Provider
ByteDance AI Models on Martini
Martini hosts ByteDance's Seedance 2 and Seedance 1 video models alongside Seedream image generation, so motion-heavy clips, image-to-video at scale, and product shots run on one canvas. Seedance 2 Omni adds reference images and video-to-video on top, making this the default stack for product video at volume.
About ByteDance
ByteDance — the company behind TikTok and CapCut — has quietly built one of the strongest generative video stacks in the market. Seedance 2 is the current flagship: a fast, motion-rich text-to-video and image-to-video model with tiered Fast and Standard generations, ultra-wide 21:9 cinematic support on the Pro tier, and 1080p output. Seedance 2 Omni adds reference images and video-to-video transformation on top of the base model. Seedream sits in the image lane as ByteDance's photoreal text-to-image model, frequently used for product, fashion, and lifestyle visuals where motion will follow.
On Martini, the ByteDance stack is positioned as the default choice for image-to-video at scale. The model is fast, the cost-per-clip is competitive, and Seedance 2 Omni's reference-image support means a single product photo can drive a consistent campaign across dozens of clips without re-prompting from scratch each time. Teams that ship daily ad creative or social video at agency volume tend to anchor on Seedance for the throughput while keeping Sora 2 or Veo for the prestige hero shot.
Pricing is metered per generation and shares the Martini wallet. Seedance 2 Offpeak gives a free-tier-eligible price point for batch work. Workspace billing handles attribution, and because Seedance and Seedream interoperate with every other provider on the canvas, you can route the workflow through a Nano Banana 2 character pass or a Runway Aleph V2V step without leaving the project. ByteDance models are particularly popular among teams that left CapCut or other native tools but want the same motion fidelity in a multi-provider environment.
Available ByteDance models on Martini
Video models
Best use cases
- Image-to-video at scale for daily ad creative or social video
- Motion-rich shots where Seedance 2 outperforms slower competitors on render time
- Product video built from a single reference photo via Seedance 2 Omni
- 21:9 cinematic clips on Seedance 2 Pro for hero placements
- Cost-sensitive batch generation using Seedance 2 Offpeak
- Photoreal product and lifestyle stills in Seedream that feed downstream Seedance motion
Recommended workflows
AI Image to Video
Seedance 2 (and Omni for references) is the fastest, most cost-effective image-to-video model in the lineup.
Multi-Shot AI Video
Seedance 2 keeps motion consistent shot-to-shot when paired with reference images for character lock.
AI Product Video Generator
Seedream stills plus Seedance 2 motion is the cleanest path from product photo to scrollable ad.
AI Ad Creative Generator
Seedance 2's throughput and Omni references make daily ad iteration practical at agency scale.
Related how-to guides
Related features
Related reading
Other providers
Frequently asked questions
How is Seedance 2 different from Seedance 1?
Seedance 2 is faster, supports 21:9 cinematic output on the Pro tier, generates 1080p natively, and exposes a Fast/Standard tier system. Seedance 1 stays available for cost-sensitive batches and legacy workflows.
What does Seedance 2 Omni add?
Omni adds video-to-video transformation and reference image support on top of the base Seedance 2 model — useful for keeping a product or character consistent across many clips.
Is Seedance 2 cheaper than Sora 2 or Veo?
Yes, in most cases Seedance 2 generates faster and at a lower cost per second, which is why teams default to it for batch ad creative. Sora 2 and Veo are usually reserved for the hero shot.
What is Seedance 2 Offpeak?
A discounted Seedance 2 tier with the same Fast/Standard workflow that is eligible for the Martini free tier — the best fit for batch generation when latency is not critical.
Can Seedream feed into Seedance 2?
Yes — drop a Seedream image node and connect it to Seedance 2 to run image-to-video on the still you just generated, all in one canvas.
Is Seedance the same model as ByteDance Doubao or CapCut?
Seedance is ByteDance's official model family. The exact build exposed via Martini matches the version available through ByteDance's API, which is what powers the public ByteDance generative products.
Build with ByteDance on Martini
Open the canvas and wire ByteDance into the rest of your stack in minutes. Free to start — no card required.