Doodle Generation

When words alone cannot explain what to change and what to keep, mark the location and area directly on an image.

What you can do

Doodle Generation is useful for local changes to an existing image. For example, you can:

Remove a stool from the scene.
Change only the color of a person's clothes while keeping the rest.
Show a spatial relationship such as “put the object behind and to the left of the person.”

The marks guide the next image generation together with your Prompt. They do not directly overwrite the original image.

Open the image you want to change in the preview area.
Click Doodle Edit in the top toolbar to open the marking canvas.
Mark the area to work on:
- Point: Indicate an exact location.
- Box: Specify a rectangular area.
- Brush: Paint an irregular area.
- Arrow: Show a direction or relationship between two places.
Adjust color or opacity, or undo or delete marks if needed.
Explain in the Prompt what to change at each mark and what to preserve. Review the model and generation settings, then submit.

For example:

Remove the stool inside the box and fill in the floor naturally. Keep the person, camera angle, and other furniture unchanged.

Do not rely on marks alone. The Prompt still needs to describe the intended change and what must stay.
If different areas need different changes, mark them separately and explain each one in the Prompt.
Image-reference support varies by model. When you open Doodle Edit, Flova uses a generation capability that supports the marks.

Doodle Generation is for local changes to an existing image, not freehand drawing that generates an image by itself.
If you delete a mark, update any Prompt text that refers to it.

Quick Generation: Start a direct generation from a blank input or an existing reference.
AI Upscale: Improve the clarity of the whole image without specifying a local change.