Debugging Seedance 2.0: Face Blocks, IP Warnings, and What to Do
Two Seedance 2.0 filters block a lot of first tries: the face guardrail on real-person references, and the copyright guardrail on trademarked characters. Both live at the model layer.
Every Seedance 2.0 post, filterable by category, sortable by date, searchable by keyword.
Two Seedance 2.0 filters block a lot of first tries: the face guardrail on real-person references, and the copyright guardrail on trademarked characters. Both live at the model layer.
Standard T2V is $0.3034/sec, Fast is $0.2419/sec for roughly 20% off. Here is the token formula, a 100 clip budget table, and when Fast loses quality.
Seedance 2.0 image-to-video holds wardrobe, face, and lighting for a full 15 seconds when you feed it the right still. The still does more work than the prompt.
Async queue pattern for Seedance 2.0: submit with webhooks, poll for status, track cost per job, and retry around the face filter without burning budget.
generate_audio is default true on Seedance 2.0. Here's when to keep it on, when to flip it off for a silent track, and what the lip sync tuning means for non English dialogue.
How to actually use the 9 image, 3 video, 3 audio reference surface on reference-to-video: file weighting, mixed vs image only, and when to skip audio refs entirely.