AIAstra’s potential for blending 3D assets with live-action footage
An author writing for Every argues that control has held back AI video tools—and says Astra’s Blender demos suggested the model could better “see” and refine visual work.
TLDR
An author writing for Every describes months of experiments with ComfyUI, Python and motion-design tools. Generative workflows made it hard to change one element without disturbing the rest, they say, while code-writing models struggled to interpret the visuals they were editing. Astra’s detailed Blender demos led them to suspect it could “see” and iterate more effectively. Their focus is compositing—combining 3D assets with live-action footage—with commercial-grade quality as the goal.