Storyboard to Video — GPT Image 2 + Seedance 2.0

How to use it — 2 stages

Stage 1 — Generate the Storyboard 🎬

⚠️ Make sure the Seedance 2.0 group is bypassed first (right-click the group → Bypass Group Nodes)

1. Write your full scene description in the prompt — describe the action like a mini script, beat by beat

2. Hit Queue

3. GPT Image 2 generates an 8-panel storyboard showing your entire sequence

Stage 2 — Animate with Seedance 2.0 🎥

⚠️ Remove the bypass from Seedance 2.0, and bypass the storyboard generator group instead

1. Upload your Storyboard image (from Stage 1) into the "Storyboard" LoadImage node

2. Upload your Character image — a clear reference of your character's face and outfit

3. Hit Run — the VLM node reads both images and writes a guided prompt, then Seedance 2.0 animates your storyboard into a full video, keeping your character consistent throughout

✍️ Writing Your Scene Description

Describe the full action sequence like a mini script — beginning to end.

A ninja sprints across a rooftop beneath the moonlit 
night sky before launching high into the air. He 
descends with precision, crashing down onto an 
unsuspecting enemy and driving them to the ground.

As the enemy falls, the ninja slips behind the corner 
of a structure, disappearing from sight. Cut to a 
close-up of his eyes peering around the wall — focused, 
alert, scanning for danger...

Tips:

• Write it as a sequence of beats — each beat becomes a storyboard panel

• Include camera details → close-up, wide shot, cut to

• Describe character actions clearly and in order

🧩 How character consistency works

1. ImageStitch combines your Character image and Storyboard side by side into one image

2. VLM node reads this combined image along with your original scene description

3. StringConcatenate appends identity rules to the prompt:

@image_1 is the character — maintain facial and 
outfit identity in the generation.

@image_2 is only for rough scene composition. 
DO NOT generate arrows or text from this image.

4. Seedance 2.0 receives the final guided prompt and generates the animation

This means your character's face and outfit stay locked in, while the storyboard only guides the scene layout and action — not the appearance.