Skip to main content

Uncensored — This model declares no content filter. SpicyAPI forwards the request without adding a platform content scan; legal and consent duties remain with the customer.Browse every model at this filtering level
bytedance/seedance-2.0-spicy/image-to-video

For Seedance 2.0 Spicy image-to-video, image_url alone satisfies the required inputs; prompt is optional and can specify the movement to add to the still.

From $0.114/s5% below the Spicy list price
Run model

Workbench

Input

Image to video
468/5000
A useful Seedance 2.0 prompt describes one visible movement, one camera direction and any desired ambience, leaving the opening still to establish the setting. The motion prompt for this 2.0 Spicy task accepts up to 5000 characters. at most 5000 characters
1/1

Drop files or media links, paste, or browse · 1 file · Max 10 MB · JPG · PNG · WEBP · GIF

  • 1
1 / 1 files
Choose a sharp opening image for 2.0 Spicy and check its framing, lighting and usage rights before you animate it; increasing output resolution cannot guarantee recovery of missing source detail. Each side of a 2.0 Spicy input image must be 300–6000 pixels.
0/1

Drop files or media links, paste, or browse · 1 file · Max 10 MB · JPG · PNG · WEBP · GIF

0 / 1 files
On 2.0 Spicy image-to-video, last_image_url supplies a target ending alongside image_url; matching first and last images can help plan a loop but do not guarantee a seamless join. Each side of a 2.0 Spicy input image must be 300–6000 pixels.
The 2.0 Spicy image endpoint offers six fixed aspect ratios plus adaptive, its default; adaptive follows the opening image, while a fixed ratio can change the framing.
affects price
Seedance 2.0 Spicy generates 4–15 whole seconds at a time, with 5 seconds and 720p selected unless you override them; 480p, 1080p and 4k are also exposed here.
Durationaffects price
Seedance 2.0 Spicy generates 4–15 whole seconds at a time, with 5 seconds and 720p selected unless you override them; 480p, 1080p and 4k are also exposed here. no lower than 4 · no higher than 15
Generate audioaffects price
Seedance 2.0 Spicy exposes generate_audio, enabled by default, for sound generated with the picture; review the soundtrack as well as the visual result.
Seed
This 2.0 Spicy endpoint accepts seed for comparisons with the same random source; matching seeds cannot guarantee identical pixels or frames across requests. no lower than 0 · no higher than 2147483647
Safety checkerOff

SpicyAPI does not run a safety checker on this model. Your prompt and media go to the model as written, and the model itself decides what comes back.

Checking your account before enabling submission and pricing.

Output

Ready for a new run
Example result
Extend
Video translation
Video editing

Related models and next tasks

Missing the model you need?

Tell us what to add and where you would use it. Requests go to the people who decide what gets onboarded next.

These previews show selected examples. Try your own prompt and settings in the Playground.

The full 2.0 Spicy contract: 4k, selectable framing, fifteen seconds

The Seedance 2.0 Spicy image to video API animates one opening still into a clip of 4 to 15 seconds at up to 4k, with a soundtrack rendered in the same pass as the frames. image_url is the only required field; duration_seconds, resolution, aspect_ratio and generate_audio carry defaults and prompt is optional, so a first call can be one URL. Fast and Mini Spicy stop at 720p; this is the 2.0 Spicy edition that reaches 4k.

Open the playground
The full 2.0 Spicy contract: 4k, selectable framing, fifteen seconds
Clip length
4–15 whole seconds, default 5
Resolution ceiling
4k, with 480p / 720p / 1080p below it
Framing
seven aspect_ratio values, adaptive by default
Opening frame
image_url required, last_image_url optional beside it
Audio
generate_audio on by default, same pass as the frames
Prompt
optional, up to 5,000 characters
Framing
Aspect ratio is yours to setSeedance 2.5 Spicy publishes no aspect_ratio property at all, so framing is not a value the caller sets there. This edition takes one of six fixed ratios, or adaptive, which keeps the geometry of the opening still. Name a ratio when the placement is already decided, then check the subject still sits where you want it inside the wider or narrower frame.
Ceiling
4k sits at the top of a four-step ladderRefine the shot at the lower resolution, then render the selected version at the delivery resolution with the same inputs and seed. Changing resolution starts a new generation, so review the final staging and motion before delivery.
Baseline
The reference point for Fast and MiniFast Spicy and Mini Spicy declare the same seven properties, the same 4 to 15 second range and the same seven aspect_ratio values. resolution is the one enum that narrows on those two, stopping at 720p, so a body written against this endpoint runs on either of them once you keep the resolution value inside their set.

What the Seedance 2.0 Spicy schema actually allows

Fifteen seconds is room for a whole beat

Fifteen seconds is room for a whole beat

duration_seconds takes whole seconds from 4 to 15, three more than the twelve-second ceiling on Seedance 1.5 Pro Spicy. Those extra seconds are what let an action begin, develop and settle inside one generation, so a scene does not open with the movement already underway. Write the prompt for the length you asked for: a four-second idea stretched to fifteen usually repeats itself.

Adaptive framing, or one of six fixed ratios

Adaptive framing, or one of six fixed ratios

aspect_ratio defaults to adaptive, which follows the proportions of the still you sent, so a portrait frame comes back portrait with nothing else set. Name 16:9, 9:16, 4:3, 3:4, 1:1 or 21:9 when the output has a fixed home and the still does not match it. A named ratio asks the model to reframe the scene, so confirm the subject has not drifted toward an edge.

  • A closing frame, on one condition

    last_image_url pins the final picture as well as the first, so the clip has to arrive at something you chose. The schema accepts it only in the same request as image_url, and a closing frame sent by itself is refused before generation. Two stills that disagree sharply on framing or light leave a lot of ground for the model to cover in between.

  • Sound arrives with the picture

    generate_audio is on unless you set it to false, and the soundtrack is produced inside the same task as the frames. There is no second call to line up and no separate audio job to poll, so room tone and movement land against the cut you are already reviewing. Turn it off when the sequence will be scored in post.

  • Keep the seed fixed for a clearer comparison

    Reuse the same image_url to carry the character’s face, wardrobe and lighting into each clip. Keep seed fixed when comparing motion prompts, and save both fields with each selected take.

Where the standard Seedance 2.0 Spicy edition fits

  • Compose for the shape you are delivering

    Fixed placements

    Compose for the shape you are delivering

    A feed tile, a story slot and a widescreen trailer are three different frames around the same scene. Set aspect_ratio up front and the clip is generated at that shape, so nothing has to survive a crop afterwards. adaptive covers the case where the opening still was already cut for its destination.

  • Give a romantic beat an opening and an ending

    Romantic narrative between adult characters

    Give a romantic beat an opening and an ending

    One authorized still holds identity, wardrobe and light while the seconds this edition adds over 1.5 Pro Spicy carry the whole arc: the approach, the turning point and the quiet that follows. Keep every character's adulthood and the consenting context unambiguous in the prompt, since the wording is the part the model reads.

  • Set the baseline before testing Fast or Mini

    A/B comparisons across editions

    Set the baseline before testing Fast or Mini

    Keep the prompt, seed, duration and resolution unchanged, then vary one setting at a time. On image-based tasks, reuse the same source images. Compare the staging and motion across a few takes to see whether the change helps.

  • One field in, an async task back

    Developers wiring up a first call

    One field in, an async task back

    Every property except image_url is optional, so the smallest valid body is a single URL and the reply is a task to poll or receive on a callback. Send an Idempotency-Key, keep the task id, and the video and its audio arrive together when the task reaches a terminal state.

Building a Seedance 2.0 Spicy API request

  1. Start from a still you are licensed to use

    image_url is the one property a request must carry, and it has to resolve without a login for the picture to reach the model. A URL that cannot be fetched comes back as an invalid request, so the link is worth testing before you spend a generation on it.

  2. Choose aspect_ratio before writing the prompt

    adaptive keeps the shape of the still, while the six fixed values reframe it. Framing changes what the same action reads as on screen, so settle the shape first and then describe the shot that belongs inside it.

  3. Set duration_seconds and resolution for the pass you are on

    duration_seconds is an integer from 4 to 15 and gives 5 when omitted; resolution gives 720p. Both set what a run costs: billing is per second, at a rate that rises with resolution. Draft low on the ladder and save 4k for the final, which renders as a new take.

  4. Decide whether generate_audio stays on

    It defaults to true, and the soundtrack is generated against the same clip in the same task. Set it to false when a composer is scoring the sequence later, or when you are judging motion alone and a rough soundtrack would pull attention away from it.

  5. Add last_image_url once the ending is decided

    A closing frame without an opening one is refused, so the two always ship together. Reach for it when the final composition matters as much as the first frame, and send the same still twice when the clip has to end where it began.

Seedance 2.0 Spicy worked examples

  • A vertical scene for a story placement

    Set aspect_ratio to 9:16 and direct a slow reveal that reads in that shape. A vertical frame holds one character close and face-forward, where a widescreen take would put half the room between the viewer and the subject.

    Vertical frame, adult character seated at a café table turns toward camera and smiles softly as evening light shifts across their face; slow push-in; ambient street sound

  • A full-length scene with room to develop

    Use the fifteen-second ceiling to let a scene develop from approach to a settled, quiet moment. Four seconds compresses the same idea into a single gesture, and the beat lands differently when there is time for what follows it.

    Two adult characters on a balcony at night move from standing apart to sitting close together, shoulders touching, both relaxed and smiling; string lights overhead; camera slowly circles

  • A silent 4k pass for a sequence scored in post

    Set resolution to 4k and generate_audio to false when the music is written later. The clip renders at the top step of the ladder with no soundtrack competing for attention while you review the motion.

    Adult character walks slowly down a lamplit hallway at night, each lamp casting the same warm glow, camera tracking alongside at a steady pace, no dialogue

Seedance 2.0 Spicy API questions

Writing the prompt when the frame carries the look

Write the prompt as direction for what moves: the action, the camera, the pace of the beat. image_url has already settled wardrobe, styling and light, so words spent on those compete with the picture instead of adding to it. Keep aspect_ratio in mind while writing, because a walk toward camera down a long hall reads differently at 9:16 than at 21:9. The field accepts up to 5,000 characters, though a short prompt is easier to revise between takes.

All Seedance models10

Related comparisons

Seedance 2.0 Spicy vs Seedance 2.5 Spicy