Seed Dance image-to-video is 9 credits and has no audio flag

The older Seed Dance I2V page is 1080p, 2–12s, 9 credits, audio null — do not treat it as Seedance 2.5 (47cr, native audio) and do not invent a /models/seedance-2-5-image-to-video URL.

ByteDance ships more than one Seedance-shaped row. The slug /models/seed-dance-image-to-video is the older still-to-motion page. The slug /models/seedance-2-5 is a later text-to-video family with a native mix. They do not share a bill, a duration ladder, a resolution ceiling, or an audio flag. Name the row. Do not fold a guessed I2V path onto the 2.5 page.

Two Seedance rows, two bills

The names look related. The catalogs are not interchangeable.

Seedance Image to Video — the Seed Dance I2V page — is ByteDance’s dancing-motion row. Catalog description: transform images into dynamic dancing videos. An image is required. The only published category is image-to-video. Output tops out at 1080p (480p / 720p / 1080p). Durations run 2s through 12s, one second at a time. Listed cost is 9 credits. Native audio is not marked on this row — audio is null. Aspect ratios include 21:9, 16:9, 4:3, 1:1, 3:4, 9:16, and auto. Styles: Cinematic, Documentary, Abstract. Frame rates 24fps and 30fps. Motion levels Medium and High.

Seedance 2.5 is a different published row. Catalog description: a native 30-second single-shot video at up to 720p from one text prompt. The published category is text-to-video. Durations run 4s through 30s. Listed cost is 47 credits. audio is true — native audio. Qualities 480p and 720p. That is the family hold for a long single pass, not the 9-credit dance row.

Nine credits is not 47. Twelve seconds is not thirty. 1080p on the older I2V page is not a 720p 2.5 ceiling. Null audio is not a native mix. If you treat them as one “Seedance” button, you will either wait for a soundtrack the 9cr row never promised, or you will spend the 47cr family on a two-second dance test.

The still is the contract already walked the older row as a plate that has to dance. This page is only the split: 9cr, audio null, versus 2.5 at 47cr with a mix — and the URL you must not invent.

Seed Dance image-to-video is 9 credits, audio null

The older page is still-to-motion. requires_image is true. No upload, no clip. The pose in the frame is the starting body. If the arms are wrong in the still, High motion will not correct them; it will commit to them.

Dance is the described job. That is more specific than generic image-to-video. If you needed a slow product orbit with no body, this is the wrong family even if the still is pretty. If you needed a body to move like a chorus, this is the family — and the still is the costume, the face, and the floor.

The floor is 2s. Start at 2s, 3s, or 4s. Climb toward 12s only when the phrase is actually that long. Every extra second copies the still. If the label is misspelled in the plate, you are buying up to twelve seconds of the misspelling. Fix the still first. Then buy time.

Nine credits is the listed figure on a row that also bills by resolution and duration in the matrix. Do not treat 9 as a reason to max the ladder “because it is cheap.” 2s at 1080p that holds is the win. 12s at 1080p of a still you do not like is the whole budget of a small job, spent on dancing.

audio is null. Do not write a silent prompt and expect a track. Do not write “bass drop” and wait for a chorus. Score later. Add music after the take if the job is picture-first. A “dancing video” without a listed soundtrack is still a picture job.

1080p is the ceiling. There is no 4K on this output list. Medium versus High motion is the aggression of the move, not a licence to invent a soundtrack.

Seedance 2.5 is 47cr with native audio — and no invented I2V URL

Seedance 2.5 is the long single-shot family: 4s–30s in one pass, 720p ceiling, listed 47 credits, native audio. The published model URL is Seedance 2.5. The published category on that page is text-to-video.

I2V is the still-to-motion job on that family when frame one has to be a JPEG you already approved. The door for that job is image-to-video. The parent model page stays /models/seedance-2-5. Do not invent a second Seedance path. Do not type /models/seedance-2-5-image-to-video and expect a live catalog row. That slug is not in the published brief. Linking it is a 404 dressed as a family.

Thirty seconds in one pass is why 2.5 holds a SKU that has to remain itself. Stitching four short Seed Dance clips around a bezel is how the badge drifts — and those clips still will not arrive with a native mix. 720p is the ceiling on the 2.5 endpoint. If the delivery spec is 1080p dance from a required still, you are on the older 9cr page. If the delivery spec is a 30-second identity pass with a mix, you are on 2.5.

Native audio on 2.5 means a mix can arrive. Write the mix you want, or write silence as a mix instruction, because a silent brief on a native-audio row can still get a soundtrack you did not choose. That problem does not exist on Seed Dance I2V: there is no mix flag to argue with. Score after.

Do not use 2.5 “so the dance holds longer.” Motion that already drifted at second five will not become true at second thirty. Use the short end of the row that matches the job.

Run I2V from the tool, not a guessed slug

Image-to-video is the still-to-motion door. Upload the plate. Pick the family the job needs.

• Required still, 2s–12s, 1080p, listed 9 credits, audio null, dance-shaped motion → /models/seed-dance-image-to-video
• Long single pass, 4s–30s, 720p, listed 47 credits, native audio, parent page /models/seedance-2-5 → same tool, that family, no invented I2V URL

The ByteDance roster is the rest of the map. This page is only the split between the older Seed Dance I2V page and Seedance 2.5. Do not paste the still into a text-to-video box and hope the model finds it. Seed Dance will not start from text. Seedance 2.5 can start from text — that is a different brief, and it will invent frame one.

If you would not ship the still, you are not allowed to animate it on either row. If you would ship the still, generate the short end first. Watch the first beat. Then buy length. Then, and only then, worry about a mix — and only on the row that actually lists one.

FAQ

Is Seed Dance the same as Seedance 2.5?

No. Seed Dance image-to-video is 1080p, 2–12s, listed 9 credits, audio null, image required. Seedance 2.5 is 720p, 4–30s, listed 47 credits, native audio, published as text-to-video. Related names. Different bills.

Does Seed Dance image-to-video include a soundtrack?

No. audio is null on that slug. Plan music or VO as a later pass. Do not wait for a chorus the row did not promise. Seedance 2.5 is the native-audio family.

Where is Seedance 2.5 image-to-video in the catalog?

It is not a published slug. Do not invent /models/seedance-2-5-image-to-video. Still-to-motion on that family runs through image-to-video with the parent page at /models/seedance-2-5.

Can I start Seed Dance from text only?

No. The catalog requires an image. Text-to-video is a different ByteDance row — including Seedance 2.5 — not this one. If you do not have a plate, lock a still first.