Gemini / Seed 2.0 Text Models

Overview

StoryFlow currently exposes two common text-understanding routes:

• Gemini 3 Flash: fast responses for rapid iteration.
• Doubao Seed 2.0 Pro: strong multimodal analysis for heavier tasks.

Both are useful as “analyze first, generate second” upstream steps.

Capabilities

• text generation (scripts, outlines, prompt refinement)
• image understanding (style extraction, reverse prompting)
• video understanding (scene summary, shot breakdown)

Inputs

• prompt (required)
• referenceImage (optional, multiple)
• referenceVideo (optional)

Parameters

These text routes usually need minimal parameters; input quality matters most.

Tips

• Use Gemini 3 Flash when speed matters.
• Use Seed 2.0 Pro when analysis complexity is higher.
• Chain text output into image/video Prompt nodes for stronger control.