AI Powered Artwork Generation

Designing an AI-Powered System for Scalable Artwork Generation

Reducing manual effort in frame selection and multi-format asset creation across thousands of media files.

COMPANY

Amagi

ROLE

Lead UX Designer

EXPERTISE

Design & strategy

YEAR

2025

Description

As media companies scale content distribution across FAST, OTT, and social platforms, generating platform-ready artwork became a key bottleneck. We chose artwork generation as the starting point for Amagi’s AI enrichment journey—given its high volume, design complexity, and manual effort.


“ The editing is fast. The artwork delays everything. ”

This project was the foundation of Amagi’s broader AI initiative, aimed at automating creative and operational workflows across the media lifecycle.

The Problem

At scale, artwork generation for media assets breaks down due to two major bottlenecks:

  1. Frame selection from long-form video:
    Editors were spending significant time scrubbing through hours of video to identify frames that visually represent the content. For media libraries with thousands of assets, this process was not only time-consuming, but cognitively taxing and inconsistent across users.

  2. Rigid source formats and multi-platform requirements:
    Most content providers supplied only 1–2 high-quality images per asset—typically in a standard 1920×1080 format. These had to be manually cropped and resized into multiple aspect ratios (1:1, 9:16, 4:3) for social media, EPGs, and OTT platforms. This transformation effort was repetitive, tedious, and often required external tools or vendor support.

My Role

As the lead designer, I owned the end-to-end design—from vision and UX strategy to research, prototyping, stakeholder alignment, and testing with key partners.

Why Artwork First?

Artwork was the most measurable, high-volume creative problem we could target to validate our AI direction with minimal risk. It had immediate visual impact, consistent inputs (media assets), and required minimal client-side training.

People Impacted (Users)

Content Editors

Needed speed and AI
assistance.

Creative Directors

Required control over branding and visuals.

Marketing Teams

Demanded platform-formatted, promo-ready images.

Platform Admins

Sought automation, auditability, and delivery reliability.

Process

To bring this vision to life, we followed a fast-paced but research-informed design process. We began with internal audits and interviews to understand bottlenecks, then rapidly mapped out a modular workflow aligned with user roles. Given tight timelines, we scoped a focused MVP while ensuring flexibility for future evolution. Feedback from strategic partners was incorporated iteratively, balancing automation with editorial control.

Design Goals

  • Help teams generate artwork quickly without compromising brand quality.

  • Automate repetitive decisions (frame choice, crop formats, overlay placement).

  • Ensure brand consistency across platforms and reduce vendor costs.

Constraints

  • Engineering bandwidth was limited.

  • We needed to ship an MVP in <8 weeks.

  • Had to integrate into Amagi NOW’s existing media management UI.

Ideation & Explorations

Research & Discovery

We conducted internal audits and interviewed editors, marketers, and platform leads to uncover bottlenecks. Recurring themes included:

  • Missed deadlines due to artwork delays

  • High vendor spend

  • Manual rework across formats and platforms

Key Design Principles

  • Editorial Control: AI suggests, humans approve.

  • Speed over Styling: Prioritize turnaround time and operational readiness.

  • Format Flexibility: 16:9, 1:1, 9:16 defaults, with override options.

Key Decisions

  • Ship a minimal but complete flow (extract → review → enrich → save).

  • Allow easy override of AI choices to build user trust.

  • Postpone generative visuals to focus on formatting and consistency.

Solution

MVP Workflow

  • Extract:
    AI detects highlight frames using speech, timecodes, and metadata.

  • Review:
    Editors see top 3–5 frames with confidence scores and preview toggles.

  • Enrich:
    Auto-apply branded templates and preview multiple aspect ratios (16:9, 9:16, 1:1).

  • Save:
    Final images are versioned and stored under each asset.

User Interactions

  • Drag-to-place overlay elements

  • Instant aspect ratio toggles

  • Regenerate frame suggestions

  • Branding templates applied automatically

Each UI decision was made to reduce friction and shorten the feedback loop for editors—from generation to final approval.

What’s Coming Next

  • Global settings for bulk image generation.

  • Layout intelligence for vertical content.

  • AI enrichment expansion to promos and metadata.

Prototype

Results

We’re currently piloting the MVP with early clients and OTT studio partners.

While metrics are still being collected, early feedback signals are strong:

✅ Reduced dependency on external vendors

✅ Faster internal artwork turnaround

✅ Higher visual consistency across platforms

What Clients Care About Most

  • Cost savings (via vendor reduction)

  • Time savings (less manual creative effort)

  • Workflow clarity and ease-of-use

“ This was the first time AI felt like a creative ally—not a threat. ”

This project earned trust with editors and unlocked roadmap buy-in for extending AI into promo generation, metadata enrichment, and smart scheduling.