Generating a Video

1. Open the editor - Navigate to your document in the dashboard and open it in the editor. Make sure the document has at least one screenshot block.
2. Click Video - In the editor toolbar, click the Video button. The Generate video view opens: the preview window and the cost bar on the left, the settings on the right. Above the settings sits the Auto / Custom mode switch (Custom adds a director's instruction field), below it three tabs - Narration, Look and effects and Direction. In Custom mode you do not have to generate right away: you can first plan the video (2 video credits), review the plan in the player and render it afterwards - see step 7.
3. Select language - In the Narration tab, choose the narration language from the dropdown. 11 languages are available (English, Polish, German, French, Spanish, and more).
4. Choose a narrator and a voice - Use the switch above the voices to pick: No narrator (6 video credits per video), Standard (6 voices, 10 video credits) or Premium (7 voices, 20 video credits) - see the "No narrator, Standard or Premium" section below. You can preview every voice directly in the view.
5. Adjust video options - In the Narration tab, turn subtitles on or off and pick their style. In the Look and effects tab, choose the visual template (Porcelain, Studio, Paper, Blueprint, Pop or Minimal), the animation style and decide whether the video should use animated annotations, auto-zoom, the synthetic cursor, key phrases on screen, sound effects, step layouts and animated text. In the Direction tab, set how the AI should plan the video: the close-up, camera and key-phrase sliders, the video length and the narration tone. See the Direction and video options section below for details.
6. Set branding preferences - At the bottom of the Look and effects tab, choose your video branding mode based on your plan:
   • clarife - default watermark (all plans)
   • none - no watermark (Pro and Business)
   • custom - your logo, color, and slogan (Business only)
7. Start generation - The bar under the preview window shows the number of steps, the estimated time, the cost per video and your balance after generation; for a long document it warns that the narration will be condensed so the video fits in 5 minutes. Click Generate video. In Custom mode there is a Plan (2 video credits) button next to it: the AI writes the script and the finished plan plays in the preview window like a video - with subtitles, key phrases, animations and effects, but without the narrator's voice (scene timings are estimates). A scene strip sits under the preview; clicking a scene jumps to it and opens the Scene panel on the right, where you refine the plan at no extra cost: the narration (with a "Restore AI version" button), the key phrases (select words in the narration and click "Add from selection"; click a phrase to give it a different on-screen text than the spoken words, e.g. "easy to edit" → "Easy editing" - the narration still decides when it appears), the screenshot layout (whole / close-up / frame), the camera and zoom target, the order in which annotations appear and the entrance animation; you can move a scene with the arrows, delete it or replace the screenshot with a text board. The preview reacts right away, and with a scene open playback stops at its end (Play replays it); changes save automatically, and "Restore the whole AI plan" undoes them all. Options greyed out in the panel need the matching effect turned on in Look and effects. When you are happy with the plan, click Render (10, 20 or 6 video credits depending on the narrator) - the narrator and the video are produced without planning again. Look options (template, subtitles, animation style, animated text, cursor, sounds, branding, narrator) can still be changed before rendering and show in the preview immediately; after changing the language, the planned effects (key phrases, layouts, animated annotations, auto-zoom) or Direction, this plan no longer matches your settings: Plan again (another 2 credits) shows a new plan, while Render then makes the video from scratch with the new settings, without a plan preview - at the cost of a regular render. The same applies when the plan has gone out of date because screenshots, annotations, code or tables changed in the document. Plan credits are not deducted from the render. Progress is shown at the top of the view through each phase; you can go back to the editor meanwhile.
8. Download the result - When complete, the video plays in the preview window. Download the MP4 and the SRT subtitle file from the Previous generations list under the preview or from the Generations panel in the dashboard.

ℹ️ Info: The list of Premium voices depends on the language - Adam and James are not offered for English, so for English you pick from 5 voices. The bar under the preview tells you how many credits the chosen video costs and blocks the start when your balance is too low (Premium needs 20 video credits, Standard 10, No narrator 6). No narrator does not force subtitles on - if you turn them off the video is silent, and the view reminds you of that. Credits are charged when the generation starts and return to your balance when it fails.

The Direction tab decides how the AI plans the video - it works in both Auto and Custom mode. Three sliders (less / AI decides / more) control the number of close-ups and window frames, the camera movement and the number of key phrases; Video length (concise / standard / detailed) sets how many sentences a scene gets and whether chapter cards and a closing recap appear; Narration tone (neutral / friendly / formal) changes the style of speech. Custom mode adds a director's instruction - up to 500 characters, with ready-made hints (e.g. "more close-ups on forms", "don't comment on interface colors"); the AI treats it as style guidance. The chosen mode is remembered in your browser. Effects from the Look and effects tab:

ℹ️ Info: All options in the table are off by default. Auto-zoom and the synthetic cursor need animated annotations turned on - until then they are greyed out and the note under them says what is missing. Animated annotations, auto-zoom, the cursor and the close-ups of step layouts work with annotations added in the editor - a document without annotations renders without these effects. Key phrases, sound effects and animated text do not need annotations. The Direction sliders do not turn effects on - they limit or boost their use once enabled (e.g. "less" close-ups = at most one per video, "more" = up to every other screenshot scene). The visual template does not affect the plan - in Custom mode you can switch it without a new plan and see it in the preview right away. Scene settings that concern the screenshot (layout, camera, zoom target, annotation order, animation) are stored with the document: they apply to this plan and come back after "Plan again"; the narration, key phrases and scene order belong to the plan. If you edit the document text after planning, the preview reminds you that you can fix the narration in the scenes - such a plan can still be rendered.

You can save a set of settings (language, narrator and voice, subtitles, animation style, effects, Direction, branding) as a preset and load it with one click in any document. Presets live at the top of the Generate video view: a dropdown plus a menu with Save as, Overwrite, Rename and Delete. You can keep up to 10 presets; they are personal and not shared within a workspace. When you change settings after loading a preset, an orange dot appears next to its name - Overwrite saves the new settings, and choosing No preset restores the starting state (every option off, Direction on "AI decides", the document's language). The view remembers the preset you used last and loads it the next time you open it.

You can also copy the settings of a finished video without saving a preset. The same dropdown ends with a group From a previous video of this document - the latest videos with their date and a short summary of settings; picking one loads its settings and the top of the view shows From video with the date. The sliders icon next to a video in the Previous generations list and the Use these settings button in the video details on the dashboard do the same. Copied settings are not stored - if you want to come back to them, click Save as to turn them into a preset. If a preset or a video contains the Premium narrator while your balance is too low, or a branding your plan does not offer, the settings are adjusted and the view tells you so.

During generation, the progress bar moves through distinct phases:

ℹ️ Info: The progress bar shows real-time rendering progress. Phase-specific animations keep the bar moving smoothly even during longer renders.

⚠️ Warning: You can only run one video generation at a time. Wait for the current generation to complete before starting another. A video step is a screenshot, an image, or a text section separated by dividers; the intro and outro count too. Split guides that exceed 30 steps or contain a lot of text into parts - the bar under the preview window tells you when a document is too long. For a long document the AI condenses the narration so the video fits in 5 minutes - the bar warns you before you start; you can then choose the concise length in Direction or split the document.

All your generated videos are saved in the generation history. Find it in the Generations panel in the dashboard (all documents) and in the Previous generations list of the Generate video view (this document's videos only - the preview icon next to a video plays it in the preview window). Each entry shows:

• Generation date and time, number of steps, duration and file size
• Colored settings labels: narrator and voice (including No narrator), subtitles (static or karaoke), branding other than clarife, a template other than Porcelain, Direction settings other than the defaults, and the effects that were on (up to 5, the rest under "+N")
• Download links for MP4, SRT and audio
• Generation status (completed, failed, or in progress)

Above the list you will find filters: Subtitles, Narrator (No narrator, Standard, Premium or a specific voice), Language, Branding, Template and Effects, plus a Search videos box that understands setting names (e.g. "karaoke", "premium", a voice name, a template name or the name of your custom branding). The filters work in both places - the dashboard and the Generate video view. Plans from Custom mode appear in both lists as Plan tiles with Open (preview in the Generate video view) and Render buttons; the Status filter has a Plan entry. On the dashboard, the preview icon opens the video details: the player, the scenario and a Generation settings section with the full list of options used and a Use these settings button that opens the Generate video view with the same settings.

ℹ️ Info: Labels, settings filters and the Generation settings section apply to videos generated from August 2026 on. Older videos keep all their downloads but have no stored settings - when you filter by options they do not appear in the list.

Yes. Generation continues in the cloud. When you reopen the document, the result will be available if it completed.

Failed generations do not use video credits - the 6, 10 or 20 credits charged return to your balance automatically. This also applies to the Premium narrator: if its narration cannot be recorded, the generation ends with an error and a refund (clarife does not switch to the Standard narrator instead). If the document is too long for a 5-minute video even after the narration is condensed, the generation ends with a message and a refund - choose the concise length in Direction or split the document. A failed plan in Custom mode refunds its 2 video credits; a failed render from a plan refunds the render credits and the plan stays - you can render again. Check the error message - common causes include network issues or extremely long documents. You can retry right away.

No. Once started, the generation runs to completion. If it fails, your video credits are refunded automatically. A plan needs no cancelling: it stays in the list until you render or delete it (Close plan in the view hides it without deleting).

The document is auto-saved. The generation uses the latest saved version of your document content and screenshots.