Generate Images in Studio
Write a prompt in a Studio chat and get one to four images back, without opening a canvas.
Before you start
- You need an image model your account can use. If none is available, the composer says Choose an available image model to generate.
- Every Generate and Retry sends real requests to an image model, one for each image, and counts toward your usage.
The whole flow
Watch it once from the canvas library to finished images, or click a step under the replay to jump to it.
5 stepsFrom the library to finished images in a Studio chat
Open Studio
In the workspace sidebar, click Studio. The sidebar folds to a narrow rail, and Studio opens a new chat with your Chats panel beside it. The new chat says What do you want to make?
To open the full sidebar again, click the Gavana logo at the top of the rail, or press Cmd + B (Ctrl + B on Windows).
Describe the image
Click the box at the bottom that reads Describe what you want to make… and write what you want to see: the subject, the setting, the light and the mood.
Choose the settings
The row under the prompt holds the settings for this run.
| Control | What it changes | Options |
|---|---|---|
| Model | Which AI model makes the images | The image models your account can use |
| Image settings (the sliders button) | Shape, detail and quality | Aspect ratio: Auto, 1:1 · Square, 3:2 · Landscape, 2:3 · Portrait, 4:3 · Landscape, 3:4 · Portrait, 16:9 · Widescreen, 9:16 · Vertical. Resolution: Auto, 1K, 2K, 4K. Quality: Auto, High, Medium, Low |
| Fewer images and More images (the – and + buttons) | How many images this run makes | 1 to 4 |
The Image settings menu stays open while you pick, so you can set all three in one go. Click outside it to close it. The button then shows what you chose, for example 16:9 · 2K.
Generate
Click the round arrow button (Generate, or Generate 2 images when you make more than one), or press Cmd + Enter (Ctrl + Enter on Windows). The button stays off until you’ve written a prompt.
A run card appears with your prompt, a line such as GPT Image 2.5 Sunburst · 16:9 · 2k · ×2, and one tile for each image. Each tile reads Generating… with a timer until its image arrives. The first run in a new chat also names the chat after your prompt and lists it under Today in the Chats panel.
Every image is saved to Assets as well.
Open an image
Click a finished image (or hover it and click Open image) to see it full size. Click Previous image and Next image, or press ← and →, to move through every image in the chat. Press Esc to close the viewer.
More you can do
Stop a run, then retry it
While a run is generating, click Stop under its card. Images that hadn’t finished read Stopped.
Once every tile has finished, failed or stopped, a Retry button appears under any card with a failed or stopped image. Retry adds a new run card with the same settings that makes only the missing images. The original card stays as it was, and a card offers Retry only once.
Stop and RetryStop a run, then make its missing images again
Retry starts new paid requests, one for each image it makes again.
Start from an earlier run
Under any run card:
- Reuse settings puts that run’s prompt, model, image settings and image count back in the composer, so you can change one thing and generate again.
- Copy prompt copies just the prompt and says Prompt copied.
ReuseStart from a run you already made
Download an image
Hover a finished image and click Download to save the original file.
If an image fails
A failed tile shows what went wrong, for example The image could not be made. Try again. Use Retry under the card to make the failed images again.
If a tile reads Saving…, the image was made and Studio is still saving it. You don’t need to do anything.
Keyboard shortcuts
| Key | Where | What it does |
|---|---|---|
| Cmd / Ctrl + Enter | Prompt box | Generates |
| ← / → | Image viewer | Previous or next image in the chat |
| Esc | Image viewer | Closes the viewer |
| Cmd / Ctrl + B | Studio | Opens or folds the workspace sidebar |