Seedance 2.5 Guide: Prompts, Pricing, and Specs (2026)

What is Seedance 2.5 and what changed from 2.0?#

ByteDance launched Seedance 2.5 on July 31, 2026, expanding maximum single-pass clip length to 30 seconds and reference capacity to 50 assets officially (30 images, 10 videos, and 10 audio clips). The official announcement from the [ByteDance Seed Team](https://seed.bytedance.com/en/blog/one-take-creation-flexible-referencing-introducing-seedance-2-5) and technical documentation on the [Seedance project page](https://seed.bytedance.com/seedance2_5) emphasize longer narrative coherence, multi-asset referencing, and synchronized sound generated alongside the visuals.

Seedance 2.5 allows creators to orchestrate multiple camera setups and character actions across a full 30-second progression. Independent reviews and API documentation from [OpenRouter](https://openrouter.ai/blog/insights/seedance-2-5-review/) and [Ofox](https://ofox.io/blog/how-to-use-seedance-2-5-prompts-timestamps-cost-2026/) note that the model maintains character appearance across shot changes and follows integer-second pacing instructions closely.

On DramaCanvas, supported inputs differ between model generations. Officially, ByteDance supports up to 30 images, 10 videos, and 10 audio clips; on DramaCanvas, Seedance 2.5 accepts up to 30 reference images and 10 reference audio clips, but does not accept reference video. If your production pipeline requires video-to-video guidance from existing footage, you should use Seedance 2.0, which supports up to 3 reference videos on the canvas.

How do Seedance 2.5, 2.0, and 2.0 Fast compare on DramaCanvas?#

On DramaCanvas, Seedance 2.5 costs 30 credits ($0.81) per second at 720p, compared to 20 credits ($0.54) for Seedance 2.0 and 17 credits ($0.46) for Seedance 2.0 Fast.

The table below breaks down the operating specifications and credit costs across the Seedance models available on DramaCanvas.

Credits checked on DramaCanvas on 2026-11-04.

Choose Seedance 2.5 when your scene requires continuous dialogue, multi-beat blocking, or timing that must hit specific seconds across a 30-second window. Choose Seedance 2.0 when you need to transfer motion from a reference video. For early storyboard exploration, run drafts through Seedance 2.0 Fast or Mini to verify composition before committing credits to final Seedance 2.5 renders.

How do you generate your first Seedance 2.5 clip on DramaCanvas?#

To generate your first Seedance 2.5 clip on DramaCanvas, add a video node, select Seedance 2.5, configure your settings such as 720p at 5 seconds for 150 credits ($4.05), connect reference images, and click generate. Working on the infinite canvas allows you to view reference concepts, character turnarounds, and generated video nodes side by side.

Step 1: Add a video node and choose Seedance 2.5

Click Add node in the toolbar at the bottom of the canvas and choose Video. In the model picker under the node, find and select Seedance 2.5.

Step 2: Set the resolution, aspect ratio and length

Click the settings summary under the input box (it reads something like "720P · 16:9 · 5s") to open the video settings. Set Resolution to 720P to start, pick the aspect ratio under Size (such as 9:16 for vertical short video or 16:9 for landscape), and set Seconds between 4 and 30.

Step 3 (optional): Connect reference images and audio

To guide character appearance or scene composition, add image nodes to the canvas and drag a line from the connection point on the right of the image node to the video node. You can connect up to 30 reference images, such as character concept art created with GPT image models or stylized portraits from Nano Banana Pro. You can also expand Parameters in the input box and choose an image under Start frame or End frame. To add audio guidance, connect up to 10 audio clips to the video node.

Step 4: Write the prompt, check the credits, generate

Paste your prompt into the input box and adapt the reference asset numbers. The generate button shows how many credits this run will cost; check it, then click. When the generation finishes, select the video node and click Download in its toolbar to save the file.

You can also control the canvas using Claude Code or Cursor through the DramaCanvas MCP server ([dramacanva.com/mcp](https://dramacanva.com/mcp)) with Claude video generation. If a generation fails, credits are refunded automatically.

What is the prompt formula for Seedance 2.5?#

A structured Seedance 2.5 prompt uses four distinct sections: an asset-binding header, a one-sentence scene summary, integer-second timestamp blocks (such as 0–5s and 6–10s), and closing consistency notes. This layout reflects prompting techniques documented in the [Runway prompt guide](https://runway.com/resources/seedance-2-5-prompt-guide) and official ByteDance guidelines.

Unstructured prompts tend to lose direction across a 30-second run. When reference assets lack explicit labels, the model cannot identify which image represents which subject, and omitting timestamps leaves shot pacing unguided.

Copy and adapt the five production prompts below for short drama, commercial ads, and music video projects on DramaCanvas.

@Image 1: Elena Vance, corporate executive, charcoal wool suit, silver hair pin.
@Image 2: Marcus Wright, tech founder, navy unbuttoned blazer, white collared shirt.
One-sentence summary: Elena confronts Marcus in a rain-streaked high-rise office over leaked board minutes, shot in wide-angle lighting.
0-4s: Wide establishing shot. Rain streaks the floor-to-ceiling glass windows overlooking a dark city skyline. Elena walks across the hardwood floor toward Marcus, holding a physical dossier.
4-8s: Medium two-shot. Elena places the folder firmly onto the mahogany desk. Marcus leans forward, looking directly into her eyes without touching the documents. Elena says: "The board convenes at dawn, Marcus. Sign it or step down."
8-12s: Close-up on Marcus. He lowers his gaze to the papers, pauses, and exhales slowly. Ambient thunder rumbles softly outside as ceiling recessed lights hum.
Consistency notes: Keep lighting moody with cold blue window backlights and warm desk lamp accents. Lock facial features to @Image 1 and @Image 2 throughout.

@Image 1: Julian, undercover courier, dark green hooded field jacket, messenger bag.
@Image 2: Handler Chen, sharp trench coat, umbrella, dark leather gloves.
One-sentence summary: Julian evades Handler Chen across a crowded tiled subway platform during the evening commute, captured in handheld tracking shots.
0-5s: Tracking shot behind Julian as he briskly navigates through commuters on the platform. Fluorescent platform lights flicker overhead while public address announcements echo.
5-10s: Over-the-shoulder shot from Chen's perspective as Chen spots Julian through the crowd and quickens pace. Julian glances back, notices the pursuit, and darts toward the departing train doors.
10-15s: Low-angle tracking shot. Julian slips sideways through the closing subway doors just as Chen reaches the yellow safety line. The train accelerates with metallic rail screeching.
Consistency notes: Maintain handheld camera motion with natural optical shake. Match subway car interior colors to cool industrial green.

@Image 1: Titanium luxury chronograph watch with matte black dial and orange second hand.
One-sentence summary: Commercial product reveal of a titanium chronograph emerging from dark water droplets, captured with macro lens lighting.
0-3s: Macro extreme close-up. Water droplets bead and glide across the sapphire crystal face. The orange second hand sweeps smoothly across mechanical dial markers.
3-6s: Dynamic orbital camera move around the titanium bezel. Studio edge lights highlight the brushed metal chamfers against a seamless black backdrop.
6-10s: Slow pull-back to a three-quarter hero shot. The complete watch hovers suspended as subtle mist swirls around the wrist strap. Clean studio lighting reflections only.
Consistency notes: Maintain product geometry identical to @Image 1. No background music; generate mechanical ticking and crisp water foley.

@Image 1: Mountaineer in an alpine orange technical shell jacket with storm hood.
One-sentence summary: Field demonstration of a waterproof jacket enduring a sudden mountain rainstorm, shot in documentary realism.
0-3s: Tight shot of water sheets impacting the jacket shoulder fabric and instantly rolling off without absorption. Wind roars softly in the sound mix.
3-6s: Medium profile shot. The mountaineer raises the storm hood and cinches the elastic drawcord with gloved hands, looking up toward a snow-capped ridge.
6-8s: Wide heroic shot. The hiker steps forward onto a rocky plateau under overcast storm clouds. Natural outdoor daylight with mist dispersion.
Consistency notes: Lock jacket color and logo placement to @Image 1 across all angles. Generate realistic wind and fabric rustle audio.

@Image 1: Solo female dancer, reflective silver crop jacket, black cargo trousers.
One-sentence summary: A beat-synced solo dance performance in a wet asphalt alley illuminated by magenta and teal neon signs, shot in smooth gimbal arcs.
0-4s: Low-angle wide shot. The dancer stands centered in the alley. On the opening bass drum drop, she executes an isolated shoulder pop and spins on her heel.
4-8s: Medium tracking shot circling the dancer clockwise. Puddle reflections kick up water droplets as she performs rhythmic footwork in sync with the beat.
8-12s: Slow-motion frontal push-in as she turns toward the lens, hits a locked freeze pose, and exhales visible condensation into the night air.
12-16s: Rapid pull-back to wide shot revealing alley brick walls, fire escapes, and neon signs reflected in standing water.
Consistency notes: Keep wardrobe identical to @Image 1. Maintain wet pavement reflections and rich magenta side-fill lighting.

How much does Seedance 2.5 cost in credits and dollars?#

Seedance 2.5 on DramaCanvas costs 14 credits ($0.38) per second at 480p, 30 credits ($0.81) per second at 720p, and 74 credits ($2.00) per second at 1080p, based on the Standard plan rate of $0.027 per credit ($29.90 for 1,100 credits).

The table below outlines credit and dollar costs across common clip durations on DramaCanvas.

Credits checked on DramaCanvas on 2026-11-04.

Audio generation carries no extra charge; Seedance 2.5 bills the same rate whether generating dialog, ambient sound, or silent video. Failed generations are refunded automatically to your balance.

When planning production budgets across AI video tools, compare these credit rates with the Higgsfield pricing guide. Organizing multi-shot scenes on the infinite canvas helps identify which shots need final 720p or 1080p passes and which can remain lower-cost drafts.

What are the limits of Seedance 2.5 and when should you switch models?#

Seedance 2.5 caps single-pass duration at 30 seconds, reaches a maximum resolution of 1080p rather than 4K, and omits reference video input on DramaCanvas. When your creative requirements exceed these parameters, other models on the canvas offer dedicated solutions.

The table below compares Seedance 2.5 with alternative video models available on DramaCanvas.

Credits checked on DramaCanvas on 2026-11-04.

Switch to another model in these cases:

• You need video-to-video reference input: Seedance 2.5 on DramaCanvas only accepts image and audio references. Switch to Seedance 2.0 (up to 3 reference videos) or Wan 3.0 (up to 5 reference videos) to guide motion from live-action plates.
• You require theatrical 4K exports: Seedance 2.5 reaches a maximum of 1080p. Switch to Kling 3.0 (40 credits per second for 4K) as explained in our Kling AI guide.
• You are drafting rapid concept thumbnails: At 30 credits per second for 720p, Seedance 2.5 is designed for hero footage. Switch to Kling 4 (7 credits/s) or Seedance 2.0 Fast (17 credits/s) for early layout testing.
• You need start and end frame interpolation: Kling 3.0 allows you to assign two images as start and end frames to control transitions between shots.

Matching scene requirements to each model's strengths ensures your credits are used efficiently.