How to Use Seedance 2.0 for AI Video (Without a CapCut Subscription)

Seedance 2.0 is ByteDance's AI video model, built around one idea: it generates synchronized audio and video in a single pass. Sound effects land on cue. Dialogue matches lip movements. Background music syncs to visual beats. All of that happens automatically, with no separate audio step.

Most people know Seedance through CapCut Pro, where clips draw from a monthly pool of 1,200 AI points on a $19.99/month plan. In Slates the same model runs on credits from 3.5 a second at 480p: a 10-second draft is 35 credits and 720p is 75, with no pool to run out of.

Here's everything you need to know about what Seedance 2.0 can do, what it costs, and how it compares to the other models available right now.

What Is Seedance 2.0 and Who Made It?

Seedance 2.0 is an AI video generation model built by ByteDance (the company behind TikTok and CapCut). ByteDance launched it in February 2026, and a newer Seedance 2.5 followed in July.

Its headline feature is the unified audio-video architecture. Seedance generates sound and picture together rather than adding audio in a second step. The audio isn't just slapped on top. It's environment-aware. A whisper in a large room gets reverb. A door slam sounds different in a hallway than in a bedroom. Music syncs to visual cuts automatically.

It also takes a lot of references at once. Through its "Omni Reference" mode, you can feed it up to 9 images, 3 video clips, and 3 audio clips, with a limit of 12 reference files total, plus your text prompt.

• ›Text-to-video: Describe what you want, get video with matching audio
• ›First and last frame: Control the start and end of the clip with reference images
• ›Omni Reference: Combine multiple images, video clips, and audio files as references in one generation
• ›Lip sync: Dialogue with mouth movement matched to the audio
• ›Multi-shot generation: Multiple camera cuts within a single 15-second clip

How Much Does Seedance 2.0 Cost?

In Slates a 10-second Seedance 2.0 clip is 35 credits at 480p and 75 at 720p, $1.40 and $3.00 on the $10 pack. On CapCut Pro the same generation comes out of a 1,200-point monthly pool, and on Dreamina (ByteDance's own creative platform) Seedance 2.0 sits inside a monthly plan with credit caps, on promo pricing that changes every few weeks.

Here is what Seedance costs in Slates next to Kling. Veo isn't in Slates, so its row is Google's own API rate, for reference:

A credit is $0.04 on the $10 pack and less on bigger packs.

Audio is always included with Seedance. There's no toggle and no surcharge. A 480p Seedance clip with sound is 3.5 credits a second; Kling Standard with sound is 6.3 at 720p, against Seedance's 7.5.

Higgsfield sells Seedance 2.0 through credit-capped plans. Plus is $59/month ($47 annual) for 1,200 credits and Ultra is $129/month ($99 annual) for 3,000, and each Seedance generation eats through credits fast. In Slates, Seedance runs on Slates Credits priced by resolution, with no monthly subscription, no credit caps, and no expiring allowances.

What about CapCut Pro?

CapCut Pro runs $19.99/month ($179.99/year) and gives you access to Seedance 2.0 through 1,200 AI points a month, with extra points sold from $4.99 per 100. A clip's cost depends on its length and resolution, and the points expire on the plan's clock.

If you're generating more than a handful of clips per week, the subscription model gets expensive fast. Slates Credits, from 3.5 credits a second, have no monthly caps, expiring credits, or resolution restrictions. For a full breakdown of how subscription pricing compares to pay-per-use across all models, see AI Video Generator Pricing: Subscription vs Pay-Per-Use.

What Can Seedance 2.0 Actually Do?

Seedance generates video clips from 4 to 15 seconds long, at 480p, 720p, 1080p or native 4K. It supports six aspect ratios: 21:9, 16:9, 4:3, 1:1, 3:4, and 9:16. Every generation includes synchronized audio at no extra cost.

In Slates the price follows the resolution: 3.5 credits a second at 480p, 7.5 at 720p, 18.5 at 1080p and 39 at native 4K. Draft at 480p and render the keeper at the resolution you deliver.

The three generation modes

Text-to-video is the simplest. Describe your scene, and Seedance generates the video with matching audio. You can specify camera movements, lighting, and even multi-shot sequences in your prompt.

First and last frame gives you more control. Upload a reference image for the start frame, end frame, or both, and Seedance generates the motion between them. This is useful for product animations, scene transitions, and maintaining visual consistency across a sequence.

Omni Reference is where Seedance really separates itself. You can combine up to 9 reference images, 3 video clips (15 seconds total), and 3 audio clips (15 seconds total), with a limit of 12 reference files total, plus your text prompt. Want a specific character wearing a specific outfit in a specific location with a specific backing track? Feed all of those as references.

Native audio co-generation

This is the feature that makes Seedance genuinely different. The model generates audio and video together in one pass, not as separate steps stitched together afterward.

That means:

• ›Sound effects land precisely when the visual event happens
• ›Dialogue gets lip-synced automatically
• ›Background music matches the mood and timing of visual cuts
• ›Environmental acoustics adapt to the scene (reverb in a cathedral, dry sound in a small room)

For comparison, Kling 3.0 charges extra for audio (2.1 more credits a second at 720p in Slates). Veo 3.1 comes with audio only on Google's own API. Seedance includes it at the base rate, always.

How Does Seedance 2.0 Compare to Kling 3.0 and Veo 3.1?

Seedance gives you sound on every clip, Kling gives you the most hands-on camera control, and Veo, on Google's own API, gives you native 4K with audio in one pass. They're built for different shots. Here's a quick comparison across the things that actually matter for production work.

When to pick Seedance: You need video with audio, you want to use multiple reference files, or you're drafting at volume, where the 480p price adds up. It's also the best choice for music-driven content and product animations.

When to pick Kling: You need a camera that follows written direction closely, multi-character dialogue (Omni mode), or the AI Director for scripted multi-shot sequences. Kling gives you more hands-on control over exactly what happens in the frame.

What about Veo? Veo 3.1 does native 4K with audio on Google's own API, but Slates retired Veo on October 2, 2026. Inside Slates, Seedance 2.0 and Kling both render native 4K.

For a deeper breakdown, see Kling vs Veo vs Sora: Which AI Video Model Should You Use?.

What Are Seedance 2.0's Limitations?

The real limitations are clip length, identity drift between generations, and a warm color bias. Content filtering used to top this list, and it is the thing most write-ups still lead with, so it is worth being precise about what it does and does not block.

ByteDance restricts faces in references, and that is not the end of the story. After pushback from Hollywood studios, ByteDance stopped accepting reference images that contain real human faces, to guard against impersonation and unauthorized use of likeness. That is where the "Seedance can't do characters" reputation comes from.

In Slates you say up front whether your reference has a face, and the price quote shows the cost before you generate. Faceless clips are the default and the cheapest. An AI-generated character's face costs more credits, and a real person's face, with their consent, asks you to confirm you hold the rights to that likeness first. Public figures and celebrities fail on every model. So character work on Seedance is a setting, not a wall.

15-second cap per generation on Seedance 2.0. There's an "extend" feature on consumer platforms (Dreamina, CapCut), but it often causes character inconsistency between the original clip and the extension. For clean longer sequences, generate separate clips and edit them together. Slates also ships Seedance 2.5, which runs a single 30-second take at 480p, 720p or 1080p. It has no native 4K and costs more than 2.0 at every resolution the two share, so it is a second seat rather than an upgrade: reach for it when length is the point, and stay on 2.0 for 4K or for the same resolution at a lower price.

Identity drift across clips. Characters can shift slightly between separate generations. Faces change subtly, clothing details drift. Every AI video model has this problem right now, but keep it in mind for multi-shot projects.

Runs on Slates Credits. In Slates, Seedance 2.0 runs on Slates Credits, the same as every other model in the app.

Warm color bias. Some users report a consistent warm tint affecting skin tones and neutral scenes. Not a dealbreaker, but something to correct in post if color accuracy matters.

How Do You Use Seedance 2.0 in Slates?

Seedance 2.0 is available in Slates as of April 2026. Download the app, select Seedance 2.0 from the model dropdown, set your resolution, duration and aspect ratio, and generate. Check pricing for current credit rates.

Seedance 2.0 and Seedance 2.5 both run on Slates Credits. Audio is always included. No extra settings needed. You pick your mode (text-to-video, first/last frame, or omni reference), write your prompt, and hit generate.

The cost estimator above shows exactly what each generation costs. Select Seedance 2.0 from the dropdown and adjust the resolution and clip length to see your per-clip and monthly costs.

Seedance 2.0 output generated in Slates. Audio included automatically.

Frequently Asked Questions

Last updated: October 17, 2026