The composer is the prompt box at the top of AI Studio. Describe what you want, attach photos if you like, pick a model, and make images or short video clips without choosing a tool. Use it for one-off ideas that don't fit a tool, or to animate a photo.
Before you start
- Open AI Studio at agentr.ee/workspace/studio. The composer is the box that reads "Describe what you want to create…".
- Any photos you want to use must be image files under 12 MB. You can't attach video files.
- Some ready-made prompts use your city. Set your market in Settings, Audience first.
The composer is open to every plan, and each model costs the same on Free and Pro. The Generate button shows the total credit cost next to a lightning icon. Check it every time: it changes with the model, the number of images and the clip length.
How to make an image or clip
- Click the model button in the composer's toolbar and pick a model under Images or Videos.
- Type what you want in the box. Press Shift+Enter for a new line.
- To add photos, click + and choose Upload from computer or Media Library. You can also drag images onto the box, where you'll see Drop images to attach, or paste them in.
- Click the aspect button and pick a shape, for example 4:5 or 9:16.
- For an image model, pick how many to make, from 1 image to 4 images. For a video model, pick the clip length in seconds.
- Check the number on the Generate button.
- Click Generate, or press Enter.
Choosing a model
| Model | Makes | What the menu says | Photos you can attach |
|---|---|---|---|
| Nano Banana 2 | Images | Best for photos of you, keeps the face | Up to 6 |
| GPT Image 2 | Images | Good with text and graphics | Up to 4 |
| WAN 3.0 | Video | Animates one photo, faithful and cheap | Exactly 1 |
| Kling 3 | Video | Strong motion from a start frame | Exactly 1 |
| Seedance 2.5 | Video | Cinematic, from a first frame | Exactly 1 |
Each video model animates one photo, used as the clip's first frame. Clip lengths are 2 to 30 seconds on WAN 3.0, 3 to 15 on Kling 3 and 4 to 30 on Seedance 2.5.
Under each attached photo you'll see a label. For image models it's @image1, @image2 and so on, in the order you attached them, so you can refer to them in your prompt, for example "use the house in @image2 as the background". For video it reads First frame, or Start on Kling 3. Click the small × to remove a photo.
Using ready-made prompts
Type / at the start of the box, or click / Prompts, to open the prompt library. Prompts are grouped under headings such as Headshots, Listing photos, Social graphics, Video and Viral. Type to search, use the arrow keys to move, and press Enter to choose.
The prompt shows as a pink chip above the box, and the model switches to the one it's made for. Add your own details if you like, such as "navy blazer, big smile". If the chip says it needs your market, click Set your market in the line under the box and add your city.
Running, finished and failed cards
Your results show under the composer in Your generations. A running card shows the model name and a progress bar. The bar is an estimate: under a minute and a half for most images, and about 2 to 3 minutes for a clip. You can make up to 4 images or clips at once, and a 4-image run uses all 4 slots.
Click a finished card to open it. You can Download it, Use in a new generation (images only), Reuse prompt and settings, or Delete it. Delete removes it right away, with no second check.
A failed card reads Generation failed. Its credits come back automatically, and the card says how many. Click Retry to run it again with the same settings, as a new charge, or Remove to clear it.
Tips for better results
- Pick Nano Banana 2 for pictures of you. It keeps your face.
- Pick GPT Image 2 when the image needs words or graphics on it.
- Refer to your photos by their labels, like @image1, so the model knows which is which.
- To animate an image you made, open it, click Use in a new generation and switch to a video model.
Common questions
Why is the Generate button grayed out?
A short line under the box says why. Common reasons are a video model with no photo, too many photos, a prompt that needs your market, 4 runs already going, or not enough credits. Click the link in that line to fix it.
Can I attach a video?
No. The composer takes images only. You'll see "Video inputs aren't supported yet. Add images."
Where are my composer results saved?
Images are saved to the AI Studio tab of your Media Library. Click All in Media Library under Your generations to open it. Video clips aren't in the Media Library. Find them on your Generations page at agentr.ee/workspace/generations, under Video.
Related articles
Read How to get started with AI Studio, How to find your AI creations, How AI credits work in AI Studio and How to set your market so AI content is local.