Building AI video shots in 3D first: a Mars landing scene workflow with Intangible AI Canvas
umesh_ai · x · 2026-11-12
umeshai walks through his workflow for a Mars landing scene using Intangible AI's new Canvas: start from a brief and references, build the scene in 3D, experiment with camera positions, even swap the rover's sensor color from blue to amber and regenerate. His favorite part: being able to go back and tweak the actual scene before regenerating shots — previs thinking applied to AI video.
Workflow highlights
- Brief + references → 3D scene build
- Iterate framing/camera angles before any generation
- Tweak scene details (sensor color) and regenerate
Related event: Intangible AI Canvas lets you frame AI video in 3D first(3 posts)→
More from Multimodal
- Firefly's Quick Cut auto-syncs video cuts to the beat, saving hours of manual editing — LudovicCreator · 2026-11-12
- Adobe Firefly's browser-based multi-track video editor: full timeline, no install needed — LudovicCreator · 2026-11-12
- Cutting a 20-second brand teaser entirely in Firefly: 6 clips, beat-synced, no editing skills — LudovicCreator · 2026-11-12
- One Prompt, Under an Hour: Claude + Blender MCP Builds Playable 3D Game — StrainNo9529 · 2026-11-12
- One prompt, under an hour: new Fable builds a 3D action game with controller support — Cojj25 · 2026-11-12
- Izwi: open-source offline voice cloning in Rust, runs on 16GB M1 Pro — zinyando · 2026-11-12