VictorInFocus
← All workAI Previs / Cinematography / Visual Continuity / Image Generation

Camera Coverage from a Single Still(09)

/ About the project

I used one cinematic still as the visual source for an entire set of camera angles, prompting GPT Image 2 to move around the scene while preserving its identity.

The experiment produced reverse-angle, frontal and close-up coverage without rebuilding the character or location. Each prompt changed the camera position and framing while holding onto the same wardrobe, architecture, amber light and film texture.

I approached the process as a previs study: testing whether a single frame could become a coherent visual world with enough continuity to support editorial choices before animation or production.

/ Year

2026

/ Tools

GPT Image 2

/ On X

View original post

43 likes · 5 replies via X

0 reposts · 0 views added by VictorInFocus

A man in a dark coat seen from behind as he enters an amber-lit room, generated as reverse-angle coverage from a single source still.

Moving the Camera, Holding the World

The goal was not simply to generate variations. I wanted to treat GPT Image 2 like a virtual camera: changing viewpoint, shot size and composition while keeping the underlying scene intact. The character, coat, doorway, textured walls and yellow-green grade all had to feel as though they belonged to the same moment.

Frontal medium shot of the same man standing in the doorway under matching amber light.
Tight profile close-up of the same man, preserving the wardrobe, lighting and filmic texture of the wider views.

Previs as a Continuity Test

Starting from one still made every new angle a continuity problem. I prompted for specific camera relationships rather than isolated images, then judged each result by whether it could cut against the others. The exercise became a compact previs workflow for exploring coverage before committing to motion.

New enquiry

Work with VictorInFocus

Add budget, timing and project type Optional