Purpose
Genjutsu AI is a standalone AI video tool for planning new shots and performing reference-based character editing. It is aimed at creators who want to quickly produce cinematic videos, vertical short videos, and product ads.
Main capabilities
- Text-to-video: Describe the subject, scene, lighting, and camera movement in a way similar to briefing a cinematographer. The model can understand camera language such as push-in and pull-out, orbit, and handheld follow.
- First-and-last-frame video: Generate video from the first frame and last frame.
- Character replacement: Upload a photo to replace the person in the photo into an existing video scene, maintaining the same motion, background, and camera movement. For example, the Genjutsu STORM effect can place a person into a cinematic crowd shot; you can also upload a dance clip and have the performer take on the appearance of the person in the photo.
- Motion transfer: Reference the original performance to generate a new character.
- Aspect ratio and duration: Each video starts from 9:16 by default, suitable for TikTok, Reels, and Shorts; you can switch to 16:9 (YouTube) or 1:1 (feed posts) with one click.
- Model selection: Choose models by shot, and view credit consumption before rendering.
- History: Recent creations are kept in history, allowing you to compare different versions, download satisfactory results, and continue iterating.
- AI voiceover: Provides narration for short videos, marked as "coming soon."
How to use
- Describe: Write down what you want to see, or start from an idea card, specifying the subject, scene, lighting, and camera movement.
- Choose a model: Select the video model, aspect ratio, and duration, and view the credit cost before rendering.
- Download and publish: Track rendering progress, download the MP4, and publish it to any platform.
Use cases
- Vertical short videos such as TikTok, Reels, and YouTube Shorts.
- Product ads and UGC-style videos.
- Place a single photo into viral video scenes.
- Character replacement and motion transfer based on reference videos.
Limitations and billing
- Registration, browsing templates, and opening examples are free; video generation uses prepaid balance.
- Packages start at $4.90 for the first order, with no subscription and no auto-renewal. Text-to-video, first-and-last-frame video, and character replacement share one balance.
- Each generation is billed by API, duration, and resolution, and the cost can be viewed before generation.
- Technical failures automatically restore deducted generation credits; completed generation results consume credits, even if they do not match creative expectations. Failed generations are automatically refunded.
- Photos are used only to create videos.
- Genjutsu AI is not affiliated with Higgsfield. Higgsfield Genjutsu edits existing footage through motion transfer or targeted object replacement; Genjutsu AI is a standalone tool, and it is recommended to compare workflows before choosing.
- Seedance creates new footage from text descriptions; reference editing, on the other hand, starts from real-shot clips and offers different kinds of control. To distinguish the Seedance text-to-video option from the current creation editor, you can read the generator guide.
- Regarding details such as available AI models, the specific dimensions and durations supported, and commercial use licenses, no clear list or terms explanation is currently provided in the materials.
