What this tool does
This is a prompt generator for the moment you open Google Flow (Veo), Sora or Kling and stare at an empty box. Pick what you want to film, fill in who, what and where, and you get an English prompt that already covers framing, camera movement, light, mood and sound. Nothing is sent to an AI: the page assembles the prompt from its own phrase templates, so the same choices always give the same text.
Each video maker asks for a slightly different shape. Veo wants spoken words in quotes, Sora wants dialogue in its own block below the scene, Kling asks you to keep negative instructions to a minimum and Runway asks you to describe what you want rather than what you don't. When you pick where the prompt will go, the wording and the clip lengths on offer follow that maker's own guide.
You get more than one paragraph. Alongside the prompt there is a timed shot list whose beats add up to the length you chose, a list of things to avoid for makers with a separate negative prompt field, a first-frame image prompt for image-to-video, and a short motion line to send with that image.
How to use it
- Pick a starting point such as Pet or Food. Sensible camera, light and ratio choices are filled in, along with an example you can overwrite.
- Rewrite who or what, the action and the place in your own words.
- Choose where you will paste the prompt. Only the clip lengths that maker offers stay on screen. Then choose the aspect ratio and the sound: dialogue, ambient sound, background music or none.
- With dialogue, type who speaks and the line. If the line is long for the chosen length, the count turns orange and a tip shows the comfortable maximum.
- Read any notes under "Before you generate", then press Copy all, or Copy next to a single block, and paste it into your video maker.
Keeping the same look across scenes
AI video drifts: faces, clothes and rooms change a little from clip to clip. Under "Several scenes, same look" choose two to four scenes and describe the look and the place once. That description is placed at the top of every scene prompt word for word, and only the action and the line change from scene to scene. Sora's own guide gives the same advice: reuse the same phrasing for continuity.
If you make a reference image first and build the scenes from it, paste the same description into your image tool too. The less the wording changes, the less the result drifts.
Dialogue that gets a voice
Short lines sync better. The tool counts about two English words per second of video, so an 8-second clip fits about 16 words, and a scene with a reply gets a little more room in total. Korean lines are counted at about three characters per second.
Pushy lines and shoving gestures sometimes come back without sound. Dropping the exclamation mark and choosing a gentler line usually helps, which is why the default voice is warm and gentle.
What video makers refuse or garble
An age next to the word girl or boy, real people and brand names, weapons and blood, and well-known cartoon characters are often blocked or come out wrong. On-screen text and subtitles often come out as garbled letters. The tool spots common examples (not every word) and suggests what to write instead, but it never stops you.
Your data
Everything is assembled inside this browser. Nothing you type is sent to a server or to an AI. What you last typed, your 10 most recent prompts and your favorites stay only in this browser: Clear recent empties the recent list, the star removes a favorite, and Start over clears the form. A link made with Copy link contains what you typed after the # sign. Browsers don't send that part to this site's server, but the chat or mail app you send it through, and anyone who gets it, can read it.
Questions and answers
Does an AI write the prompt?
No. The page joins its own phrase templates according to your choices. It works offline once loaded, and the same choices always produce the same prompt. Press Other combination to try a different camera, light and mood from the starting point's sensible set.
Does this page make the video?
No. It writes the text you paste. The video is made in Google Flow, Sora, Kling, Runway or another maker, with that service's own plan or credits. This site is not affiliated with any of them.
Where do the things to avoid go?
If your maker has a separate negative prompt field, paste the list there. For makers without one, the list is already at the end of the prompt. Kling asks for as few negative instructions as possible, so only subtitles (and background music, unless you chose music) plus anything under Also avoid are ruled out there. Runway asks you to describe only what you want, so nothing is added for Runway.
Why can't I pick 12 seconds for Veo?
Lengths follow each maker's documentation as of November 2026: Veo 3.1 makes 4, 6 or 8 seconds, Omni Flash in Flow up to 10, Sora 2 4, 8, 12, 16 or 20, Kling 4.0 3 to 30, Runway Gen-4.5 2 to 10 and Hailuo 2.3 6 or 10. For Kling and Runway the picker shows the common lengths (Kling 5–30, Runway 4–10). Makers change often, so check the length picker on their screen too.
What happens if my ratio isn't offered?
As of November 2026, Veo, Omni Flash, Sora and Runway make only 16:9 and 9:16, and Kling adds 1:1. For other ratios, make the vertical version and crop it afterwards. The prompt then asks to keep the main subject centered, so it is less likely to be cut off.
Is the share link private?
Only as private as where you send it. What you typed sits after the # in the address; browsers don't send that part to this site's server, but the app you send it through and anyone who has the link can read it.
