Generating Scene Portraits

Opening the Portrait Modal

Click a scene’s portrait thumbnail or the Generate / Regenerate button on any scene row. The modal opens with the title “Generate Portrait — {sceneTitle}” and shows the current portrait (if one exists) on the left with all controls on the right.

Portrait Orientation

Scene portraits use a 16:9 landscape aspect ratio by default, producing wide cinematic establishing shots designed for wide-angle cinematic framing. This is the same ratio used for location portraits.

Location Reference Image

This is a feature unique to scene portraits. The system automatically loads the location’s existing portrait from the active production design as a reference image. This reference is sent with the generation request, maintaining visual consistency between the location’s establishing shot and the scene’s establishing shot.

If the scene’s location has no portrait yet, no reference image is sent. For best results, generate your location portraits first, then generate scene portraits — the scene will inherit the visual style and environment from the location.

Prompt Controls

• Image prompt — An editable text area. The default template is “Cinematic establishing shot of a film scene: {description}. No words, text, or lettering anywhere in the image.”
• Style prompt — Read-only, pulled from the active production design. Defaults to “Dramatic lighting, wide-angle lens, cinematic color grading.”
• Model picker — Select which AI model to use for generation.
• Aspect ratio — Defaults to 16:9 for scenes. Choose from preset ratios.
• Negative prompt — Describe what you don’t want in the image. Availability depends on the selected model.
• Seed — A number for reproducible results. Availability depends on the selected model.

Auto-Describe Inside the Modal

If a scene has no description yet, the modal shows an Auto-Describe button. Clicking it generates a visual description from the scene heading context — the setting (INT/EXT), location name, and time of day. The generated description is saved and used as the image prompt. This lets you go straight from an empty scene to a generated portrait in one step.

Uploading an Image

You can upload your own establishing shot image in three ways:

• File picker — Click the upload area to open a file browser
• Drag-and-drop — Drag an image file onto the left column
• Clipboard paste — Copy an image and press Ctrl+V / Cmd+V

Version Navigation

Every generated or uploaded portrait becomes a new version. Use the Older and Newer arrow buttons to browse through the version history. The version counter shows your position (e.g., “3 of 7”).

• Use This Version — Set an older portrait as the current active portrait
• Metadata display — Each version shows the model used, seed value, and negative prompt

Save as Design Default

When you find a seed and negative prompt combination you like, click Save as Design Default to persist those settings to the active production design. Future portrait generations for other scenes will use the same defaults.