AI Text to Image
Edit Photos with AI Using Text Prompts
The VdoBloom AI image editor is an AI tool that modifies existing photos based on written instructions. Upload an image, describe the change — swap a background, change colors, add or remove objects — and the AI applies the edit while keeping the rest of the picture intact.
The editor supports multiple editing models, including Nano Banana Edit, Flux Kontext Pro and Max, Seedream edit models, and GPT Image, with several models accepting multiple reference images in a single edit. Specialized options like erase and outpainting let you remove unwanted elements or extend an image beyond its original canvas.
How it works
- 1
Upload your image
Add the photo you want to change; many models accept several reference images at once.
- 2
Describe the edit
Tell the AI what to change in plain language — no masks or layers required for most edits.
- 3
Pick a model and apply
Choose an editing model suited to your task and run the edit; results appear in a few minutes.
Frequently asked questions
How does AI photo editing with text prompts work?
You upload a photo and type an instruction such as "make the sky sunset orange" or "remove the person in the background." The AI model interprets the instruction, edits only the relevant region, and returns a new version of the image while preserving everything you did not ask to change.
Can I edit an image using multiple reference images?
Yes. Several editing models on VdoBloom accept multiple input images in one request — for example, combining a product photo with a background reference. The maximum number of images depends on the model you choose, and the editor shows the limit for each one.
Can the AI remove objects or extend an image?
Yes. The editor includes erase tools for removing unwanted objects from a photo and outpainting for expanding the canvas in any direction. Outpainting generates new content that matches the original lighting and style, which is useful for converting an image to a different aspect ratio.
What kinds of edits give the best results?
Clear, single-purpose instructions work best: change one thing per edit, name the object or region explicitly, and describe the desired outcome rather than the technical operation. For larger transformations, chain several edits together, reviewing the output at each step.