Seedance 2 in practice: from prompt to shot in five minutes

Seedance 2 is the cinematic-control video model wired into TrendRemix's video pipeline. Unlike free-form generators where the model decides everything from camera move to scene composition, Seedance 2 responds to grounded instructions and respects them. That makes it the right model to pair with a TrendRemix structural blueprint.

Before you start

• Have a reference video already analyzed in TrendRemix, or have a finished blueprint on the canvas
• Decide your duration. The default is 5 seconds; longer durations cost proportionally more credits
• Decide aspect ratio (16:9 for YouTube, 9:16 for TikTok / Reels, 1:1 for square ads)

Step 1 — Start from a shot node

On the canvas, every analyzed segment becomes a shot node with a pre-filled aiPrompt field. The simplest way to generate is to right-click the shot you want to recreate and choose "Generate video." TrendRemix bundles that shot's prompt along with the global video prompt and your latest chat instructions, then sends it to Seedance.

Step 2 — Read what Seedance does well

• Camera moves: dolly in, dolly out, tracking shots, slow push, tilt up
• Stable subject motion in mid-distance and full-body shots
• Lighting transitions (sunlight to neon, daylight to magic hour)
• Material rendering on hard surfaces (glass, water, metal)
• Consistent character framing across a single 5-10 second shot

Step 3 — Read what Seedance still struggles with

• Close-up hand articulation (rings, gripping, gesturing) — improving but still imperfect
• On-screen text that needs to remain legible — overlay text in post instead
• Character identity preservation across cuts — handle continuity manually with multiple shot nodes
• Crowd scenes with more than 10 distinct people

Step 4 — When to use Seedance Fast

Seedance Fast trades fidelity for speed. Use it when you are exploring camera options, A/B testing alternate openings, or rendering many short variants before locking the hero shot. Once you find the angle you want, regenerate it with full Seedance 2 for the final output.

Step 5 — Avoid common pitfalls

• Do not write the prompt as a screenplay with stage directions; Seedance reads it as one fluid instruction
• Specify mood and lighting in the same sentence as the action, not separately
• If your reference uses a specific lens (35mm, anamorphic), include it explicitly in the prompt
• Avoid contradicting the global video prompt — when in conflict, the shot-level prompt wins

Where to go next

Drop a reference video into the workbench, let the analysis finish, and right-click any shot to fire off your first Seedance generation. Total time from cold start to first rendered shot: about five minutes including upload.