ShotForward LogoShotForward
Core Concepts

Core Concepts

Deep dive into the architecture and logic of ShotForward.

Storyboard

Your active workspace and orchestration canvas. It's not just a static blueprint but a dynamic environment for directing.

  • Linear Flow Visualization: Visualizes the film's linear progression. Supports drag-and-drop reordering for WYSIWYG editing.
  • AI Interaction Hub: Integrated Chat Panel. Give commands like 'Redraw this shot' or 'Generate sad BGM', and the AI Director directly manipulates the storyboard state.
  • Multi-Dimensional State: Synchronizes states for Video (V1), Global Audio (A1), and Dialogue (A2) tracks.

Scenes

Logical containers for narrative. They break down massive scripts into manageable units (usually creating a Time/Location block).

  • Structural Management: Supports Create, Delete, Rename, and Drag-and-Drop. Scene order defines the final playback flow.
  • Agent Context Window: The AI Director thinks in Scenes. This prevents context loss in long scripts and ensures tight logic within each segment.

Shots

The atomic execution unit. A synthesis of visuals, audio, and control parameters.

Visual Anchor

The core is the 'Start Image'. It defines composition and lighting. Video generation effectively animates this static anchor.

Vocal Track (A2)

Shots have a dedicated A2 track. Write scripts here, and the system generates voiceovers using the bound character's voice.

Seamless & Interpolation

Supports 'Start & End Frame' generation. Designate the next shot's image as the endpoint, and AI generates a seamless transition video connecting them.

Assets

Your local resource pool. All generated media and manually uploaded files reside here.

  • Local Storage: Files are stored on your disk. No cloud syncing of raw assets ensures absolute privacy and speed.
  • Drag & Drop Override: Unhappy with AI results? Drag local assets directly onto shot cards to override AI outputs instantly.

Characters

Consistency anchors. In ShotForward, a character is more than a name; it's a set of mandatory constraints.

Visual Anchor

Bind a Ref Image. When you mention the character, the Agent injects this image into the model (via IP-Adapter), locking the facial appearance.

Audio Anchor

Bind a voice sample. All dialogue assigned to this character uses this sample for Voice Cloning, ensuring consistent vocal tone.