Seedance Prompts: Shot-by-Shot Templates to Try

ByteDance describes Seedance 2.0 as supporting multi-shot video, native audio and mixed text, image, video and audio references. Those are model capabilities, not a promise that every CrePal mode exposes every reference input. In a shot list, use one main action per beat and repeat the visual anchors that matter: clothing, object color, location and light. For example, “same red thermos, same cafe table” is clearer than “make it consistent.”

Copyable structure: “Subject: [one subject and fixed details]. Setting: [place, light]. Shot 1: [framing + single action]. Shot 2: [framing + next action]. Audio: [dialogue, ambience or effects]. References: [asset role, only if supported].”

Product ad: “Subject: matte blue reusable bottle with a silver cap. Shot 1: close view as a hand places it on a trail map. Shot 2: wider view of the bottle beside hiking boots. Warm morning light. Audio: soft footsteps and wind; no dialogue.”

Action beat: “Subject: cyclist in a yellow rain jacket. Shot 1: side view approaching a shallow puddle. Shot 2: close view of one wheel crossing it. Overcast daylight. Audio: tire spray and distant traffic.”

Dialogue scene: “Two original adult characters at a bakery counter. Shot 1: customer asks, ‘Is this the last cinnamon roll?’ Shot 2: baker smiles and replies, ‘I saved one.’ Quiet cafe ambience.” These examples have not been run as a benchmark.

Complete product sequence to try: “Subject: one matte blue reusable bottle with a silver cap; keep the same cap and color in every shot. Setting: a wooden trail table in soft morning light. Shot 1, close view: a hand places the bottle beside a folded map; audio: one gentle tap. Shot 2, medium view: the same hand closes the same silver cap; audio: a single cap click. Shot 3, wider view: the bottle stays on the table as the camera pulls back slightly; audio: wind and distant footsteps, no dialogue. Do not add product claims or generated text.” Review the cap, table and sound timing at each transition. If one beat changes the bottle, simplify that beat before extending the sequence.

If your interface accepts reference assets, assign a role to each: character image for appearance, location image for setting, reference clip for camera movement, audio reference for sound. Confirm current input limits in the tool you actually use. Avoid contradictory references, such as “fixed camera” plus a sweeping orbit. For a failed shot, simplify its action and preserve the character description before changing models.

Google Veo, Kling and Seedance have different prompting and generation options; the best choice depends on the shot and current availability. CrePal's video creator says its AI Director selects a model for each scene. Use CrePal to review the proposed shot plan and model choice; do not assume a particular Seedance version is available without checking the project interface. Revise and export after checking audio, continuity and rights.

This is a prompt-shape comparison, not a quality ranking or a claim that all three exact versions are selectable in every CrePal project. Let the AI Director propose a model for each shot, inspect its selection, and override or simplify a shot only when the current interface permits it.

For a three-shot product scene, write one line per shot with subject, action, camera, sound and continuity anchor. Example: “Shot 1: red thermos on a cafe table; hand reaches toward it; locked medium frame; cup noise; morning window light.” In shot 2, repeat the red thermos and the same table before describing the lift. In shot 3, show the thermos placed beside a notebook, then let the scene end. A shot contract makes omissions visible before generation.

If the tool accepts references, assign each one a single job: product appearance, room layout or movement. Do not imply that every Seedance version or CrePal mode exposes the same reference inputs. Review each generated shot for object color, hand continuity, sound timing and direction of travel. If one shot fails, revise that shot's contract rather than rewriting all three. Keep the model version and settings with the prompt so the sequence can be reproduced or sensibly compared later.

If the interface accepts mixed image, video and sound references, assign each asset one job using the Seedance 2.0 reference guide. For a simpler one-shot comparison, reuse the same subject and camera instruction in the Veo prompt examples.

Caption: Character-driven short scenes. CrePal labels this use-case frame an illustrative composite, not a documented result of a prompt in this guide; actual results vary by prompt, model and revision.

The figure stays visually distinct from the blurred street lights. In a character prompt, lock the silhouette, clothing and light direction before specifying a camera move; compare those details across actual generated frames.

Next check: Reject a draft if the character changes shape or costume between shots.

CrePal's AI Director workflow can break a request into scenes and route models per shot. A Seedance prompt with image, video or audio references should therefore assign one job to each asset and check the selected route's current controls before generation. This official workflow description is not a test record for the model-specific templates below.

Related official reading: CrePal's Seedance 2.0 multi-shot guide. CrePal's Seedance 2.0 multi-shot article distinguishes identity drift from style drift and recommends repeating short character, camera, light, wardrobe and background descriptors across shots. Use those as a continuity checklist when adapting the generic Seedance shot templates here; confirm the selected model version and controls before running them. This cited 2.0 article is related guidance, not evidence that another Seedance version behaves identically.

Evidence note added 5 November 2026. The prompts in this guide are editorial examples. The linked CrePal pages document product workflows or illustrative use cases; they do not verify an output from these exact prompts. No prompt-output pair is published here.

To document a result for this topic: Record the model, prompts and clips; compare silhouette, costume and light direction across shots. Keep the exact prompt, model/version, settings, output URL, revisions and human review decision together.

Start with one example from this guide, then compare the result with your intended subject, style and details.

Create with CrePal

CrePal's Terms identify SIGNOO PTE. LTD. as the service provider. The prompt examples here are instructions to try, not recorded generation tests.

• Privacy
• Terms
• Send feedback
• Email support