Prompt bar basics

Control what gets generated from the floating bar. Job to be done: pick image or video, attach the right references, mention Elements or [@me] when needed, and submit a clear request.

1. Choose Image or Video mode.

2. Write a prompt that names subject, action, and (for video) duration or pacing if it matters.

3. Attach reference images for identity, product, or style lock.

4. Type @ to mention saved Elements or [@me].

5. Submit — the Director may reply in chat, open an approval card, or both.

Attach photos or prior frames so the model locks identity, product look, or style. Describe how each reference should be used (face lock, packshot, style frame). Uploads must finish before a submit that depends on them — the bar stays usable while uploads continue in the background. Free EdgeCut AI does not support references; use Veo, Kling, or Nano Banana when you need reference lock.

Type @ to mention saved Elements or [@me]. Keep those tags in the prompt when you want that identity in the render. The Creative Director should not cast [@me] unless you asked for it. Remove the tag if you want a fictional character instead.

• Submitting while a critical reference is still uploading.

• Expecting references on free EdgeCut AI.

• Mentioning an Element once, then rewriting the prompt without the [@Name] tag.

• Mixing too many conflicting references in one short clip.

One primary subject per first render works best. For storyboards, ask for an image grid first, then animate. Explicit model or duration requests override agent defaults. Chat questions in the Director do not spend credits; confirmed renders do.

If attachments never finish uploading or the bar will not submit after a refresh, use Help Center chat at /help or email support@edgecut.ai. For “which model should I pick?”, use the Generate articles or ask the Creative Director.