Lexus — AI R&D
Made
A set of speculative visuals and motion concepts for Lexus, made as self-initiated R&D.
Needed
An answer to whether AI could change brand storytelling: faster ideation and new pipelines, without diluting a premium brand.
Worth
Tested workflows, and a body of concept visuals: where a future Lexus campaign could go, and how fast it could get there.
The problem
Generative tools move faster than the pipelines a brand already has. The question was whether art direction and automation could work together to sharpen a luxury brand’s identity rather than flatten it into generic AI imagery.
What I built
Node-based workflows in Weavy tied image generation, video models and 3D together into one repeatable pipeline. Stills set the look, video models carried it into motion, and Blender kept form and product accurate.
How it was made
The motion starts in Blender. I block the car and the camera there — plain geometry, no look — only to fix the speed and the camera move.
That blocking pass is the reference. Everything finished was rendered in ComfyUI, with Seedance 1.5 carrying the frames.
My part
I led the research: the concept, the creative direction, the Blender blocking, and the Weavy and ComfyUI graphs behind every finished frame.
3D and motion tests
The blocking pass, then two renders made from it. Short, silent, on a loop.
Stills
Where the car could sit, and what light it holds. Open any one full size.
Made with
- Tools
- Blender
- ComfyUI
- Seedance 1.5
- Weavy
- Kling
- Veo
- Runway
- Nano Banana Pro
- Art direction & AI
- Artem Maslov