Summary
This tutorial shows how to create 3D motion graphics from plain-language prompts using Claude Fable 5.2, without opening After Effects or a timeline. The creator points to the model’s improved visual understanding, cited as rising from 45% to 73%, as the reason it can interpret loose creative direction more reliably.
The walkthrough covers setup first: selecting Fable 5.2 in Claude and adding the Higgsfield MCP connector, which gives Claude access to video and image models such as Seedance and Kling. From there, the video moves through five levels of increasing sophistication. Level one uses words only, such as a cinematic headline that hits on the beat or a SaaS-style promo built from shapes and text. Level two introduces reference images and website links, with Claude extracting a brand’s colors and tone to produce a 15-second launch promo. Level three treats the user as camera crew, specifying camera moves and lighting, previewing keyframes as stills before committing to a full render, and chaining scenes with liquid transitions.
The key lessons: show references rather than describing them, request stills before video to save time, and keep to one clear camera move per shot. Prompts are shared through the creator’s community.
📺 Source: Zinho Automates · Published November 11, 2026
🏷️ Format: Tutorial Demo
