
How does an idea like "three people running toward an off-road vehicle, fleeing a zombie chase" turn into cinematic shots?
This issue tests the GPT-6 Astra x Higgsfield creation workflow: starting from character, scene and prop design, use Blender to make a gray-model camera previsualization, then hand it to Higgsfield to generate realistic video.
Running into the vehicle, stepping on the gas, tires skidding on the ground, pursuers in the rearview mirror, handgun shooting from the passenger seat, bullet time, and a burly zombie grabbing the tail of the car — we design step by step, and also handle generation issues one by one: misplaced character positions, reversed vehicle direction, similar male voice timbres, and reference video audio being carried over.
The video shows the actual creation and modification process, to see what AI can accomplish and which details still need the director's continued oversight.
This issue's topics: GPT-6 for content creation, agentic AI, AI video generation, AI content creation.
What scene do you most want to shoot with this workflow? Let me know in the comments.
Video Chapters
Zombie Chase Final Preview
Creation Workflow Introduction
Installing Codex and Connecting Higgsfield
Installing Blender
Writing Story and Storyboards
Generating Characters, Scenes and Props
Gray-Model Modeling and Camera Design
Seedance 2.5 Video Generation
Gray-Model vs Final Comparison
Bullet Time and Zombie Grabbing the Tail
Higgsfield Feature Introduction
Supercomputer
Plans and Pricing
Full Film Appreciation
This course comes with a materials pack. Members receive the access link by email after requesting the course.