HyperFrames

HyperFrames · Launch Video Breakdown: Hook, Pacing & Motion Design

Write HTML. Render video. Built for Agents.

0:00 · The Hook · Motion Design Isn't Dead
0:00 / 0:00
AI AgentsProductionSeptember 21, 2026

Counter-intuitive Statement / Industry Debate

18.4

Cuts / min

1:28

Runtime

0K

Views

1,379

Likes

0

Reposts

471

Replies

Scene-by-scene timeline & spoken transcript

  1. The Hook

    Motion Design Isn't Dead

    “Motion design isn't dead. In fact, we think it's just starting to expand and AI will play a huge part in unlocking the possibility for more people to create video.”

    On screen
    Motion design isn't dead. Kailash @kail_designs Motion design is dead!? Saksham Manchanda @saksham91h AI just killed motion design Titouan Gillet @titouangillet_ No AI prompts were used
    Camera
    Static front view with rapid cards stacking and popping in 3D perspective.
    Motion
    Social card layered pop-ins with smooth spring-based physics and UI motion design reel previews.
  2. Problem Agitation

    Quantifiable Benchmarks vs Creative Work

    “You see, every new model comes with a report card, a table of scores against benchmarks. These tests measure how well a model solves hard, checkable problems, like code and agentic work or math and science.”

    On screen
    report card Benchmarks DeepSeek V4.1 GPQA Diamond HLE Codeforces Terminal-Bench-2.1 ONE TASK PASSED FAILED
    Camera
    Slow push-in on a cosmic benchmark table, zooming into code terminal windows and kinetic formulas.
    Motion
    Data table grid overlay on 3D sun sphere, transitioning to code diffs, terminal execution status, and flying mathematical equations.
  3. Problem Agitation

    The Creative Subjectivity Gap

    “But many people are pushing these models somewhere else that is much harder to quantify. Creative work, UI design, writing, video. There's no answer key for those. So people improve by trial and error. Our team set out to change that for video generated from code, a quantifiable way to measure how well a model makes video.”

    On screen
    Creative work. UI design, writing, video. The product development system for teams and agents Chapter One. She kept the key in a coffee tin this doesn't quite
    Camera
    Zoom through a dotted particle sphere expanding into an infinite mosaic grid of UI designs, text cards, and video clips.
    Motion
    Matrix transformation of 3D point cloud sphere to layout grid, pass/fail checkmarks over creative samples, and prompt input box typing interaction.
  4. Product Reveal

    Introducing Code2Video Benchmark

    “Introducing the Code2Video benchmark. I'm sure you're asking, well, who says a video is good?”

    On screen
    Introducing the Code2Video benchmark Who says its good?
    Camera
    Direct live-action medium shot of presenter in an outdoor grassy field with animated title overlays.
    Motion
    Kinetic bold title lockup sliding in over live presenter, followed by floating question headline.
  5. Feature Teaser

    Human Judgments & Model Judge Training

    “We started with people, two generated videos side by side followed by simple questions. Does it hold your attention? Does it match the brief? Thousands of judgments, all made by humans. Then we trained a model on those picks. Show it two videos and a question, and it tells you which one people would choose and why. This model flickers. That one ignores the brief. And because the judge can score any two videos in milliseconds, models can train against it, not just be ranked by it.”

    On screen
    A B Command Bar Shortcuts Does it hold your attention? Does it match the brief? claude-sonnet-4-6 GPT-5.5 P(A beats B) 0.85 Q1 attention Q2 brief Q3 composition Q5 stability Q6 craft This model flickers. That one ignores the brief.
    Camera
    Clean orthographic UI views switching to rapid columnar data stream, followed by 3D spherical judge node orchestrating comparison cards.
    Motion
    A/B comparison card UI, interactive voting buttons, terminal log cascades, live probability score gauge, and floating 3D mosaic of videos orbiting a point-cloud core.
  6. Call to Action

    HeyGen HyperFrames x Google DeepMind

    “Over time, this benchmark will help models produce better videos from code. This is just the beginning. Hyperframes, go make something.”

    On screen
    endless control HeyGen HyperFrames x Google DeepMind Benchmark available through Kaggle
    Camera
    Smooth center zoom into bezier curve motion graph, snapping to center logo lockup.
    Motion
    Kinetic vector curve manipulation, stylized particle cursor snapping logo shapes into position, and clean corporate co-brand reveal.

Related AI Agents Product Launches

Explore all AI Agents launches →
OpenAI
@OpenAI
AI Agents
@OpenAI
OpenAI@OpenAI
Gopuff
@gopuff
AI Agents
@gopuff
Gopuff@gopuff
Elon Musk
@elonmusk
AI Agents
@elonmusk
Elon Musk@elonmusk