Summary
Generating AI video can feel like a lottery: write a prompt, hope the model understands, and spend more credits when it does not. Julia McCoy looks at Topview’s new 3D Builder, a tool that brings previsualization, the rough 3D blocking used in big-budget filmmaking, to AI video production.
Inside Topview Film Studio, users describe a scene in a text prompt and choose a model, such as GPT-6 Astra for more advanced creative interpretation or Qwen 3.8 Flash as the low-cost option at roughly 10 cents per scene. The agent builds a 3D blockout with labeled objects, places a character, and sets up cameras automatically. The video then shows how to edit the scene by changing animation presets, adding keyframes on a timeline, placing props like street lights, and creating custom camera angles with follow and look-at controls.
McCoy notes that building scenes manually costs zero credits, with charges only when the agent generates a scene or the final video is produced. Because every shot uses the same 3D set, locations stay consistent across cuts, a common weakness when each shot is prompted separately. The video covers exporting snapshots and renders and wiring them into a video node as references.
📺 Source: Julia McCoy · Published October 13, 2026
🏷️ Format: Tutorial Demo
