaming of leading-IP adaptations and crowdfunding services for global mass audiences and institutional capital.