Runway's Sightline Turns One Text Prompt Into a Reusable 360 Virtual Set

Runway has released Sightline, an App Studio tool designed to keep AI-generated locations consistent across camera angles. It creates one 360-degree panorama from a text prompt, then lets users frame still images from different directions within that fixed environment. Because each frame comes from the same source image, walls, lighting, props, and other visible details remain anchored as the camera rotates.

Sightline addresses a common failure in generative video workflows. When a model creates every shot independently, architectural details and object placement can change between cuts. Sightline gives previsualization, compositing, and 3D teams a shared environment from which to produce multiple plates.

Build the set once

A Sightline session starts with a written description of the location. The tool generates an equirectangular panorama, which represents a full 360-degree view from one camera position. Users can then aim a virtual camera within that panorama, adjust the framing, and export the resulting views.

1. Describe the environment with a text prompt.
2. Choose the camera height, lighting state, lens, and aspect ratio.
3. Point the camera with the map-style aim control, then adjust tilt and roll.
4. Save individual frames, the full panorama, or cube-map faces for a 3D workflow.

An orbit mode automates coverage by generating still plates at intervals from 15 to 90 degrees. Configurable tilt rings add vertical angles, allowing a team to create a broad reference set in one pass.

Camera controls at a glance

Shared scenery reduces shot drift

Every exported frame is a projection of the same panorama, so visible features retain their position as the view rotates. That spatial coherence supports several production tasks:

• Previsualization and storyboards that need matching locations across shots
• Background plates for compositing characters, products, or graphics
• Skyboxes and environment maps for Unreal Engine, Unity, Blender, and Gaussian splatting pipelines
• Location references for virtual production and remote planning
• Advertising variations that reuse the same generated set across products or formats

The exported cube faces and equirectangular image can feed common 3D tools as background or lighting assets. They can also serve as visual references for scene reconstruction, projection mapping, or geometry generation.

One viewpoint defines the limits

A panorama captures the scene from a single position. Sightline can rotate the view, change its field of view, and adjust tilt or roll, while camera translation requires depth and geometry that the panorama does not contain. Dolly shots, walk-throughs, and large lateral moves therefore need a separate 3D reconstruction or projection workflow.

Changing the lens setting alters the framing and projection of the existing panorama. It cannot reveal surfaces hidden behind objects or generate physically accurate parallax. The interface also includes an option to allow people in the scene, although any generated figures remain baked into the environment rather than independently controllable actors.

Sightline concentrates on environmental consistency. Character continuity, object animation, and temporal motion still depend on the downstream image, video, compositing, or 3D tools used with its exports.

Access, credits, and hosting

Sightline runs inside Runway App Studio, the company’s system for creating hosted, task-specific tools that can use its image, video, and audio models alongside Brand Kits and Workflows. App Studio is included with paid Runway plans, and creating, running, and hosting an app consumes credits from the user’s existing balance. Each app displays its usage to help teams track spending.

Paid Runway users can access the hosted Sightline app without a separate subscription. Runway’s App Studio overview explains the broader app-building system, credit model, and workspace controls.