Seedance Mode
Starting frame for Seedance 2.0 to animate. JPEG, PNG or WebP up to 10 MB.
Describe the scene, motion, camera movement, atmosphere and action for Seedance 2.0.
Higher resolution may take longer to generate.
Auto infers the format when possible.
Adds synchronized sound effects and ambient audio when supported.
Sign in to spend tokens
Reference Image is required
Your generated video will appear here
Seedance 2.0 is ByteDance's video model, available on Prizmad in three modes — text-to-video, image-to-video, and reference-to-video with up to 9 tagged reference images. Clips run 5, 10, or 15 seconds at 480p or 720p, with optional generated audio, all with one token balance.
Seedance 2.0 is ByteDance's video generation model. On Prizmad it runs in a single workspace with three modes: Text-to-Video builds a clip from a prompt alone, Image-to-Video animates an uploaded starting frame (JPEG, PNG, or WebP up to 10 MB), and Reference-to-Video takes up to 9 reference images that you address directly in the prompt as @Image1, @Image2, and so on — a precise way to tell the model which product, person, or style element goes where in the scene.
Output is flexible where it counts for ad work: durations of 5, 10, or 15 seconds, 480p or 720p resolution, and seven aspect-ratio options including 9:16 vertical, 16:9 landscape, 1:1 square, 21:9 cinematic, and an Auto setting that infers the format when possible. An optional audio switch adds synchronized sound effects and ambient audio to the clip when supported, so a product reveal can arrive with its soundscape already attached.
On Prizmad, Seedance 2.0 sits alongside Luma Ray 3.2, , , image tools, AI avatars, voiceover, and music using . There's no ByteDance account to create and no API key to manage — you pick a mode, write the prompt, choose duration and resolution, and the clip lands in your asset library in about three minutes, ready for the video wizard or a direct download.
Image-to-Video turns one still product shot into a moving reveal — a slow push-in with realistic reflections and subtle background motion — without a camera or studio.
Generate 9:16 clips with energetic camera movement and a scroll-stopping finish, purpose-built for TikTok, Reels, and Shorts placements rather than cropped from landscape.
Reference-to-Video with @Image tags keeps your actual packaging, colors, and product details in the generated scene — critical when the ad has to match what customers receive.
The 15-second duration fits a complete mini-narrative — hook, product demo, and payoff — in one generation instead of stitching multiple clips.
With generated audio on, lifestyle scenes arrive with synchronized sound effects and ambience, giving organic-style posts a finished feel out of the box.
Text-to-Video from a prompt alone, Image-to-Video from an uploaded starting frame, and Reference-to-Video built around up to 9 reference images — all in one workspace.
In Reference-to-Video mode, refer to uploads as @Image1, @Image2, and so on directly in the prompt, so the model knows exactly which reference drives which element of the scene.
Choose 5, 10, or 15 seconds per generation — the 15-second option gives room for a full hook-demo-CTA ad structure in a single clip.
9:16, 16:9, 1:1, 4:3, 3:4, and 21:9 cover vertical social, feed, and cinematic placements, and the Auto setting infers the format when possible.
Flip the audio switch on and Seedance 2.0 adds synchronized sound effects and ambient audio to the output when supported — off by default if you plan to add your own voiceover or music.
Generations run asynchronously with an estimated time of about three minutes, so you can queue multiple concepts and compare them side by side.
Open the Seedance 2.0 tool in Prizmad's AI Studio and pick a mode: Text-to-Video, Image-to-Video (the default), or Reference-to-Video.
For Image-to-Video, upload the starting frame — JPEG, PNG, or WebP up to 10 MB. For Reference-to-Video, upload up to 9 reference images.
Write your prompt describing the scene, motion, camera movement, and atmosphere. In Reference-to-Video mode, mention @Image1, @Image2, etc. to bind each reference to its role in the scene.
Choose resolution (480p for speed, 720p for quality), duration (5, 10, or 15 seconds), and aspect ratio — or leave the ratio on Auto.
Decide whether to enable generated audio. Leave it off if you'll layer Prizmad voiceover or music on top later.
Click Generate. The clip returns in roughly three minutes and lands in your asset library.
Seedance 2.0 is listed at from 141 tokens (480p 5s) for the default configuration, and the cost scales with your settings — a 5-second 720p clip bills 303 tokens. The workspace shows the exact cost before you generate.
Seedance 2.0 is unlocked on the same token balance as Luma Ray 3.2, Bernini-R, image tools, avatars, voiceover, and music — one token balance, no ByteDance account or API key.
If your tokens run out mid-campaign, buy a one-off top-up directly from the top-up page. Every generated clip is yours with full commercial rights.
On Prizmad it's available to any account with enough tokens rather than free-standing: generations draw tokens from your token balance, starting at 141 tokens (480p 5s) for the default configuration and scaling with resolution and duration.
Cost depends on resolution and duration. The catalog lists from 141 tokens (480p 5s) for the default setup; as concrete reference points, a 5-second 720p clip costs 303 tokens. The workspace recalculates the exact price as you change settings.
Image-to-Video animates a single uploaded frame — the image becomes the literal starting point of the clip. Reference-to-Video accepts up to 9 images as identity and style references that you cite in the prompt as @Image1, @Image2, and so on; the model builds a new scene that keeps those referenced elements consistent.
Yes, optionally. The 'Generate audio' switch adds synchronized sound effects and ambient audio to the clip when supported. It's off by default, which suits workflows where you add Prizmad voiceover or background music afterward.
Clips can be 5, 10, or 15 seconds at 480p or 720p, in seven aspect ratios (9:16, 16:9, 1:1, 4:3, 3:4, 21:9) plus an Auto option. Prizmad exposes the 15-second duration across all three modes.
Both generate video from text or images on the same token balance. Seedance 2.0 offers up to 15-second clips, 9-image reference mode with @Image tagging, and optional generated audio at up to 720p; Luma Ray 3.2 tops out at 10 seconds but reaches 1080p, offers 12 aspect ratios, and adds a seamless-loop option. Testing both on one token balance is the practical way to pick per campaign.
Yes. Videos generated on Prizmad carry full commercial rights — run them as paid ads on TikTok, Meta, and YouTube, and use them on product pages and social channels with no extra licensing.
The workspace estimates about three minutes per clip. Higher resolution may take longer, and generations run asynchronously so you can queue several variations at once.
Published 2026-07-16 · Last updated 2026-07-16