Genjutsu AI Generator

Genjutsu AI brings two different starting points to video creation: describing a new shot, or supplying a reference clip and a character photo. This guide covers the text-to-video capability powered by Seedance, rather than promising that a reference edit works like a written prompt.

From one sentence to a new video

A useful prompt names a subject, an action, a setting and a camera move. For example: “A cyclist passes a rain-soaked café at dusk, slow tracking shot, warm window light.” Start with a single coherent shot; adding several locations or conflicting actions makes the result harder to control.

The AI genjutsu generator interprets those words to create new footage. It does not need to preserve an existing actor or background. Output is a generated interpretation, not a guaranteed reproduction of a real scene, and rendering time varies with model demand.

Seedance 2.0 or Seedance 2.0 Fast?

Seedance 2.0 is configured for 480p, 720p and 1080p. Seedance 2.0 Fast is configured for 480p and 720p, making it a practical option for exploring a shot before committing to a higher-resolution version. Fast does not offer 1080p in this tool's current model catalog.

Both text-to-video models support 9:16 for vertical publishing, 16:9 for landscape edits and 1:1 for square placements. Configured output durations are 5 and 10 seconds. These settings belong to the Seedance text-generation API, not the reference-video editor.

Which workflow is currently on the creation page?

The creation page currently offers Reference Video Edit only: choose an effect or upload a source clip of at least 3 seconds, add a character photo, pick 16:9 or 9:16 and generate a new interpretation of up to 10 seconds with Kling video-edit models. Text to Video is not available on the creation page at the moment, so the Seedance settings above describe the configured model catalog rather than controls you can use today.

If your goal is to keep filmed movement while changing a character, read the Higgsfield Genjutsu alternative comparison, or open the creation page and try an effect with your own photo.

Choose the format before rendering

Use 9:16 when the composition needs to work on a phone. Keep the subject away from edges reserved for captions. Choose 16:9 for a wider environment or a clip you will cut into a landscape timeline. Choose 1:1 when square composition is part of the brief, rather than relying on a later crop.

A 5-second idea is easier to evaluate than a long sequence. Move to 10 seconds only when the action needs that time. Higher resolutions and longer clips use more credits; not every model and resolution combination is available.

Credits and failed generations

Creating an account and exploring examples are free. Rendering is paid: a character swap of up to 10 seconds costs $9.90 of balance, and the cost is always shown before you generate. A failed generation automatically returns the deducted credits; a successfully rendered clip is not automatically refunded merely because you dislike its creative interpretation.

Read what is free and what costs credits before buying a pack. For briefs, format choices and ten effects from the library, see Genjutsu Video AI for shorts and ads.