Blog
Articles with first-hand material only: Lab benchmark results, prompt experiments with their outputs, how reference-video control behaves in practice, and workflows we have actually run. If it could have been written without generating anything, it is not published here.
Nothing published yet
The first posts will be Lab results, not press releases
We publish only when we have first-hand material: a completed Seedance Lab run, a prompt experiment with the outputs attached, or a production workflow we actually used. Until then, the pages below already hold everything we can verify.
Editorial policy
One primary page owns each search intent on this site. Blog posts support those pages; they never duplicate a guide's title or purpose. Every post cites its sources, names the model versions and dates involved, and carries an updated date when facts are re-checked. We do not publish automatically generated rewrites of announcements, and we do not publish posts whose only content is a claim we cannot show.
- Good: a Lab run write-up with all outputs, a prompt experiment with before/after clips, a 2.5 vs 2.0 comparison built from generated outputs, a production workflow with settings.
- Not published: 'Top 10 Seedance tips' with no generations behind them, release-day rewrites, speculation about unannounced versions presented as news.