Multi-shot story progression
Built for short-film storytelling and shot progression. Spelling out shot-switching order and subject relationships in the prompt gets you closer to the narrative structure on the first try.
Alibaba Tongyi · Wan 2.6
Wan 2.6 is the main model in Alibaba Tongyi's Wanxiang series built for multi-shot film narrative, emphasizing shot progression, native audio-video, and character-reference footage. Common official specs include up to ~15s per clip, up to 1080p at ~24fps, and support for text / image / reference-video / audio multimodal input. Compared with Wan 2.5, 2.6 is more complete on narrative, A/V sync, and reference control; compare against Wan 2.7 when you need the family's newest capability. Select Wan 2.6 in the iMini video workspace to use it.
Built for short-film storytelling and shot progression. Spelling out shot-switching order and subject relationships in the prompt gets you closer to the narrative structure on the first try.
Officially described as supporting native A/V generation. For tests that need dialogue, ambient sound, and picture in sync, 2.6 has a clear place in the Wanxiang family.
Beyond text and image, you can also bring in reference video and audio to constrain motion or mood. Character-reference footage and recast test takes go more smoothly.
Compare Wan 2.7 for the newest Wanxiang narrative capability; compare Seedance 2.5 when you need a native ~30s single pass with a larger reference scale.
Narrative completeness, native audio-video, and reference control—the key changes to check before migrating from 2.5 to 2.6.
2.6 makes shot continuity and story progression a headline feature. Compared with 2.5, cutting between wide, medium, and close shots within one pass is more usable.
Supports native A/V generation with lip-sync description. When ambient sound, dialogue, and picture need to come from the same pipeline, 2.6 fits better than 2.5.
Beyond text / image, you can also bring in reference video and audio. Character reference, motion constraints, and mood locking no longer rely on text description alone.
Common official specs—up to ~15s, 1080p, and ~24fps—are more verifiable. Test a full clip at the workspace's actually available tier before finalizing.
All are mainstream contemporary video choices. The differences are mainly Wanxiang generation, native A/V, and clip length plus reference scale.
| Dimension | Wan 2.6 | Wan 2.7 | Seedance 2.5 |
|---|---|---|---|
| Clip length | Up to ~15s | Follows workspace tier | Native ~30s single pass |
| Resolution | Up to 1080p · ~24fps | Follows workspace tier | HD tiers available (follows workspace) |
| Native audio | Native A/V and lip sync (per official description) | Follows current workspace capability | Follows current workspace capability |
| Multimodal references | Text / image / reference video / audio | Text / image / first-last frame and multi-image (follows workspace) | Full-modal multi-reference (~50 per official spec) |
| Motion & physics | Multi-shot narrative with character-reference footage | Multi-shot narrative and shot continuity first | Longer-segment consistency first |
| Prefer when | Multi-shot narrative with native A/V and reference video | Latest Wanxiang multi-shot film narrative | Generating a full ad segment / short-drama beat in one pass |
When you need native A/V and reference-video control, and a ~15s clip already suffices for multi-shot narrative.
For new projects in the Wan family that need the newest narrative and shot-continuity capability.
When you need a native ~30s single pass with larger reference scale to tell a beat in one go.
Three steps from opening the workspace to exporting a clip.
Go to iMini video creation and select Wan 2.6.
Spell out shots and audio-visual intent; upload reference video and images when you need to lock a character or action.
Export once narrative, A/V, and subject consistency meet the bar. If not, add references or revise the shot descriptions first.
Quotas, specs, migration, and how to choose.
Multi-shot narrative, native audio-video, and reference-video control. Open the video workspace to use it—no separate vendor API key needed.
No separate vendor API key needed.