HUBAL › User manual › Media AI
Media AI — Media Kreator
Media AI is a media studio on the project canvas: images, video clips, voice-over, music, montage, subtitles and 3D models.
Media AI is part of the paid Media Kreator add-on: 19 zł net per month for the whole company, on every plan (Free included), however many people have access. An administrator turns it on in Settings → Subscription → Add-ons → Media Kreator → Add package — before paying, they accept the Media Kreator add-on terms on behalf of the organisation. Without accepted terms, creating media is paused (finished results can still be viewed); when a new version of the terms appears, the administrator accepts it with Accept terms in the same place. After the purchase the organisation owner has access straight away; you add more people or whole permission presets with Who has access.
Costs and storage. Every generation uses AI credits from the company pool — you see the cost before you click Generate. In Settings → Subscription → AI credits → Credit split an administrator decides how much of the pool media may use. Finished files take up the organisation's file storage.
Tiles on the canvas
Choose the kind of tile. Objects you selected on the canvas beforehand become its inputs.
The + input button on the tile: an image, a clip, a recording, a note or another tile. One tile's result can feed the next — Generate with predecessors generates the whole chain in order.
The result appears in the tile, usually after 1–2 minutes (a 3D model: 1–5 minutes). A further message in the tile refines the result, and the arrows take you back to an earlier version (Use this version).
| Tile | What it makes |
|---|---|
| Image | A picture from a description. A connected image can be a Reference, an image to edit (Edit this image) or a style reference — change the role by clicking it on the tile. |
| Clip | A short video from a description or from a starting image (a second image = the last frame). |
| Voice-over | A recording of the text read in the chosen voice, including My voice. |
| Music | An instrumental track from a description (about 30 s, no vocals). |
| Edit (montage) | A video made of images, clips, voice-over and music — a simple montage or the Timeline with tracks, captions and title cards. |
| Subtitles | Subtitles from the speech in a connected video; download them as an .srt file. |
| 3D model | A GLB model from a photo of an object or from a description — rotate it in the tile and download it. |
| Prompt | A shared description for other tiles, in two fields: Content and Style. Read from media reads the content and style of connected images and videos. Save prompts in the library (Mine or Team). |
When a tile needs an image and you only have a video, use Use a frame from the video… — pause the video on the shot you want and the frame lands on the canvas as an image.
My voice
You can clone only your own voice, recorded live in HUBAL. In a Voice-over tile choose the ElevenLabs — Mój głos (My voice) model and click Record your voice. Read the displayed text out loud together with the sentence with a random code (30 s to 3 minutes, in a quiet place), name the voice and confirm the statement that it is your own voice recorded just now. Cloning usually takes up to a minute and uses credits. A new voice is private — only you can read with it until you share it; in Manage voices you can share it with the organisation, rename it or delete it. A deleted voice is also removed from the voice provider.
Download and saving to the Drive
Download a finished result to your computer or Save to Drive — into the project files or your private Drive. Images, videos, recordings and 3D models can be up to 5 GB on the Drive (other files up to 100 MB).
AI content label. Images, videos and recordings from Media AI carry embedded Content Credentials (the C2PA standard) — signed information that the file was made with AI. Downloading and saving to the Drive keep it. Anyone can check it at contentcredentials.org/verify.
People with Media Kreator access can also ask the assistant on the project canvas, or an AI assistant based on the Marketing template, e.g. "prepare three images for a post with our logo". Before a bigger job the assistant gives an estimated cost, and the results land on the project's Media AI canvas.