Image Upscaler
Upload an image and choose the 2× or 4× setting to reconstruct detail with AI. Actual output dimensions depend on the selected model.
Examples to try
How to Upscale an Image with AI
Choose a 2× or 4× setting to enlarge an image with AI-reconstructed detail. Output resolution depends on the model, and reconstructed details may differ from the original.
- 01
Upload a clear source image
Add one JPG, PNG, or WebP up to 5MB. A clear subject and a well-exposed image give the model more useful detail.
- 02
Choose 2× or 4× enlargement
Choose 2× for moderate enlargement or 4× for a higher output resolution where supported. Actual pixel dimensions depend on the model.
- 03
Run the tool and download
Start the edit, compare the before and after, then download the finished image or find it later in My Creations.
The other tools
Try another AI tool optimized for a specific image task.
Image upscaler FAQ
Answers about 2× and 4× enlargement, AI detail reconstruction, supported formats, text, printing, and private uploads.
How much bigger does it get?
Choose the 2× or 4× setting. Nano Banana Pro uses 2K or 4K output respectively; other models use a supported resolution. Actual pixel dimensions are not guaranteed to be exactly twice or four times the original dimensions.
What does one run cost?
With Nano Banana Pro, 2× uses 2K output for 95 credits and 4× uses 4K output for 155 credits per successful image. Other models have their own credit costs, shown in the model picker. A run that fails costs nothing.
Will it invent detail that wasn't there?
To a degree — that is how AI upscaling works. It reconstructs plausible detail rather than magnifying pixels, so faces and text can shift slightly. For documents, rescan rather than upscale.
What happens to the image I upload?
It is stored so the model can fetch it and so the run stays in your library. It is private to your account and never appears in the public gallery.
Which images benefit most from AI upscaling?
Photos, illustrations, generated images, and product shots with a recognizable subject usually work well. Extremely compressed files cannot recover every lost detail.
Can an AI image upscaler sharpen text and logos?
It can improve edges, but small letters and exact logo geometry may change because the model reconstructs detail. Use a vector original for brand marks whenever one exists.
Enlarge Photos with AI-Reconstructed Detail
Upload a photo or generated image, choose a 2× or 4× setting, and download the result. Check its dimensions and reconstructed details before using it in a layout or print.





