Comfy has integrated Seedance 2 in the v0.19 release, expanding its video generation capabilities for users and developers.

Available templates and input modes

The update introduces three generation templates: Text-2-video, Reference-2-video and FLF, each addressing different input scenarios for creators.

• Text-2-video produces video content from textual prompts without additional media inputs.
• Reference-2-video (r2v) accepts reference images; current implementations use four images in practice but support could extend to a larger set.
• FLF generates motion between two provided frames, focusing on interpolation and temporal coherence.

Reference handling and differences

In this build r2v appears to rely primarily on image references, whereas Dreamina's omnireference system supports mixed media inputs up to twelve items including video and audio.

Face and content restrictions

For the global release the model underwent content checks and by default resists generating realistic human faces, reflecting platform-level safety filtering and moderation choices.

Generation modes and speed

Seedance 2 ships in two modes: a Fast variant for quick drafts with generate times around thirty to sixty seconds, and a Base variant intended for higher-quality renders of longer duration.

Pricing

Pricing is presented in credit units and differs between Fast and Base modes, with costs scaling by duration and resolution.

• Seedance 2 Fast: 7 sec 480p — 119 credits; 7 sec 720p — 255 credits.
• Seedance 2 Fast: 15 sec 480p — 256 credits; 15 sec 720p — 548 credits.
• Seedance 2 (Base): 7 sec 480p — 149 credits; 7 sec 720p — 320 credits.
• Seedance 2 (Base): 15 sec 480p — 319 credits; 15 sec 720p — 685 credits.

Platform availability and free tiers

The model is accessible through multiple hosts including Fal, Replicate, Freepik and Runway, with notable price differences across providers for comparable outputs.

For comparison, a four sec 720p render on Freepik costs 1128 credits, highlighting variability in commercial pricing between services.

Bytedance offerings and free quotas

Bytedance, the developer of the generator, exposes the model within CapCut and Dreamina ecosystems with limited free credits and varied generation limits for users.

In the mobile CapCut application a Fast four sec 480p clip can be generated at no charge, while Dreamina grants 120 daily free credits on the website for basic experiments.

The CapCut web interface provides 450 free credits and can produce Base model videos; a five sec 720p clip costs about 85 credits, and users may encounter resource limits prompting subscription offers.

Example prompts and output

The same textual prompt was used to produce both Fast and Base variants in demonstration clips, showing differences in motion fidelity and resolution between modes.