The clip is the prompt
Reference footage carries the direction — no prompt engineering needed to describe a camera move.
AI Video Generator
Reference in, edit out — a video generator directed by footage, not by prompt engineering.
AI 模型 4 點數
GPT Image 2萬用圖像生成，最高 4K 指示0/2000
請避免使用受版權保護的角色、標誌、音樂及敏感或違規內容，否則可能被攔截。
解析度
長寬比
批量生成1
預估費用：4 點數
你的剪輯會顯示在這裡
選擇參考、加入主體，然後按「生成」。
Why AI Video Generator
Why creators choose AI Video Generator
AI Video Generator runs inside the Katana AI studio, so you can generate, edit, and export images without juggling separate tools or subscriptions.
Your subjects stay yours
Subject photos anchor identity — faces, products and outfits survive the rebuild.
Motion that already works
Presets and your own clips supply choreography that is proven to land on feeds.
Share-ready MP4
Vertical, landscape or square — download and post with your own soundtrack.
The studio
An AI video generator directed by footage
Most AI video generators start from a text box. This one starts from a clip: upload reference footage and subject photos, and the output is a rebuilt edit rather than a guess at your description.
Footage in, edit out
The reference clip decides how the video moves — camera, cuts, pacing — so nothing about the motion is left to interpretation. Subject photos decide who or what appears. An optional instruction can steer emphasis, but the direction is already done.
One pipeline, no model-hunting
The studio runs a single tuned pipeline — Katana Edit — instead of a model picker. Resolution (480p to 1080p) and aspect ratio (vertical to widescreen) are the only levers, because duration follows the reference automatically.
A companion image generator shares the same credit balance for thumbnails, covers and subject stills.
Credits and cost
Renders bill per second of output at the chosen resolution — the estimate shows before you generate, and failed runs refund automatically. One-time packs never renew; free starter credits cover a first draft.
How to use AI Video Generator
How to use AI Video Generator in three steps
Upload the reference clip
4–15 seconds of footage — yours or a preset — sets the motion and rhythm.
Add your subject photos
1–8 images of the person, product or outfit that should appear.
Generate and review
Watch the rebuild, check the subjects, download the MP4.
Who uses AI Video Generator
Who uses AI Video Generator for image work
Creators & editors
Rebuild trending formats with your own cast — no reshoot.
Brands & sellers
Product photos step into proven ad-style motion.
Social teams
Format-matched clips sized and paced for feeds.
Fans & communities
Put yourself inside the edits you already love — with rights confirmed.
More models
AI Video Generator and the rest of the studio in one place
Katana Edit handles the video pipeline end to end; companion image models cover thumbnails, covers and stills.
Reference-to-videoMotion transferSubject photosOptional audio4–15s editsImage generationMP4 exportPreset library
FAQ
AI Video Generator FAQ
01What can I generate?
Reference-to-video edits: a 4–15 second clip rebuilt around your subjects, in vertical, landscape or square formats up to 1080p.
02Is the generator free?
New accounts get free starter credits. After that, renders bill per second — you see the exact estimate before every run.
03What format do videos download as?
Standard MP4, ready for editors and social platforms.
04Which engine runs it?
Katana Edit — the single tuned pipeline this studio is built around, so there is no model-hunting.