Model · ByteDance
Seedance, the deepest video family in the catalogue
Seedance is ByteDance's video model, and it is the most complete family available here: five engines, from the cheapest to the most capable. Its edge comes down to one number — Seedance 2.5 accepts up to thirty reference images, where most models take one or two.
What Seedance is good at
Thirty reference images
This changes the nature of the work. With a single start frame you hope the model understands the character. With thirty references you impose it — product, set, character and palette, held across the whole sequence.
Up to 30 seconds
Seedance 2.5 reaches 30 seconds where most video models stop at 10 or 15. For a product sequence or a demo, that avoids stitching three shots that do not match.
Text, image, reference or clip
Depending on the engine: text alone, a start frame, reference images, reference clips, even a reference audio track. The inputs combine rather than exclude each other.
The same publisher as Seedream
Seedream makes the image, Seedance animates it. Two models from one publisher share a way of handling material and light that two different vendors do not.
The five Seedance engines available here
From most capable to cheapest, all in the same picker.
- Seedance 2.5The most complete: text, image or reference to video, up to 30 input images and 10 reference clips, 4–30s, 480p to 1080p, optional audio.
- Seedance 2 Reference to VideoReference or text to video, up to 9 images and 3 reference clips, 4–15s, optional audio.
- Seedance 1.5 ProHigh-quality image-to-video, 1–2 input images required, 4–12s, 480p to 1080p, seven ratios.
- Seedance 2.0 FastThe fast, cheaper variant, 4–15s, 480p or 720p.
- Seedance 2.0 MiniThe most compact, for iterating before a final pass.
What Beemm Vision adds
A model that takes thirty references is only useful if you can hand them over without uploading them one by one.
- Images generated at the previous step wire straight into the video node — no export, no re-upload.
- A multi-image import node feeds the references in one move.
- Compare Seedance, Kling, Wan and MiniMax on one brief before committing to a long generation.
- Derive the video into several formats without redoing the settings.
- One credit balance for image and video alike.
Frequently asked questions
- What is the difference between Seedance and Seedream?
- Same publisher, two jobs: Seedream generates images, Seedance generates video. Both families are in the catalogue, so you can generate an image and then animate it while keeping a consistent look.
- What is the maximum length?
- Seedance 2.5 goes up to 30 seconds. The other engines in the family stop at 12 or 15. The picker shows the exact range for each before you launch.
- How many reference images can I give it?
- Up to 30 for Seedance 2.5, and up to 9 for Seedance 2 Reference to Video. That is what lets you impose a character, a set and a palette rather than hope for them.
- Does Seedance generate sound?
- Yes, several engines in the family can produce a soundtrack with the video. It is an option on the generation, not a separate step.
Generate a video with Seedance
Start from text, from an image, or from a set of references you have just generated.
Keep reading
- Luma Uni-1 on Beemm VisionGenerate with Luma Uni-1 and Uni-1 Max: one unified model that handles generation and image editing the same way, inside a multi-model canvas.
- Recraft on Beemm VisionRecraft V4.1 Pro generates images and, above all, VECTOR output: logos, icons, pictograms. The only model in the catalogue that returns editable curves.
- Wan on Beemm VisionGenerate video with Alibaba's Wan 3.0, 3.0 Prime, 2.7 and 2.6 Flash: up to 30 seconds, 10 reference images, 5 reference clips, reference audio.
- The full catalogueEvery image and video model available in the canvas.