Core Features

StoryFlow is a node-based, multi‑model creative workstation. Below is a concise overview of what it does and how to use it effectively.

Highlights

Infinite Canvas: Pan, zoom, and freely arrange nodes to design complex pipelines.
Visual Programming: Connect inputs and outputs; data flows through your graph without code.
Multi‑Model Engine: Use Gemini (text), Banana Pro/Z‑Image (image), multiple video models (Sora/Veo/Hailuo/Kling/Seedance/Wan), and audio models (Suno/MiniMax).
Smart Wiring: Color‑coded handles (Text=🟡, Image=🟢, Video=🔵) and input validation prevent mismatches.
Dynamic Parameters: The Prompt Node shows model‑specific controls like aspect ratio, resolution, and duration.
Groups & Batch Results: Create and ungroup frames with shortcuts; batch generations automatically land in a shared group. Drag a result node out of the frame to detach it and keep large runs tidy.
Cloud Workflows: Auto‑save, snapshots, public sharing, and remixing.
Asset Library: Drag‑and‑drop images/videos and reuse across projects.
Security by Design: Backend‑for‑Frontend proxy; no API keys in the browser.

Input Text: Store outlines, prompts, and context.
Input Image: Upload files or paste URLs; auto‑resize to aspect ratio.
Input Video: Import clips for reference or remix.
Input Audio: Import audio for audio-capable generation flows.
Prompt (Process): The “brain” that calls AI models; accepts text and media; exposes parameters.
Split Output: Turn a list/paragraphs into multiple text outputs (scenes/items).
Result: Universal viewer for text/image/video with quick download.
Global Context: Store project-wide constraints and reuse across prompts.

Gemini 3 Pro / Flash (Text): Reasoning, planning, and prompt refinement. Accepts optional reference images/videos for analysis.
Banana Pro (Image): High‑fidelity text‑to‑image/img‑to‑img; up to 8 reference images; resolutions 1K/2K/4K.
Z‑Image (Image): Lightweight image generation with basic aspect ratio options.
Sora 2 (Video): Text‑to‑video, image‑to‑video (First Frame), and video remix. Duration 10s/15s.
Sora 2 Pro (Video): Higher‑end video generation for longer clips. Duration 15s/25s.
Veo 3.1 Fast (Video): Faster results; supports First/Last frames and a Reference mode.
Hailuo 2.3 (Video): High‑fidelity image‑to‑video model for realistic motion, lighting changes, and expressive faces; supports 6s/10s and 768P/1080P.
Kling 2.6 (Video): Text‑to‑video with optional sound; supports aspect ratios 1:1, 16:9, 9:16 and 5s/10s durations.
Seedance 1.5 Pro (Video): Cinematic video generation with flexible durations and optional audio generation.
Wan 2.6 (Video): Versatile generation supporting text, image, and video reference (remix).
Suno V5 (Audio): Music generation, extend/remix, and lyric‑driven creation.
MiniMax Voice Clone (Audio): Clone a voice from an audio sample for reuse.
MiniMax TTS (Audio): Text‑to‑speech generation (supports preset and custom voice IDs).

Press Space to create a Prompt node.
Type a prompt, e.g., “A cyberpunk street market at night, neon lights, raining”.
Choose a model:
- For images: Nano Banana Pro (try aspectRatio=16:9, resolution=2K).
- For video: Sora 2 (pick duration=10s).
Click Run. Results appear in a connected Result node.
Chain nodes: connect the Result into another Prompt to iterate or switch models.

First/Last Frames (Veo): Connect 1–2 images to control start/end frames. 1080p requires 16:9.
Reference Mode (Veo): Use REFERENCE_2_VIDEO with veo3_fast and 1–3 images; 16:9 only.
Sora References: The backend makes local images publicly accessible when needed; no manual upload required.
Banana Pro Quality: Use multiple reference images (up to 8) to guide style and composition.
Remix: Feed an existing video into Prompt (Sora) to repaint while preserving motion.