Frequently asked questions

A line was repeated, cut short or said wrong

A picture fix does not change the voice or the words (the provider does not edit audio). If the words are wrong, regenerate that scene: correct the words on the card and press “Regenerate scene”. The cost is the scene’s cost.

• Write the words in short sentences; about 13 characters are spoken per second (a 10-second scene fits ≈ 130 characters).
• Write numbers and abbreviations the way they are read: “twenty percent” rather than “20%”, “and so on” rather than “etc.”.
• If the brand name takes a suffix, add that form to the pronunciation in the brand voice too.
• In a video with subtitles, the card marks a scene whose speech differs from your script with a “script differs” badge and shows the difference; regenerate that scene.

What to do: Regenerate scene

Change the background, light, outfit or an object

“Fix picture” changes the scene’s picture and keeps the words and the voice. Fixes add up: each new fix carries the earlier ones too (at most 10 per scene). If the change should be in every scene, use “Apply to the whole video”: the same prompt is applied to each scene separately and the cost is one scene’s cost per scene.

What to do: Fix picture · Apply to the whole video

Good and bad prompt examples for a picture fix

A good prompt says in one sentence what changes and what stays the same.

Good example

• Turn the background into a rooftop terrace in evening light; keep the person, the outfit and the camera angle the same.
• Make the person’s jacket red and change nothing else.
• Add a steaming cup of coffee on the table.

Bad example

• “Make it nicer.” — it does not say what should change.
• “Change everything, make it more professional.” — the scene is reinterpreted and the person and place may drift.
• “Have the woman say ‘welcome’.” — words do not change with a picture fix; regenerate the scene.

The brand name is pronounced wrong

On the client card, in Brand voice → Pronunciation, write one name per line: spelling, a colon, pronunciation (e.g. ClientPoster: Klayint Poster). The pronunciation only goes into the prompts of scenes where someone speaks; regenerate that scene with the new pronunciation. For a different pronunciation in a single video, use the brand voice components on the video screen.

What to do: Add a pronunciation to the brand voice, then regenerate the scene

A word in the subtitles is misspelled

You can correct the subtitle text in the card’s “Fix subtitles” section; the line timings are kept and there is no extra cost. If the video is merged again (a scene is fixed or trimmed), the subtitles are taken from the audio again and the manual fix is lost — fix the subtitles last.

What to do: Fix subtitles

There is a pause at the start or end of a scene, the video is too long

Each scene is generated as a separate clip; when ClientPoster joins the clips it trims the time before and after the words in scenes with speech by itself. You can turn this off and on free of charge in the card’s “Automatic cutting” section (videos with background music are not cut). If that is not enough, press “Open in the trim tool” in the scene’s “Trim manually” section: drag the handles (the video shows the frame under the handle) and choose “Add to changes”. Prepare several scenes this way and save them together with “Save and merge the video”; the video is merged once, at no extra cost.

What to do: Trim manually

Writing the scene description in sync with the words

A scene gets shorter to fit its words: movement before or after the words (walking in, turning away) can be trimmed. Write the movement at the moment the words are spoken — like “walks up to the table and lifts the cup while speaking”.

What cannot be done on screen today

Changing the speaker’s voice, tone or pace afterwards; changing the background music; reordering scenes or adding a new scene to a delivered video. For these, open the wizard with “Generate again” on the content’s detail page and produce the video again.

1. Content type

Choose «Long video». The video's length comes from the number of scenes: each scene is 10 seconds. You set the number of scenes on the Scenes step with «Add scene» or the scene count box.

2. Characters and locations

Define once who acts and where it takes place; you do not rewrite them in every scene.

• Elif — add her photo (drag and drop, choose from your computer or from My images) and describe her look: «in her thirties, beautiful and attractive; olive skin, blue eyes; shoulder-length dark brown wavy hair, cream sweater». Describe her voice too: «a soft, warm and cheerful female voice».
• Ali — a photo is not required, a description is enough: «in his forties, a barista with short dark hair and a short beard; white shirt, black apron». Pick the age group that matches the description: «Middle-aged». Describe his voice too: «a warm, friendly and cheerful male voice».
• Café — location card: name «Café», short description «a small café with big windows, wooden tables and morning light; in front of the window, a step back from the glass, a small table with two chairs»; a photo if you like. We added one in this example: it helps the place look the same in scenes that are produced separately.
• Adding characters or locations is optional: if you add none, the scene text describes the people and the place.
• A video can have at most 5 characters and 10 locations; at most 3 people appear and 2 people speak in one scene.

3. Video prompt

The general video prompt is added to every scene's prompt: write what the whole video shares, e.g. «a realistic, cinematic picture, warm colours, soft natural light, calm pace».

• In a long video, if you described every scene you can leave the general prompt empty; each scene then goes with its own text and the post text is written from the scenes.
• In a short video the general prompt is required: it is the short video's only scene.

4. Scenes

On each scene card write what happens in that scene (at most 3,000 characters). Mention people by name and do not rewrite their look — the character card says it.

• Characters and locations named in the scene text are ticked automatically — also with suffixes such as «Ali's»; you can untick or tick them by hand. The card warns you about anyone named but left unticked.
• Pick the location with «Where does this scene take place»: its description and photo go into every scene it is picked for, unchanged; the scene text may add details.
• Write the dialogue in the scene's dialogue lines: who says it and what. The line is spoken in that character's voice, in sync with their lips.
• In a scene with dialogue and no background music, the pause before and after the words is trimmed automatically (automatic cutting; you can turn it off in the wizard’s Settings step or once the video is ready). Write the action together with the words («speaks while setting down the cup»); an action that ends before the words may not stay in the video. Once the video is ready you can change the trim with «Trim manually».
• Scenes are produced separately; the AI may read where the table is or what is on it differently in each scene. Write what must stay the same in the same words in every scene («the small table in front of the window») and add the place's layout to the place card's description. Write a character who has no lines as «silently», and the feeling of the speech in the scene text too («speaks with a warm smile and a friendly voice»).
• The AI can struggle with moves that touch a door, a table or a chair: a person may seem to pass through an object. Keep the move simple, say where the camera is («The camera stays inside the café, opposite the entrance door») and split the move over scenes if needed — in this example Elif sits down in scene 2.
• Once the video is ready you can regenerate a scene that didn't work on its own: in the «Correct the video» card use «Regenerate the scene» — the scene's text, lines and location come pre-filled; change them and regenerate, and you pay for that scene only. For a small change to the picture, use «Correct the picture».

Card or scene text?

If they conflict, the person's identity — face, hair, age and voice — comes from the character card; what they do in that scene, where they are, the camera and their clothing come from the scene text. If the card says «cream sweater» and you write «Elif has taken off her sweater, in a white T-shirt» in scene 3, she appears in the T-shirt in that scene but her face does not change.

Limits

Text boxes do not cut text over the limit: the counter turns red and you are asked to shorten it before you continue.

• A scene's text is at most 3,000 characters.
• A scene's prompt — cards, scene text, dialogue and general prompt together — is at most 16,000 characters; if it does not fit, the production notes, the target audience, the oldest change notes and the general prompt are shortened first; your scene text and dialogue are not shortened.