Seedance 2.0 Spec Sheet: Max Length, Resolution, 4K, Aspect Ratios and Reference Limits | SeedanceTips

Seedance 2.0 generates clips of 4 to 15 seconds at 480p, 720p, 1080p or 4K, in six aspect ratios from 21:9 to 9:16, at 24 fps. Those are the limits ByteDance publishes for the dreamina-seedance-2-0-260128 model on BytePlus ModelArk (data checked 2026-10-08). 4K is exclusive to the full Seedance 2.0 model; Seedance 2.0 Fast and Seedance 2.0 Mini share the same 15-second ceiling but stop at 720p.
This page is a spec sheet, not a hands-on test. Every number below comes from ByteDance’s own documentation: the BytePlus ModelArk model list, the video generation tutorial (page stamp “Last updated: September 28, 2026”) and the Dreamina Seedance 2.0 series tutorial (page stamp “Last updated: September 29, 2026”), plus the consumer Dreamina Seedance 2.0 page where the consumer app differs. If you want prices rather than limits, the Seedance tiers guide covers the API rate card and the Seedance pricing review covers subscription credits.
Sources: BytePlus ModelArk model list and video generation tutorial, both read on 2026-10-08. The three 2.0 models share one input specification; the docs group them as “Dreamina Seedance 2.0 series”, and the 2.0 series tutorial describes the difference between them as a trade-off between generation quality and cost.
Two things in this table catch people out. First, Fast is not a cheaper route to 1080p: the tutorial’s pixel table carries the note “Dreamina Seedance 2.0 Fast and Dreamina Seedance 2.0 Mini do not support 1080p”. Second, the reference limits are per request, not per project. Each generation can use the full allowance again.
The longest single Seedance 2.0 generation is 15 seconds. The video generation tutorial (checked 2026-10-08) sets the duration parameter “in whole seconds” with these ranges:
The -1 value is the one most people miss. The tutorial explains it directly: “A value of -1 enables intelligent duration selection, allowing the model to choose an appropriate video length within the supported range.” In practice that means you hand the length decision to the model and get something between 4 and 15 seconds back.
There is no fractional-second option on 2.0. The frames parameter, which converts to a duration at 24 frames per second and allows odd lengths, is documented for the Seedance 1.0 models only.
The 15-second ceiling applies to one generation, not to a finished video. The 2.0 series tutorial lists video extension as a supported task: you can extend a clip forward or backward, or “stitch multiple video clips (up to 3 clips) into a coherent video”. Because each input video in a 2.0 request must be 2–15 seconds and all input videos together must stay within 15 seconds, extension is a way to chain shots, not a way to submit a long timeline in one call.
If a single uninterrupted take longer than 15 seconds is the requirement, the official answer is a different model. Seedance 2.5 is listed at 4–30 seconds on the same model list. Our multi-shot storytelling guide covers how to plan a sequence across several 15-second clips instead.
Seedance 2.0 has four output resolutions on the API: 480p, 720p, 1080p and 4K. The model list (checked 2026-10-08) gives the color depth for each:
The pixel table in the tutorial is blunt about who gets 4K: “Only Dreamina Seedance 2.0 supports 4K”. No other Seedance model on the BytePlus model list, including Seedance 2.5, lists a 4K option on the API.
Three practical differences come with the 4K setting, all documented by BytePlus as of 2026-10-08:
Older Seedance 2.0 coverage, including some earlier pages on this site, describes the model as “native 2K”. That was the launch-era description. Seedance 2.0 gained 4K output at the Volcano Engine FORCE conference on June 23, 2026, according to third-party launch coverage from The Decoder, and the current BytePlus documentation lists 4K with no 2K setting at all. For anything you build today, treat 480p / 720p / 1080p / 4K as the complete list.
The ratio parameter accepts 16:9, 4:3, 1:1, 3:4, 9:16, 21:9 and adaptive. Adaptive keeps the shape of your input instead of forcing a fixed ratio. ByteDance’s own 4K example in the 2.0 series tutorial uses "ratio": "adaptive" with a reference image.
Resolution and ratio together decide the exact frame size. These are the output dimensions the video generation tutorial lists for the Seedance 2.0 series (checked 2026-10-08):
A few details worth knowing before you set up an edit timeline:
For image-to-video, the tutorial documents a center-crop rule. When the uploaded image does not match the requested ratio, the model crops it to that ratio, and the crop window is always centered on the original and stays fully inside it, with no black borders. The tutorial’s own advice is to set a ratio “as close as possible to the aspect ratio of the actual uploaded image”. If your subject sits near an edge, crop the source yourself first or switch to adaptive. Step-by-step setup is in our image-to-video tutorial.
Seedance 2.0’s multimodal reference mode (the docs call it “omni reference”) is where most failed requests come from, because each file type has its own count, length, size and dimension window. Here is the full set for the 2.0 series, from the video generation tutorial’s Limitations section (checked 2026-10-08):
The whole request body is capped at 64 MB, and the docs advise against Base64 encoding for large files. Images and audio can be sent as a URL, a Base64 string or an asset ID; videos as a URL or an asset ID only.
The capability table in the tutorial lists image reference, video reference, and four mixed combinations for the 2.0 series: image + audio, image + video, video + audio, and image + video + audio. Audio-only reference is marked as not supported on Seedance 2.0, Fast and Mini, with the note “must be used with an image or video”. Seedance 2.5 is the only model in the table that accepts audio as the sole reference.
One limit is not about numbers. The tutorial states: “Dreamina Seedance 2.5 and Dreamina Seedance 2.0 series m