VictorInFocus
← All workAI Film / Workflow Study / Action / Spatial Storyboarding

Depth-Map-Style Storyboards for AI Combat(14)

/ About the project

An experiment in using a depth-map-style storyboard to give an AI video model clearer spatial information for a combat sequence.

I developed the scene through Midjourney, a GPT storyboard, a depth-styled visual pass and Seedance 2.0. This was not a true depth map; it was an alternative storyboard treatment designed to describe separation, staging and camera depth more explicitly.

The resulting movement felt more physically grounded, particularly in the contact, positioning and continuity between the characters. It is still an exploratory workflow, but the improvement was strong enough to keep developing.

/ Year

2026

/ Tools

Midjourney · GPT · Seedance 2.0

/ On X

View original post

50 likes · 3 replies via X

A Spatial Storyboard, Not a Depth Map

The key experiment was an alternative storyboard pass that looked like a depth map without functioning as true depth data. I used tonal separation to describe which figures and surfaces should read as nearer or farther from camera, giving the video model a more explicit picture of the scene's spatial structure.

Original cinematic combat still used to establish the sequence and composition.
Storyboard development frame used to define character placement and action.

More Believable Contact

I moved from Midjourney to a GPT storyboard, then created the depth-styled pass before animating in Seedance 2.0. In the finished clip, character spacing, physical contact and movement through the frame felt more coherent. It is not a definitive technical test, but it offered a useful way to give spatially complex action more structure before generation.

Depth-map-style storyboard frame separating characters and scene depth through tonal values.

New enquiry

Work with VictorInFocus

Add budget, timing and project type Optional